LUCENE-3312: Merge up to trunk

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/branches/lucene3312@1377246 13f79535-47bb-0310-9956-ffa450edef68

build.xml

@@ -265,7 +265,6 @@
       <param name="tests.jvms" value="1"/>
       <!-- Also override some other props to be fast, ignoring what's set on command line: -->
       <param name="tests.multiplier" value="1"/>
-      <param name="tests.slow" value="false"/>
       <param name="tests.nightly" value="false"/>
       <param name="tests.weekly" value="false"/>
       <param name="tests.multiplier" value="1"/>
@@ -276,7 +275,7 @@
   <!-- we need this extra condition, as we want to match only on "true", not solely if property is set: -->
   <property name="disable.javadocs-lint" value="false" />
   <condition property="-disable.javadocs-lint">
-    <equals arg1="${disable.javadocs-lint}" arg2="true"/>
+    <istrue value="${disable.javadocs-lint}"/>
   </condition>
   <target name="-jenkins-javadocs-lint" unless="-disable.javadocs-lint">
     <antcall target="javadocs-lint"/> 

dev-tools/idea/lucene/analysis/morfologik/morfologik.iml

@@ -6,6 +6,7 @@
     <exclude-output />
     <content url="file://$MODULE_DIR$">
       <sourceFolder url="file://$MODULE_DIR$/src/java" isTestSource="false" />
+      <sourceFolder url="file://$MODULE_DIR$/src/resources" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/src/test" isTestSource="true" />
     </content>
     <orderEntry type="inheritedJdk" />

dev-tools/idea/lucene/analysis/phonetic/phonetic.iml

@@ -6,6 +6,7 @@
     <exclude-output />
     <content url="file://$MODULE_DIR$">
       <sourceFolder url="file://$MODULE_DIR$/src/java" isTestSource="false" />
+      <sourceFolder url="file://$MODULE_DIR$/src/resources" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/src/test" isTestSource="true" />
     </content>
     <orderEntry type="inheritedJdk" />

dev-tools/maven/pom.xml.template

@@ -47,7 +47,7 @@
     <java.compat.version>1.6</java.compat.version>
     <jetty.version>8.1.2.v20120308</jetty.version>
     <slf4j.version>1.6.4</slf4j.version>
-    <tika.version>1.1</tika.version>
+    <tika.version>1.2</tika.version>
     <httpcomponents.version>4.1.3</httpcomponents.version>
 
     <!-- RandomizedTesting library system properties -->

dev-tools/scripts/smokeTestRelease.py

@@ -546,7 +546,7 @@ def testNotice(unpackPath):
   if solrNotice.find(expected) == -1:
     raise RuntimeError('Solr\'s NOTICE.txt does not have the verbatim copy, plus header/footer, of Lucene\'s NOTICE.txt')
   
-def readSolrOutput(p, startupEvent, logFile):
+def readSolrOutput(p, startupEvent, failureEvent, logFile):
   f = open(logFile, 'wb')
   try:
     while True:
@@ -556,8 +556,13 @@ def readSolrOutput(p, startupEvent, logFile):
       f.write(line)
       f.flush()
       # print 'SOLR: %s' % line.strip()
-      if line.decode('UTF-8').find('Started SocketConnector@0.0.0.0:8983') != -1:
+      if not startupEvent.isSet() and line.find(b'Started SocketConnector@0.0.0.0:8983') != -1:
         startupEvent.set()
+  except:
+    print()
+    print('Exception reading Solr output:')
+    traceback.print_exc()
+    failureEvent.set()
   finally:
     f.close()
     
@@ -572,7 +577,8 @@ def testSolrExample(unpackPath, javaPath, isSrc):
   server = subprocess.Popen(['java', '-jar', 'start.jar'], stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=env)
 
   startupEvent = threading.Event()
-  serverThread = threading.Thread(target=readSolrOutput, args=(server.stderr, startupEvent, logFile))
+  failureEvent = threading.Event()
+  serverThread = threading.Thread(target=readSolrOutput, args=(server.stderr, startupEvent, failureEvent, logFile))
   serverThread.setDaemon(True)
   serverThread.start()
 
@@ -609,6 +615,9 @@ def testSolrExample(unpackPath, javaPath, isSrc):
         # Shouldn't happen unless something is seriously wrong...
         print('***WARNING***: Solr instance didn\'t respond to SIGKILL; ignoring...')
 
+  if failureEvent.isSet():
+    raise RuntimeError('exception while reading Solr output')
+    
   os.chdir('..')
     
 def checkJavadocpath(path):

lucene/CHANGES.txt

@@ -15,6 +15,15 @@ New Features
   that you must store term vector positions to store payloads.
   (Robert Muir)
 
+* LUCENE-3892: Add a new BlockPostingsFormat that bulk-encodes docs,
+  freqs and positions in large (size 128) packed-int blocks for faster
+  search performance.  This was from Han Jiang's 2012 Google Summer of
+  Code project (Han Jiang, Adrien Grand, Robert Muir, Mike McCandless)
+
+* LUCENE-4323: Added support for an absolute maximum CFS segment size
+  (in MiB) to LogMergePolicy and TieredMergePolicy.
+  (Alexey Lef via Uwe Schindler)
+
 API Changes
 
 * LUCENE-4299: Added Terms.hasPositions() and Terms.hasOffsets().
@@ -50,6 +59,24 @@ API Changes
 * LUCENE-4307: Renamed IndexReader.getTopReaderContext to 
   IndexReader.getContext.  (Robert Muir)
 
+* LUCENE-4316: Deprecate Fields.getUniqueTermCount and remove it from
+  AtomicReader. If you really want the unique term count across all
+  fields, just sum up Terms.size() across those fields. This method
+  only exists so that this statistic can be accessed for Lucene 3.x 
+  segments, which don't support Terms.size().  (Uwe Schindler, Robert Muir)
+
+* LUCENE-4321: Change CharFilter to extend Reader directly, as FilterReader
+  overdelegates (read(), read(char[], int, int), skip, etc). This made it
+  hard to implement CharFilters that were correct. Instead only close() is
+  delegated by default: read(char[], int, int) and correct(int) are abstract
+  so that its obvious which methods you should implement.  The protected 
+  inner Reader is 'input' like CharFilter in the 3.x series, instead of 'in'.  
+  (Dawid Weiss, Uwe Schindler, Robert Muir)
+
+* LUCENE-3309: The expert FieldSelector API, used to load only certain
+  fields in a stored document, has been replaced with the simpler
+  StoredFieldVisitor API.  (Mike McCandless)
+
 Bug Fixes
 
 * LUCENE-4297: BooleanScorer2 would multiply the coord() factor
@@ -83,6 +110,14 @@ Bug Fixes
   containing non-BMP Unicode characters.  (Dawid Weiss, Robert Muir,
   Mike McCandless)
 
+Optimizations
+
+* LUCENE-4317: Improve reuse of internal TokenStreams and StringReader
+  in oal.document.Field.  (Uwe Schindler, Chris Male, Robert Muir)
+
+* LUCENE-4327: Support out-of-order scoring in FilteredQuery for higher
+  performance.  (Mike McCandless, Robert Muir)
+
 Build
 
 * LUCENE-3985: Upgrade to randomizedtesting 2.0.0. Added support for 

lucene/MIGRATE.txt

@@ -619,3 +619,10 @@ you can now do this:
   postings in simple arrrays (byte[] for terms, int[] for docs, freqs,
   positions, offsets).  Note that this only covers postings, whereas
   Instantiated covered all other parts of the index as well.
+
+* LUCENE-3309: The expert FieldSelector API has been replaced with
+  StoredFieldVisitor.  The idea is the same (you have full control
+  over which fields should be loaded).  Instead of a single accept
+  method, StoredFieldVisitor has a needsField method: if that method
+  returns true then the field will be loaded and the appropriate
+  type-specific method will be invoked with that fields's value.

lucene/analysis/common/src/java/org/apache/lucene/analysis/charfilter/MappingCharFilter.java

@@ -67,8 +67,8 @@ public class MappingCharFilter extends BaseCharFilter {
 
   @Override
   public void reset() throws IOException {
-    super.reset();
-    buffer.reset(in);
+    input.reset();
+    buffer.reset(input);
     replacement = null;
     inputOff = 0;
   }

lucene/analysis/common/src/java/org/apache/lucene/analysis/fa/PersianCharFilter.java

@@ -34,7 +34,7 @@ public class PersianCharFilter extends CharFilter {
   
   @Override
   public int read(char[] cbuf, int off, int len) throws IOException {
-    final int charsRead = super.read(cbuf, off, len);
+    final int charsRead = input.read(cbuf, off, len);
     if (charsRead > 0) {
       final int end = off + charsRead;
       while (off < end) {
@@ -45,6 +45,17 @@ public class PersianCharFilter extends CharFilter {
     }
     return charsRead;
   }
+  
+  // optimized impl: some other charfilters consume with read()
+  @Override
+  public int read() throws IOException {
+    int ch = input.read();
+    if (ch == '\u200C') {
+      return ' ';
+    } else {
+      return ch;
+    }
+  }
 
   @Override
   protected int correct(int currentOff) {

lucene/analysis/common/src/java/org/apache/lucene/analysis/pattern/PatternReplaceCharFilter.java

@@ -72,7 +72,7 @@ public class PatternReplaceCharFilter extends BaseCharFilter {
   private void fill() throws IOException {
     StringBuilder buffered = new StringBuilder();
     char [] temp = new char [1024];
-    for (int cnt = in.read(temp); cnt > 0; cnt = in.read(temp)) {
+    for (int cnt = input.read(temp); cnt > 0; cnt = input.read(temp)) {
       buffered.append(temp, 0, cnt);
     }
     transformedInput = new StringReader(processPattern(buffered).toString());

lucene/analysis/common/src/test/org/apache/lucene/analysis/core/TestRandomChains.java

@@ -782,31 +782,51 @@ public class TestRandomChains extends BaseTokenStreamTestCase {
     @Override
     public int read(char[] cbuf, int off, int len) throws IOException {
       readSomething = true;
-      return in.read(cbuf, off, len);
+      return input.read(cbuf, off, len);
     }
 
     @Override
     public int read() throws IOException {
       readSomething = true;
-      return in.read();
+      return input.read();
     }
 
     @Override
     public int read(CharBuffer target) throws IOException {
       readSomething = true;
-      return in.read(target);
+      return input.read(target);
     }
 
     @Override
     public int read(char[] cbuf) throws IOException {
       readSomething = true;
-      return in.read(cbuf);
+      return input.read(cbuf);
     }
 
     @Override
     public long skip(long n) throws IOException {
       readSomething = true;
-      return in.skip(n);
+      return input.skip(n);
+    }
+
+    @Override
+    public void mark(int readAheadLimit) throws IOException {
+      input.mark(readAheadLimit);
+    }
+
+    @Override
+    public boolean markSupported() {
+      return input.markSupported();
+    }
+
+    @Override
+    public boolean ready() throws IOException {
+      return input.ready();
+    }
+
+    @Override
+    public void reset() throws IOException {
+      input.reset();
     }
   }
   

lucene/analysis/common/src/test/org/apache/lucene/analysis/fa/TestPersianCharFilter.java

@@ -0,0 +1,48 @@
+package org.apache.lucene.analysis.fa;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.Reader;
+
+import org.apache.lucene.analysis.Analyzer;
+import org.apache.lucene.analysis.BaseTokenStreamTestCase;
+import org.apache.lucene.analysis.MockTokenizer;
+
+public class TestPersianCharFilter extends BaseTokenStreamTestCase {
+  private Analyzer analyzer = new Analyzer() {
+    @Override
+    protected TokenStreamComponents createComponents(String fieldName, Reader reader) {
+      return new TokenStreamComponents(new MockTokenizer(reader));
+    }
+
+    @Override
+    protected Reader initReader(String fieldName, Reader reader) {
+      return new PersianCharFilter(reader);
+    }
+  };
+  
+  public void testBasics() throws Exception {
+    assertAnalyzesTo(analyzer, "this is a\u200Ctest",
+        new String[] { "this", "is", "a", "test" });
+  }
+  
+  /** blast some random strings through the analyzer */
+  public void testRandomStrings() throws Exception {
+    checkRandomData(random(), analyzer, 1000*RANDOM_MULTIPLIER);
+  }
+}

lucene/build.xml

@@ -238,7 +238,7 @@
       <check-missing-javadocs dir="build/docs/demo" level="class"/>
       <check-missing-javadocs dir="build/docs/facet" level="class"/>
       <check-missing-javadocs dir="build/docs/grouping" level="class"/>
-      <!-- highlighter: problems -->
+      <check-missing-javadocs dir="build/docs/highlighter" level="class"/>
       <check-missing-javadocs dir="build/docs/join" level="class"/>
       <check-missing-javadocs dir="build/docs/memory" level="class"/>
       <check-missing-javadocs dir="build/docs/misc" level="class"/>

lucene/common-build.xml

@@ -135,7 +135,7 @@
   <property name="javac.target" value="1.6"/>
   <property name="javac.source.backwards" value="1.6"/>
   <property name="javac.target.backwards" value="1.6"/>
-  <property name="javac.args" value="-Xlint -Xlint:-deprecation -Xlint:-serial"/>
+  <property name="javac.args" value="-Xlint -Xlint:-deprecation -Xlint:-serial -Xlint:-options"/>
   <property name="bootclasspath" value=""/>
   <property name="javadoc.link" value="http://download.oracle.com/javase/6/docs/api/"/>
   <property name="javadoc.link.junit" value="http://junit.sourceforge.net/javadoc/"/>
@@ -713,6 +713,7 @@
     <or>
       <isset property="tests.class" />
       <isset property="tests.method" />
+      <istrue value="${tests.showSuccess}"/>
     </or>
   </condition>
   <property name="tests.showOutput" value="onerror"/>

lucene/core/build.xml

@@ -62,16 +62,40 @@
   	<sequential>
       <exec dir="src/java/org/apache/lucene/util/automaton"
             executable="${python.exe}" failonerror="true">
-        <arg line="createLevAutomata.py @{n} True"/>
+        <arg value="createLevAutomata.py"/>
+        <arg value="@{n}"/>
+        <arg value="True"/>
       </exec>
       <exec dir="src/java/org/apache/lucene/util/automaton"
             executable="${python.exe}" failonerror="true">
-        <arg line="createLevAutomata.py @{n} False"/>
+        <arg value="createLevAutomata.py"/>
+        <arg value="@{n}"/>
+        <arg value="False"/>
       </exec>
       <fixcrlf srcdir="src/java/org/apache/lucene/util/automaton" includes="*ParametricDescription.java" encoding="UTF-8"/>
     </sequential>
   </macrodef>
 
+  <target name="createPackedIntSources">
+    <exec dir="src/java/org/apache/lucene/util/packed"
+          executable="${python.exe}" failonerror="true">
+      <arg value="gen_BulkOperation.py"/>
+    </exec>
+    <exec dir="src/java/org/apache/lucene/util/packed"
+          executable="${python.exe}" failonerror="true">
+      <arg value="gen_Direct.py"/>
+    </exec>
+    <exec dir="src/java/org/apache/lucene/util/packed"
+          executable="${python.exe}" failonerror="true">
+      <arg value="gen_Packed64SingleBlock.py"/>
+    </exec>
+    <exec dir="src/java/org/apache/lucene/util/packed"
+          executable="${python.exe}" failonerror="true">
+      <arg value="gen_PackedThreeBlocks.py"/>
+    </exec>
+    <fixcrlf srcdir="src/java/org/apache/lucene/util/packed" includes="BulkOperation*.java,Direct*.java,Packed64SingleBlock.java,Packed*ThreeBlocks.py" encoding="UTF-8"/>
+  </target>
+
   <target name="createLevAutomata" depends="check-moman,clone-moman,pull-moman">
   	<createLevAutomaton n="1"/>
   	<createLevAutomaton n="2"/>
@@ -86,14 +110,21 @@
   <target name="clone-moman" unless="moman.cloned">
   	<exec dir="src/java/org/apache/lucene/util/automaton" 
           executable="${hg.exe}" failonerror="true">
-      <arg line="clone -r ${moman.rev} ${moman.url} moman"/>
+      <arg value="clone"/>
+      <arg value="-r"/>
+      <arg value="${moman.rev}"/>
+      <arg value="${moman.url}"/>
+      <arg value="moman"/>
     </exec>
   </target>
 
   <target name="pull-moman" if="moman.cloned">
     <exec dir="src/java/org/apache/lucene/util/automaton/moman" 
           executable="${hg.exe}" failonerror="true">
-      <arg line="pull -f -r ${moman.rev}"/>
+      <arg value="pull"/>
+      <arg value="-f"/>
+      <arg value="-r"/>
+      <arg value="${moman.rev}"/>
     </exec>
   </target>
 </project>

lucene/core/src/java/org/apache/lucene/analysis/Analyzer.java

@@ -30,8 +30,42 @@ import java.util.Map;
  * policy for extracting index terms from text.
  * <p>
  * In order to define what analysis is done, subclasses must define their
- * {@link TokenStreamComponents} in {@link #createComponents(String, Reader)}.
+ * {@link TokenStreamComponents TokenStreamComponents} in {@link #createComponents(String, Reader)}.
  * The components are then reused in each call to {@link #tokenStream(String, Reader)}.
+ * <p>
+ * Simple example:
+ * <pre class="prettyprint">
+ * Analyzer analyzer = new Analyzer() {
+ *  {@literal @Override}
+ *   protected TokenStreamComponents createComponents(String fieldName, Reader reader) {
+ *     Tokenizer source = new FooTokenizer(reader);
+ *     TokenStream filter = new FooFilter(source);
+ *     filter = new BarFilter(filter);
+ *     return new TokenStreamComponents(source, filter);
+ *   }
+ * };
+ * </pre>
+ * For more examples, see the {@link org.apache.lucene.analysis Analysis package documentation}.
+ * <p>
+ * For some concrete implementations bundled with Lucene, look in the analysis modules:
+ * <ul>
+ *   <li><a href="{@docRoot}/../analyzers-common/overview-summary.html">Common</a>:
+ *       Analyzers for indexing content in different languages and domains.
+ *   <li><a href="{@docRoot}/../analyzers-icu/overview-summary.html">ICU</a>:
+ *       Exposes functionality from ICU to Apache Lucene. 
+ *   <li><a href="{@docRoot}/../analyzers-kuromoji/overview-summary.html">Kuromoji</a>:
+ *       Morphological analyzer for Japanese text.
+ *   <li><a href="{@docRoot}/../analyzers-morfologik/overview-summary.html">Morfologik</a>:
+ *       Dictionary-driven lemmatization for the Polish language.
+ *   <li><a href="{@docRoot}/../analyzers-phonetic/overview-summary.html">Phonetic</a>:
+ *       Analysis for indexing phonetic signatures (for sounds-alike search).
+ *   <li><a href="{@docRoot}/../analyzers-smartcn/overview-summary.html">Smart Chinese</a>:
+ *       Analyzer for Simplified Chinese, which indexes words.
+ *   <li><a href="{@docRoot}/../analyzers-stempel/overview-summary.html">Stempel</a>:
+ *       Algorithmic Stemmer for the Polish Language.
+ *   <li><a href="{@docRoot}/../analyzers-uima/overview-summary.html">UIMA</a>: 
+ *       Analysis integration with Apache UIMA. 
+ * </ul>
  */
 public abstract class Analyzer {
 

lucene/core/src/java/org/apache/lucene/analysis/CharFilter.java

@@ -17,7 +17,7 @@ package org.apache.lucene.analysis;
  * limitations under the License.
  */
 
-import java.io.FilterReader;
+import java.io.IOException;
 import java.io.Reader;
 
 /**
@@ -25,17 +25,39 @@ import java.io.Reader;
  * They can be used as {@link java.io.Reader} with additional offset
  * correction. {@link Tokenizer}s will automatically use {@link #correctOffset}
  * if a CharFilter subclass is used.
+ * <p>
+ * This class is abstract: at a minimum you must implement {@link #read(char[], int, int)},
+ * transforming the input in some way from {@link #input}, and {@link #correct(int)}
+ * to adjust the offsets to match the originals.
+ * <p>
+ * You can optionally provide more efficient implementations of additional methods 
+ * like {@link #read()}, {@link #read(char[])}, {@link #read(java.nio.CharBuffer)},
+ * but this is not required.
  */
-public abstract class CharFilter extends FilterReader {
+// the way java.io.FilterReader should work!
+public abstract class CharFilter extends Reader {
+  /** 
+   * The underlying character-input stream. 
+   */
+  protected final Reader input;
 
   /**
    * Create a new CharFilter wrapping the provided reader.
-   * @param in a Reader, can also be a CharFilter for chaining.
+   * @param input a Reader, can also be a CharFilter for chaining.
    */
-  public CharFilter(Reader in) {
-    super(in);
+  public CharFilter(Reader input) {
+    super(input);
+    this.input = input;
   }
   
+  /** 
+   * Closes the underlying input stream.
+   */
+  @Override
+  public void close() throws IOException {
+    input.close();
+  }
+
   /**
    * Subclasses override to correct the current offset.
    *
@@ -50,6 +72,6 @@ public abstract class CharFilter extends FilterReader {
    */
   public final int correctOffset(int currentOff) {
     final int corrected = correct(currentOff);
-    return (in instanceof CharFilter) ? ((CharFilter) in).correctOffset(corrected) : corrected;
+    return (input instanceof CharFilter) ? ((CharFilter) input).correctOffset(corrected) : corrected;
   }
 }

lucene/core/src/java/org/apache/lucene/analysis/tokenattributes/PositionIncrementAttribute.java

@@ -19,7 +19,7 @@ package org.apache.lucene.analysis.tokenattributes;
 
 import org.apache.lucene.util.Attribute;
 
-/** The positionIncrement determines the position of this token
+/** Determines the position of this token
  * relative to the previous Token in a TokenStream, used in phrase
  * searching.
  *

lucene/core/src/java/org/apache/lucene/analysis/tokenattributes/PositionIncrementAttributeImpl.java

@@ -20,7 +20,7 @@ package org.apache.lucene.analysis.tokenattributes;
 import org.apache.lucene.analysis.TokenStream;
 import org.apache.lucene.util.AttributeImpl;
 
-/** The positionIncrement determines the position of this token
+/** Determines the position of this token
  * relative to the previous Token in a {@link TokenStream}, used in phrase
  * searching.
  *

lucene/core/src/java/org/apache/lucene/analysis/tokenattributes/PositionLengthAttribute.java

@@ -19,7 +19,7 @@ package org.apache.lucene.analysis.tokenattributes;
 
 import org.apache.lucene.util.Attribute;
 
-/** The positionLength determines how many positions this
+/** Determines how many positions this
  *  token spans.  Very few analyzer components actually
  *  produce this attribute, and indexing ignores it, but
  *  it's useful to express the graph structure naturally

lucene/core/src/java/org/apache/lucene/codecs/BlockTermsReader.java

@@ -18,6 +18,7 @@ package org.apache.lucene.codecs;
  */
 
 import java.io.IOException;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.Iterator;
 import java.util.TreeMap;
@@ -39,7 +40,6 @@ import org.apache.lucene.util.ArrayUtil;
 import org.apache.lucene.util.Bits;
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.DoubleBarrelLRUCache;
-import org.apache.lucene.util.UnmodifiableIterator;
 
 /** Handles a terms dict, but decouples all details of
  *  doc/freqs/positions reading to an instance of {@link
@@ -185,7 +185,7 @@ public class BlockTermsReader extends FieldsProducer {
 
   @Override
   public Iterator<String> iterator() {
-    return new UnmodifiableIterator<String>(fields.keySet().iterator());
+    return Collections.unmodifiableSet(fields.keySet()).iterator();
   }
 
   @Override
@@ -308,7 +308,7 @@ public class BlockTermsReader extends FieldsProducer {
       private int metaDataUpto;
 
       public SegmentTermsEnum() throws IOException {
-        in = (IndexInput) BlockTermsReader.this.in.clone();
+        in = BlockTermsReader.this.in.clone();
         in.seek(termsStartPointer);
         indexEnum = indexReader.getFieldEnum(fieldInfo);
         doOrd = indexReader.supportsOrd();

lucene/core/src/java/org/apache/lucene/codecs/BlockTreeTermsReader.java

@@ -21,6 +21,7 @@ import java.io.ByteArrayOutputStream;
 import java.io.IOException;
 import java.io.PrintStream;
 import java.io.UnsupportedEncodingException;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.Iterator;
 import java.util.Locale;
@@ -45,7 +46,6 @@ import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.IOUtils;
 import org.apache.lucene.util.RamUsageEstimator;
 import org.apache.lucene.util.StringHelper;
-import org.apache.lucene.util.UnmodifiableIterator;
 import org.apache.lucene.util.automaton.CompiledAutomaton;
 import org.apache.lucene.util.automaton.RunAutomaton;
 import org.apache.lucene.util.automaton.Transition;
@@ -200,7 +200,7 @@ public class BlockTreeTermsReader extends FieldsProducer {
 
   @Override
   public Iterator<String> iterator() {
-    return new UnmodifiableIterator<String>(fields.keySet().iterator());
+    return Collections.unmodifiableSet(fields.keySet()).iterator();
   }
 
   @Override
@@ -400,7 +400,7 @@ public class BlockTreeTermsReader extends FieldsProducer {
       rootBlockFP = (new ByteArrayDataInput(rootCode.bytes, rootCode.offset, rootCode.length)).readVLong() >>> BlockTreeTermsWriter.OUTPUT_FLAGS_NUM_BITS;
 
       if (indexIn != null) {
-        final IndexInput clone = (IndexInput) indexIn.clone();
+        final IndexInput clone = indexIn.clone();
         //System.out.println("start=" + indexStartFP + " field=" + fieldInfo.name);
         clone.seek(indexStartFP);
         index = new FST<BytesRef>(clone, ByteSequenceOutputs.getSingleton());
@@ -746,7 +746,7 @@ public class BlockTreeTermsReader extends FieldsProducer {
         // }
         runAutomaton = compiled.runAutomaton;
         compiledAutomaton = compiled;
-        in = (IndexInput) BlockTreeTermsReader.this.in.clone();
+        in = BlockTreeTermsReader.this.in.clone();
         stack = new Frame[5];
         for(int idx=0;idx<stack.length;idx++) {
           stack[idx] = new Frame(idx);
@@ -1236,7 +1236,7 @@ public class BlockTreeTermsReader extends FieldsProducer {
       // Not private to avoid synthetic access$NNN methods
       void initIndexInput() {
         if (this.in == null) {
-          this.in = (IndexInput) BlockTreeTermsReader.this.in.clone();
+          this.in = BlockTreeTermsReader.this.in.clone();
         }
       }
 

lucene/core/src/java/org/apache/lucene/codecs/BlockTreeTermsWriter.java

@@ -71,7 +71,7 @@ import org.apache.lucene.util.fst.Util;
 */
 
 /**
- * block-based terms index and dictionary writer.
+ * Block-based terms index and dictionary writer.
  * <p>
  * Writes terms dict and index, block-encoding (column
  * stride) each term's metadata for each set of terms
@@ -724,7 +724,7 @@ public class BlockTreeTermsWriter extends FieldsConsumer {
           // Write term stats, to separate byte[] blob:
           bytesWriter2.writeVInt(term.stats.docFreq);
           if (fieldInfo.getIndexOptions() != IndexOptions.DOCS_ONLY) {
-            assert term.stats.totalTermFreq >= term.stats.docFreq;
+            assert term.stats.totalTermFreq >= term.stats.docFreq: term.stats.totalTermFreq + " vs " + term.stats.docFreq;
             bytesWriter2.writeVLong(term.stats.totalTermFreq - term.stats.docFreq);
           }
         }

lucene/core/src/java/org/apache/lucene/codecs/FieldsProducer.java

@@ -22,7 +22,7 @@ import java.io.IOException;
 
 import org.apache.lucene.index.Fields;
 
-/** Abstract API that produces terms, doc, freq, prox and
+/** Abstract API that produces terms, doc, freq, prox, offset and
  *  payloads postings.  
  *
  * @lucene.experimental

lucene/core/src/java/org/apache/lucene/codecs/FixedGapTermsIndexReader.java

@@ -34,7 +34,7 @@ import java.io.IOException;
 import org.apache.lucene.index.IndexFileNames;
 
 /** 
- * TermsIndexReader for simple every-nth terms indexes.
+ * TermsIndexReader for simple every Nth terms indexes.
  *
  * @see FixedGapTermsIndexWriter
  * @lucene.experimental 
@@ -278,7 +278,7 @@ public class FixedGapTermsIndexReader extends TermsIndexReaderBase {
         this.termsStart = termsStart;
         termBytesStart = termBytes.getPointer();
 
-        IndexInput clone = (IndexInput) in.clone();
+        IndexInput clone = in.clone();
         clone.seek(indexStart);
 
         // -1 is passed to mean "don't load term index", but
@@ -309,8 +309,8 @@ public class FixedGapTermsIndexReader extends TermsIndexReaderBase {
           }
         } else {
           // Get packed iterators
-          final IndexInput clone1 = (IndexInput) in.clone();
-          final IndexInput clone2 = (IndexInput) in.clone();
+          final IndexInput clone1 = in.clone();
+          final IndexInput clone2 = in.clone();
 
           try {
             // Subsample the index terms

lucene/core/src/java/org/apache/lucene/codecs/MultiLevelSkipListReader.java

@@ -59,30 +59,36 @@ public abstract class MultiLevelSkipListReader {
   private int skipInterval[];         // skipInterval of each level
   private int[] numSkipped;           // number of docs skipped per level
     
-  private int[] skipDoc;              // doc id of current skip entry per level 
+  protected int[] skipDoc;            // doc id of current skip entry per level 
   private int lastDoc;                // doc id of last read skip entry with docId <= target
   private long[] childPointer;        // child pointer of current skip entry per level
   private long lastChildPointer;      // childPointer of last read skip entry with docId <= target
   
   private boolean inputIsBuffered;
-  
-  public MultiLevelSkipListReader(IndexInput skipStream, int maxSkipLevels, int skipInterval) {
+  private final int skipMultiplier;
+
+  protected MultiLevelSkipListReader(IndexInput skipStream, int maxSkipLevels, int skipInterval, int skipMultiplier) {
     this.skipStream = new IndexInput[maxSkipLevels];
     this.skipPointer = new long[maxSkipLevels];
     this.childPointer = new long[maxSkipLevels];
     this.numSkipped = new int[maxSkipLevels];
     this.maxNumberOfSkipLevels = maxSkipLevels;
     this.skipInterval = new int[maxSkipLevels];
+    this.skipMultiplier = skipMultiplier;
     this.skipStream [0]= skipStream;
     this.inputIsBuffered = (skipStream instanceof BufferedIndexInput);
     this.skipInterval[0] = skipInterval;
     for (int i = 1; i < maxSkipLevels; i++) {
       // cache skip intervals
-      this.skipInterval[i] = this.skipInterval[i - 1] * skipInterval;
+      this.skipInterval[i] = this.skipInterval[i - 1] * skipMultiplier;
     }
     skipDoc = new int[maxSkipLevels];
   }
 
+  // skipMultiplier and skipInterval are the same:
+  protected MultiLevelSkipListReader(IndexInput skipStream, int maxSkipLevels, int skipInterval) {
+    this(skipStream, maxSkipLevels, skipInterval, skipInterval);
+  }
   
   /** Returns the id of the doc to which the last call of {@link #skipTo(int)}
    *  has skipped.  */
@@ -157,7 +163,7 @@ public abstract class MultiLevelSkipListReader {
     numSkipped[level] = numSkipped[level + 1] - skipInterval[level + 1];
     skipDoc[level] = lastDoc;
     if (level > 0) {
-        childPointer[level] = skipStream[level].readVLong() + skipPointer[level - 1];
+      childPointer[level] = skipStream[level].readVLong() + skipPointer[level - 1];
     }
   }
 
@@ -187,7 +193,12 @@ public abstract class MultiLevelSkipListReader {
   
   /** Loads the skip levels  */
   private void loadSkipLevels() throws IOException {
-    numberOfSkipLevels = MathUtil.log(docCount, skipInterval[0]);
+    if (docCount <= skipInterval[0]) {
+      numberOfSkipLevels = 1;
+    } else {
+      numberOfSkipLevels = 1+MathUtil.log(docCount/skipInterval[0], skipMultiplier);
+    }
+
     if (numberOfSkipLevels > maxNumberOfSkipLevels) {
       numberOfSkipLevels = maxNumberOfSkipLevels;
     }
@@ -208,7 +219,7 @@ public abstract class MultiLevelSkipListReader {
         toBuffer--;
       } else {
         // clone this stream, it is already at the start of the current level
-        skipStream[i] = (IndexInput) skipStream[0].clone();
+        skipStream[i] = skipStream[0].clone();
         if (inputIsBuffered && length < BufferedIndexInput.BUFFER_SIZE) {
           ((BufferedIndexInput) skipStream[i]).setBufferSize((int) length);
         }

lucene/core/src/java/org/apache/lucene/codecs/MultiLevelSkipListWriter.java

@@ -26,6 +26,8 @@ import org.apache.lucene.util.MathUtil;
 /**
  * This abstract class writes skip lists with multiple levels.
  * 
+ * <pre>
+ *
  * Example for skipInterval = 3:
  *                                                     c            (skip level 2)
  *                 c                 c                 c            (skip level 1) 
@@ -45,6 +47,7 @@ import org.apache.lucene.util.MathUtil;
  * 
  * While this class takes care of writing the different skip levels,
  * subclasses must define the actual format of the skip data.
+ * </pre>
  * @lucene.experimental
  */
 
@@ -54,15 +57,23 @@ public abstract class MultiLevelSkipListWriter {
   
   // the skip interval in the list with level = 0
   private int skipInterval;
+
+  // skipInterval used for level > 0
+  private int skipMultiplier;
   
   // for every skip level a different buffer is used 
   private RAMOutputStream[] skipBuffer;
 
-  protected MultiLevelSkipListWriter(int skipInterval, int maxSkipLevels, int df) {
+  protected MultiLevelSkipListWriter(int skipInterval, int skipMultiplier, int maxSkipLevels, int df) {
     this.skipInterval = skipInterval;
+    this.skipMultiplier = skipMultiplier;
     
     // calculate the maximum number of skip levels for this document frequency
-    numberOfSkipLevels = MathUtil.log(df, skipInterval);
+    if (df <= skipInterval) {
+      numberOfSkipLevels = 1;
+    } else {
+      numberOfSkipLevels = 1+MathUtil.log(df/skipInterval, skipMultiplier);
+    }
     
     // make sure it does not exceed maxSkipLevels
     if (numberOfSkipLevels > maxSkipLevels) {
@@ -70,6 +81,11 @@ public abstract class MultiLevelSkipListWriter {
     }
   }
   
+  // skipMultiplier and skipInterval are the same:
+  protected MultiLevelSkipListWriter(int skipInterval, int maxSkipLevels, int df) {
+    this(skipInterval, skipInterval, maxSkipLevels, df);
+  }
+
   protected void init() {
     skipBuffer = new RAMOutputStream[numberOfSkipLevels];
     for (int i = 0; i < numberOfSkipLevels; i++) {
@@ -95,7 +111,7 @@ public abstract class MultiLevelSkipListWriter {
    * @param skipBuffer the skip buffer to write to
    */
   protected abstract void writeSkipData(int level, IndexOutput skipBuffer) throws IOException;
-  
+
   /**
    * Writes the current skip data to the buffers. The current document frequency determines
    * the max level is skip data is to be written to. 
@@ -104,11 +120,15 @@ public abstract class MultiLevelSkipListWriter {
    * @throws IOException
    */
   public void bufferSkip(int df) throws IOException {
-    int numLevels;
+
+    assert df % skipInterval == 0;
+    int numLevels = 1;
+    df /= skipInterval;
    
     // determine max level
-    for (numLevels = 0; (df % skipInterval) == 0 && numLevels < numberOfSkipLevels; df /= skipInterval) {
+    while ((df % skipMultiplier) == 0 && numLevels < numberOfSkipLevels) {
       numLevels++;
+      df /= skipMultiplier;
     }
     
     long childPointer = 0;
@@ -150,5 +170,4 @@ public abstract class MultiLevelSkipListWriter {
     
     return skipPointer;
   }
-
 }

lucene/core/src/java/org/apache/lucene/codecs/StoredFieldsReader.java

@@ -22,8 +22,8 @@ import java.io.IOException;
 import org.apache.lucene.index.StoredFieldVisitor;
 
 /**
- * Codec API for reading stored fields:
- * 
+ * Codec API for reading stored fields.
+ * <p>
  * You need to implement {@link #visitDocument(int, StoredFieldVisitor)} to
  * read the stored fields for a document, implement {@link #clone()} (creating
  * clones of any IndexInputs used, etc), and {@link #close()}

lucene/core/src/java/org/apache/lucene/codecs/TermVectorsWriter.java

@@ -20,6 +20,7 @@ package org.apache.lucene.codecs;
 import java.io.Closeable;
 import java.io.IOException;
 import java.util.Comparator;
+import java.util.Iterator;
 
 import org.apache.lucene.index.AtomicReader;
 import org.apache.lucene.index.DocsAndPositionsEnum;
@@ -187,19 +188,21 @@ public abstract class TermVectorsWriter implements Closeable {
   }
   
   /** Safe (but, slowish) default method to write every
-   *  vector field in the document.  This default
-   *  implementation requires that the vectors implement
-   *  both Fields.size and
-   *  Terms.size. */
+   *  vector field in the document. */
   protected final void addAllDocVectors(Fields vectors, MergeState mergeState) throws IOException {
     if (vectors == null) {
       startDocument(0);
       return;
     }
 
-    final int numFields = vectors.size();
+    int numFields = vectors.size();
     if (numFields == -1) {
-      throw new IllegalStateException("vectors.size() must be implemented (it returned -1)");
+      // count manually! TODO: Maybe enforce that Fields.size() returns something valid?
+      numFields = 0;
+      for (final Iterator<String> it = vectors.iterator(); it.hasNext(); ) {
+        it.next();
+        numFields++;
+      }
     }
     startDocument(numFields);
     
@@ -208,7 +211,9 @@ public abstract class TermVectorsWriter implements Closeable {
     TermsEnum termsEnum = null;
     DocsAndPositionsEnum docsAndPositionsEnum = null;
     
+    int fieldCount = 0;
     for(String fieldName : vectors) {
+      fieldCount++;
       final FieldInfo fieldInfo = mergeState.fieldInfos.fieldInfo(fieldName);
 
       assert lastFieldName == null || fieldName.compareTo(lastFieldName) > 0: "lastFieldName=" + lastFieldName + " fieldName=" + fieldName;
@@ -225,9 +230,14 @@ public abstract class TermVectorsWriter implements Closeable {
       final boolean hasPayloads = terms.hasPayloads();
       assert !hasPayloads || hasPositions;
       
-      final int numTerms = (int) terms.size();
+      int numTerms = (int) terms.size();
       if (numTerms == -1) {
-        throw new IllegalStateException("terms.size() must be implemented (it returned -1)");
+        // count manually. It is stupid, but needed, as Terms.size() is not a mandatory statistics function
+        numTerms = 0;
+        termsEnum = terms.iterator(termsEnum);
+        while(termsEnum.next() != null) {
+          numTerms++;
+        }
       }
       
       startField(fieldInfo, numTerms, hasPositions, hasOffsets, hasPayloads);
@@ -263,6 +273,7 @@ public abstract class TermVectorsWriter implements Closeable {
       }
       assert termCount == numTerms;
     }
+    assert fieldCount == numFields;
   }
   
   /** Return the BytesRef Comparator used to sort terms

lucene/core/src/java/org/apache/lucene/codecs/TermsIndexReaderBase.java

@@ -51,7 +51,7 @@ public abstract class TermsIndexReaderBase implements Closeable {
   /** 
    * Similar to TermsEnum, except, the only "metadata" it
    * reports for a given indexed term is the long fileOffset
-   * into the main terms dictionary file:
+   * into the main terms dictionary file.
    */
   public static abstract class FieldIndexEnum {
 

lucene/core/src/java/org/apache/lucene/codecs/VariableGapTermsIndexReader.java

@@ -170,7 +170,7 @@ public class VariableGapTermsIndexReader extends TermsIndexReaderBase {
 
     private void loadTermsIndex() throws IOException {
       if (fst == null) {
-        IndexInput clone = (IndexInput) in.clone();
+        IndexInput clone = in.clone();
         clone.seek(indexStart);
         fst = new FST<Long>(clone, fstOutputs);
         clone.close();

lucene/core/src/java/org/apache/lucene/codecs/block/BlockPostingsFormat.java

@@ -0,0 +1,422 @@
+package org.apache.lucene.codecs.block;
+
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.IOException;
+
+import org.apache.lucene.codecs.BlockTreeTermsReader;
+import org.apache.lucene.codecs.BlockTreeTermsWriter;
+import org.apache.lucene.codecs.CodecUtil;
+import org.apache.lucene.codecs.FieldsConsumer;
+import org.apache.lucene.codecs.FieldsProducer;
+import org.apache.lucene.codecs.MultiLevelSkipListWriter;
+import org.apache.lucene.codecs.PostingsFormat;
+import org.apache.lucene.codecs.PostingsReaderBase;
+import org.apache.lucene.codecs.PostingsWriterBase;
+import org.apache.lucene.index.DocsEnum;
+import org.apache.lucene.index.FieldInfo.IndexOptions;
+import org.apache.lucene.index.SegmentReadState;
+import org.apache.lucene.index.SegmentWriteState;
+import org.apache.lucene.store.DataOutput;
+import org.apache.lucene.util.IOUtils;
+import org.apache.lucene.util.packed.PackedInts;
+
+/**
+ * Block postings format, which encodes postings in packed int blocks 
+ * for faster decode.
+ *
+ * <p><b>NOTE</b>: this format is still experimental and
+ * subject to change without backwards compatibility.
+ *
+ * <p>
+ * Basic idea:
+ * <ul>
+ *   <li>
+ *   <b>Packed Block and VInt Block</b>: 
+ *   <p>In packed block, integers are encoded with the same bit width ({@link PackedInts packed format}), 
+ *      the block size (i.e. number of integers inside block) is fixed. </p>
+ *   <p>In VInt block, integers are encoded as {@link DataOutput#writeVInt VInt}, 
+ *      the block size is variable.</p>
+ *   </li>
+ *
+ *   <li> 
+ *   <b>Block structure</b>: 
+ *   <p>When the postings is long enough, BlockPostingsFormat will try to encode most integer data 
+ *      as packed block.</p> 
+ *   <p>Take a term with 259 documents as example, the first 256 document ids are encoded as two packed 
+ *      blocks, while the remaining 3 as one VInt block. </p>
+ *   <p>Different kinds of data are always encoded separately into different packed blocks, but may 
+ *      possible be encoded into a same VInt block. </p>
+ *   <p>This strategy is applied to pairs: 
+ *      &lt;document number, frequency&gt;,
+ *      &lt;position, payload length&gt;, 
+ *      &lt;position, offset start, offset length&gt;, and
+ *      &lt;position, payload length, offsetstart, offset length&gt;.</p>
+ *   </li>
+ *
+ *   <li>
+ *   <b>Skipper setting</b>: 
+ *   <p>The structure of skip table is quite similar to Lucene40PostingsFormat. Skip interval is the 
+ *      same as block size, and each skip entry points to the beginning of each block. However, for 
+ *      the first block, skip data is omitted.</p>
+ *   </li>
+ *
+ *   <li>
+ *   <b>Positions, Payloads, and Offsets</b>: 
+ *   <p>A position is an integer indicating where the term occurs at within one document. 
+ *      A payload is a blob of metadata associated with current position. 
+ *      An offset is a pair of integers indicating the tokenized start/end offsets for given term 
+ *      in current position. </p>
+ *   <p>When payloads and offsets are not omitted, numPositions==numPayloads==numOffsets (assuming a 
+ *      null payload contributes one count). As mentioned in block structure, it is possible to encode 
+ *      these three either combined or separately. 
+ *   <p>For all the cases, payloads and offsets are stored together. When encoded as packed block, 
+ *      position data is separated out as .pos, while payloads and offsets are encoded in .pay (payload 
+ *      metadata will also be stored directly in .pay). When encoded as VInt block, all these three are 
+ *      stored in .pos (so as payload metadata).</p>
+ *   <p>With this strategy, the majority of payload and offset data will be outside .pos file. 
+ *      So for queries that require only position data, running on a full index with payloads and offsets, 
+ *      this reduces disk pre-fetches.</p>
+ *   </li>
+ * </ul>
+ * </p>
+ *
+ * <p>
+ * Files and detailed format:
+ * <ul>
+ *   <li><tt>.tim</tt>: <a href="#Termdictionary">Term Dictionary</a></li>
+ *   <li><tt>.tip</tt>: <a href="#Termindex">Term Index</a></li>
+ *   <li><tt>.doc</tt>: <a href="#Frequencies">Frequencies and Skip Data</a></li>
+ *   <li><tt>.pos</tt>: <a href="#Positions">Positions</a></li>
+ *   <li><tt>.pay</tt>: <a href="#Payloads">Payloads and Offsets</a></li>
+ * </ul>
+ * </p>
+ *
+ * <a name="Termdictionary" id="Termdictionary"></a>
+ * <dl>
+ * <dd>
+ * <b>Term Dictionary</b>
+ *
+ * <p>The .tim file format is quite similar to Lucene40PostingsFormat, 
+ *  with minor difference in MetadataBlock</p>
+ *
+ * <ul>
+ * <!-- TODO: expand on this, its not really correct and doesnt explain sub-blocks etc -->
+ *   <li>TermDictionary(.tim) --&gt; Header, DirOffset, PostingsHeader, PackedBlockSize, 
+ *                                   &lt;Block&gt;<sup>NumBlocks</sup>, FieldSummary</li>
+ *   <li>Block --&gt; SuffixBlock, StatsBlock, MetadataBlock</li>
+ *   <li>SuffixBlock --&gt; EntryCount, SuffixLength, {@link DataOutput#writeByte byte}<sup>SuffixLength</sup></li>
+ *   <li>StatsBlock --&gt; StatsLength, &lt;DocFreq, TotalTermFreq&gt;<sup>EntryCount</sup></li>
+ *   <li>MetadataBlock --&gt; MetaLength, &lt;DocFPDelta, 
+ *                            &lt;PosFPDelta, PosVIntBlockFPDelta?, PayFPDelta?&gt;?, 
+ *                            SkipFPDelta?&gt;<sup>EntryCount</sup></li>
+ *   <li>FieldSummary --&gt; NumFields, &lt;FieldNumber, NumTerms, RootCodeLength, 
+ *                           {@link DataOutput#writeByte byte}<sup>RootCodeLength</sup>, SumDocFreq, DocCount&gt;
+ *                           <sup>NumFields</sup></li>
+ *   <li>Header, PostingsHeader --&gt; {@link CodecUtil#writeHeader CodecHeader}</li>
+ *   <li>DirOffset --&gt; {@link DataOutput#writeLong Uint64}</li>
+ *   <li>PackedBlockSize, EntryCount, SuffixLength, StatsLength, DocFreq, MetaLength, 
+ *       PosVIntBlockFPDelta, SkipFPDelta, NumFields, FieldNumber, RootCodeLength, DocCount --&gt; 
+ *       {@link DataOutput#writeVInt VInt}</li>
+ *   <li>TotalTermFreq, DocFPDelta, PosFPDelta, PayFPDelta, NumTerms, SumTotalTermFreq, SumDocFreq --&gt; 
+ *       {@link DataOutput#writeVLong VLong}</li>
+ * </ul>
+ * <p>Notes:</p>
+ * <ul>
+ *    <li>Here explains MetadataBlock only, other fields are mentioned in 
+ *   <a href="../lucene40/Lucene40PostingsFormat.html#Termdictionary">Lucene40PostingsFormat:TermDictionary</a>
+ *    </li>
+ *    <li>PackedBlockSize is the fixed block size for packed blocks. In packed block, bit width is 
+ *        determined by the largest integer. Smaller block size result in smaller variance among width 
+ *        of integers hence smaller indexes. Larger block size result in more efficient bulk i/o hence
+ *        better acceleration. This value should always be a multiple of 64, currently fixed as 128 as 
+ *        a tradeoff. It is also the skip interval used to accelerate {@link DocsEnum#advance(int)}.
+ *    <li>DocFPDelta determines the position of this term's TermFreqs within the .doc file. 
+ *        In particular, it is the difference of file offset between this term's
+ *        data and previous term's data (or zero, for the first term in the block).On disk it is 
+ *        stored as the difference from previous value in sequence. </li>
+ *    <li>PosFPDelta determines the position of this term's TermPositions within the .pos file.
+ *        While PayFPDelta determines the position of this term's &lt;TermPayloads, TermOffsets?&gt; within 
+ *        the .pay file. Similar to DocFPDelta, it is the difference between two file positions (or 
+ *        neglected, for fields that omit payloads and offsets).</li>
+ *    <li>PosVIntBlockFPDelta determines the position of this term's last TermPosition in last pos packed
+ *        block within the .pos file. It is synonym for PayVIntBlockFPDelta or OffsetVIntBlockFPDelta. 
+ *        This is actually used to indicate whether it is necessary to load following
+ *        payloads and offsets from .pos instead of .pay. Every time a new block of positions are to be 
+ *        loaded, the PostingsReader will use this value to check whether current block is packed format
+ *        or VInt. When packed format, payloads and offsets are fetched from .pay, otherwise from .pos. 
+ *        (this value is neglected when total number of positions i.e. totalTermFreq is less or equal 
+ *        to PackedBlockSize).
+ *    <li>SkipFPDelta determines the position of this term's SkipData within the .doc
+ *        file. In particular, it is the length of the TermFreq data.
+ *        SkipDelta is only stored if DocFreq is not smaller than SkipMinimum
+ *        (i.e. 8 in BlockPostingsFormat).</li>
+ * </ul>
+ * </dd>
+ * </dl>
+ *
+ * <a name="Termindex" id="Termindex"></a>
+ * <dl>
+ * <dd>
+ * <b>Term Index</b>
+ * <p>The .tim file format is mentioned in
+ *   <a href="../lucene40/Lucene40PostingsFormat.html#Termindex">Lucene40PostingsFormat:TermIndex</a>
+ * </dd>
+ * </dl>
+ *
+ *
+ * <a name="Frequencies" id="Frequencies"></a>
+ * <dl>
+ * <dd>
+ * <b>Frequencies and Skip Data</b>
+ *
+ * <p>The .doc file contains the lists of documents which contain each term, along
+ * with the frequency of the term in that document (except when frequencies are
+ * omitted: {@link IndexOptions#DOCS_ONLY}). It also saves skip data to the beginning of 
+ * each packed or VInt block, when the length of document list is larger than packed block size.</p>
+ *
+ * <ul>
+ *   <li>docFile(.doc) --&gt; Header, &lt;TermFreqs, SkipData?&gt;<sup>TermCount</sup></li>
+ *   <li>Header --&gt; {@link CodecUtil#writeHeader CodecHeader}</li>
+ *   <li>TermFreqs --&gt; &lt;PackedBlock&gt; <sup>PackedDocBlockNum</sup>,  
+ *                        VIntBlock? </li>
+ *   <li>PackedBlock --&gt; PackedDocDeltaBlock, PackedFreqBlock?
+ *   <li>VIntBlock --&gt; &lt;DocDelta[, Freq?]&gt;<sup>DocFreq-PackedBlockSize*PackedDocBlockNum</sup>
+ *   <li>SkipData --&gt; &lt;&lt;SkipLevelLength, SkipLevel&gt;
+ *       <sup>NumSkipLevels-1</sup>, SkipLevel&gt;, SkipDatum?</li>
+ *   <li>SkipLevel --&gt; &lt;SkipDatum&gt; <sup>TrimmedDocFreq/(PackedBlockSize^(Level + 1))</sup></li>
+ *   <li>SkipDatum --&gt; DocSkip, DocFPSkip, &lt;PosFPSkip, PosBlockOffset, PayLength?, 
+ *                        OffsetStart?, PayFPSkip?&gt;?, SkipChildLevelPointer?</li>
+ *   <li>PackedDocDeltaBlock, PackedFreqBlock --&gt; {@link PackedInts PackedInts}</li>
+ *   <li>DocDelta, Freq, DocSkip, DocFPSkip, PosFPSkip, PosBlockOffset, PayLength, OffsetStart, PayFPSkip 
+ *       --&gt; 
+ *   {@link DataOutput#writeVInt VInt}</li>
+ *   <li>SkipChildLevelPointer --&gt; {@link DataOutput#writeVLong VLong}</li>
+ * </ul>
+ * <p>Notes:</p>
+ * <ul>
+ *   <li>PackedDocDeltaBlock is theoretically generated from two steps: 
+ *     <ol>
+ *       <li>Calculate the difference between each document number and previous one, 
+ *           and get a d-gaps list (for the first document, use absolute value); </li>
+ *       <li>For those d-gaps from first one to PackedDocBlockNum*PackedBlockSize<sup>th</sup>, 
+ *           separately encode as packed blocks.</li>
+ *     </ol>
+ *     If frequencies are not omitted, PackedFreqBlock will be generated without d-gap step.
+ *   </li>
+ *   <li>VIntBlock stores remaining d-gaps (along with frequencies when possible) with a format 
+ *       mentioned in
+ *   <a href="../lucene40/Lucene40PostingsFormat.html#Frequencies">Lucene40PostingsFormat:Frequencies</a>
+ *   </li>
+ *   <li>PackedDocBlockNum is the number of packed blocks for current term's docids or frequencies. 
+ *       In particular, PackedDocBlockNum = floor(DocFreq/PackedBlockSize) </li>
+ *   <li>TrimmedDocFreq = DocFreq % PackedBlockSize == 0 ? DocFreq - 1 : DocFreq. 
+ *       We use this trick since the definition of skip entry is a little different from base interface.
+ *       In {@link MultiLevelSkipListWriter}, skip data is assumed to be saved for
+ *       skipInterval<sup>th</sup>, 2*skipInterval<sup>th</sup> ... posting in the list. However, 
+ *       in BlockPostingsFormat, the skip data is saved for skipInterval+1<sup>th</sup>, 
+ *       2*skipInterval+1<sup>th</sup> ... posting (skipInterval==PackedBlockSize in this case). 
+ *       When DocFreq is multiple of PackedBlockSize, MultiLevelSkipListWriter will expect one 
+ *       more skip data than BlockSkipWriter. </li>
+ *   <li>SkipDatum is the metadata of one skip entry.
+ *      For the first block (no matter packed or VInt), it is omitted.</li>
+ *   <li>DocSkip records the document number of every PackedBlockSize<sup>th</sup> document number in
+ *       the postings (i.e. last document number in each packed block). On disk it is stored as the 
+ *       difference from previous value in the sequence. </li>
+ *   <li>DocFPSkip records the file offsets of each block (excluding )posting at 
+ *       PackedBlockSize+1<sup>th</sup>, 2*PackedBlockSize+1<sup>th</sup> ... , in DocFile. 
+ *       The file offsets are relative to the start of current term's TermFreqs. 
+ *       On disk it is also stored as the difference from previous SkipDatum in the sequence.</li>
+ *   <li>Since positions and payloads are also block encoded, the skip should skip to related block first,
+ *       then fetch the values according to in-block offset. PosFPSkip and PayFPSkip record the file 
+ *       offsets of related block in .pos and .pay, respectively. While PosBlockOffset indicates
+ *       which value to fetch inside the related block (PayBlockOffset is unnecessary since it is always
+ *       equal to PosBlockOffset). Same as DocFPSkip, the file offsets are relative to the start of 
+ *       current term's TermFreqs, and stored as a difference sequence.</li>
+ *   <li>PayLength indicates the length of last payload.</li>
+ *   <li>OffsetStart indicates the first value of last offset pair.</li>
+ * </ul>
+ * </dd>
+ * </dl>
+ *
+ * <a name="Positions" id="Positions"></a>
+ * <dl>
+ * <dd>
+ * <b>Positions</b>
+ * <p>The .pos file contains the lists of positions that each term occurs at within documents. It also
+ *    sometimes stores part of payloads and offsets for speedup.</p>
+ * <ul>
+ *   <li>PosFile(.pos) --&gt; Header, &lt;TermPositions&gt; <sup>TermCount</sup></li>
+ *   <li>Header --&gt; {@link CodecUtil#writeHeader CodecHeader}</li>
+ *   <li>TermPositions --&gt; &lt;PackedPosDeltaBlock&gt; <sup>PackedPosBlockNum</sup>,  
+ *                            VIntBlock? </li>
+ *   <li>VIntBlock --&gt; PosVIntCount, &lt;PosDelta[, PayLength?], PayData?, 
+ *                        OffsetStartDelta?, OffsetLength?&gt;<sup>PosVIntCount</sup>
+ *   <li>PackedPosDeltaBlock --&gt; {@link PackedInts PackedInts}</li>
+ *   <li>PosVIntCount, PosDelta, OffsetStartDelta, OffsetLength --&gt; 
+ *       {@link DataOutput#writeVInt VInt}</li>
+ *   <li>PayData --&gt; {@link DataOutput#writeByte byte}<sup>PayLength</sup></li>
+ * </ul>
+ * <p>Notes:</p>
+ * <ul>
+ *   <li>TermPositions are order by term (terms are implicit, from the term dictionary), and position 
+ *       values for each term document pair are incremental, and ordered by document number.</li>
+ *   <li>PackedPosBlockNum is the number of packed blocks for current term's positions, payloads or offsets. 
+ *       In particular, PackedPosBlockNum = floor(totalTermFreq/PackedBlockSize) </li>
+ *   <li>PosVIntCount is the number of positions encoded as VInt format. In particular, 
+ *       PosVIntCount = totalTermFreq - PackedPosBlockNum*PackedBlockSize</li>
+ *   <li>The procedure how PackedPosDeltaBlock is generated is the same as PackedDocDeltaBlock 
+ *       in chapter <a href="#Frequencies">Frequencies and Skip Data</a>.</li>
+ *   <li>PosDelta is the same as the format mentioned in 
+ *   <a href="../lucene40/Lucene40PostingsFormat.html#Positions">Lucene40PostingsFormat:Positions</a>
+ *   </li>
+ *   <li>OffsetStartDelta is the difference between this position's startOffset from the previous 
+ *       occurrence (or zero, if this is the first occurrence in this document).</li>
+ *   <li>OffsetLength indicates the length of the current offset (endOffset-startOffset).</li>
+ *   <li>PayloadData is the blob of metadata associated with current position.</li>
+ * </ul>
+ * </dd>
+ * </dl>
+ *
+ * <a name="Payloads" id="Payloads"></a>
+ * <dl>
+ * <dd>
+ * <b>Payloads and Offsets</b>
+ * <p>The .pay file will store payloads and offsets associated with certain term-document positions. 
+ *    Some payloads and offsets will be separated out into .pos file, for speedup reason.</p>
+ * <ul>
+ *   <li>PayFile(.pay): --&gt; Header, &lt;TermPayloads, TermOffsets?&gt; <sup>TermCount</sup></li>
+ *   <li>Header --&gt; {@link CodecUtil#writeHeader CodecHeader}</li>
+ *   <li>TermPayloads --&gt; &lt;PackedPayLengthBlock, SumPayLength, PayData&gt; <sup>PackedPayBlockNum</sup>
+ *   <li>TermOffsets --&gt; &lt;PackedOffsetStartDeltaBlock, PackedOffsetLengthBlock&gt; <sup>PackedPayBlockNum</sup>
+ *   <li>PackedPayLengthBlock, PackedOffsetStartDeltaBlock, PackedOffsetLengthBlock --&gt; {@link PackedInts PackedInts}</li>
+ *   <li>SumPayLength --&gt; {@link DataOutput#writeVInt VInt}</li>
+ *   <li>PayData --&gt; {@link DataOutput#writeByte byte}<sup>SumPayLength</sup></li>
+ * </ul>
+ * <p>Notes:</p>
+ * <ul>
+ *   <li>The order of TermPayloads/TermOffsets will be the same as TermPositions, note that part of 
+ *       payload/offsets are stored in .pos.</li>
+ *   <li>The procedure how PackedPayLengthBlock and PackedOffsetLengthBlock are generated is the 
+ *       same as PackedFreqBlock in chapter <a href="#Frequencies">Frequencies and Skip Data</a>. 
+ *       While PackedStartDeltaBlock follows a same procedure as PackedDocDeltaBlock.</li>
+ *   <li>PackedPayBlockNum is always equal to PackedPosBlockNum, for the same term. It is also synonym 
+ *       for PackedOffsetBlockNum.</li>
+ *   <li>SumPayLength is the total length of payloads written within one block, should be the sum
+ *       of PayLengths in one packed block.</li>
+ *   <li>PayLength in PackedPayLengthBlock is the length of each payload, associated with current 
+ *       position.</li>
+ * </ul>
+ * </dd>
+ * </dl>
+ * </p>
+ *
+ * @lucene.experimental
+ */
+
+public final class BlockPostingsFormat extends PostingsFormat {
+  /**
+   * Filename extension for document number, frequencies, and skip data.
+   * See chapter: <a href="#Frequencies">Frequencies and Skip Data</a>
+   */
+  public static final String DOC_EXTENSION = "doc";
+
+  /**
+   * Filename extension for positions. 
+   * See chapter: <a href="#Positions">Positions</a>
+   */
+  public static final String POS_EXTENSION = "pos";
+
+  /**
+   * Filename extension for payloads and offsets.
+   * See chapter: <a href="#Payloads">Payloads and Offsets</a>
+   */
+  public static final String PAY_EXTENSION = "pay";
+
+  private final int minTermBlockSize;
+  private final int maxTermBlockSize;
+
+  /**
+   * Fixed packed block size, number of integers encoded in 
+   * a single packed block.
+   */
+  // NOTE: must be multiple of 64 because of PackedInts long-aligned encoding/decoding
+  public final static int BLOCK_SIZE = 128;
+
+  public BlockPostingsFormat() {
+    this(BlockTreeTermsWriter.DEFAULT_MIN_BLOCK_SIZE, BlockTreeTermsWriter.DEFAULT_MAX_BLOCK_SIZE);
+  }
+
+  public BlockPostingsFormat(int minTermBlockSize, int maxTermBlockSize) {
+    super("Block");
+    this.minTermBlockSize = minTermBlockSize;
+    assert minTermBlockSize > 1;
+    this.maxTermBlockSize = maxTermBlockSize;
+    assert minTermBlockSize <= maxTermBlockSize;
+  }
+
+  @Override
+  public String toString() {
+    return getName() + "(blocksize=" + BLOCK_SIZE + ")";
+  }
+
+  @Override
+  public FieldsConsumer fieldsConsumer(SegmentWriteState state) throws IOException {
+    PostingsWriterBase postingsWriter = new BlockPostingsWriter(state);
+
+    boolean success = false;
+    try {
+      FieldsConsumer ret = new BlockTreeTermsWriter(state, 
+                                                    postingsWriter,
+                                                    minTermBlockSize, 
+                                                    maxTermBlockSize);
+      success = true;
+      return ret;
+    } finally {
+      if (!success) {
+        IOUtils.closeWhileHandlingException(postingsWriter);
+      }
+    }
+  }
+
+  @Override
+  public FieldsProducer fieldsProducer(SegmentReadState state) throws IOException {
+    PostingsReaderBase postingsReader = new BlockPostingsReader(state.dir,
+                                                                state.fieldInfos,
+                                                                state.segmentInfo,
+                                                                state.context,
+                                                                state.segmentSuffix);
+    boolean success = false;
+    try {
+      FieldsProducer ret = new BlockTreeTermsReader(state.dir,
+                                                    state.fieldInfos,
+                                                    state.segmentInfo.name,
+                                                    postingsReader,
+                                                    state.context,
+                                                    state.segmentSuffix,
+                                                    state.termsIndexDivisor);
+      success = true;
+      return ret;
+    } finally {
+      if (!success) {
+        IOUtils.closeWhileHandlingException(postingsReader);
+      }
+    }
+  }
+}

lucene/core/src/java/org/apache/lucene/codecs/block/BlockPostingsWriter.java

@@ -0,0 +1,563 @@
+package org.apache.lucene.codecs.block;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import static org.apache.lucene.codecs.block.BlockPostingsFormat.BLOCK_SIZE;
+import static org.apache.lucene.codecs.block.ForUtil.MAX_DATA_SIZE;
+import static org.apache.lucene.codecs.block.ForUtil.MAX_ENCODED_SIZE;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.lucene.codecs.CodecUtil;
+import org.apache.lucene.codecs.PostingsWriterBase;
+import org.apache.lucene.codecs.TermStats;
+import org.apache.lucene.index.CorruptIndexException;
+import org.apache.lucene.index.FieldInfo;
+import org.apache.lucene.index.FieldInfo.IndexOptions;
+import org.apache.lucene.index.IndexFileNames;
+import org.apache.lucene.index.SegmentWriteState;
+import org.apache.lucene.store.IndexOutput;
+import org.apache.lucene.store.RAMOutputStream;
+import org.apache.lucene.util.ArrayUtil;
+import org.apache.lucene.util.BytesRef;
+import org.apache.lucene.util.IOUtils;
+import org.apache.lucene.util.packed.PackedInts;
+
+
+/**
+ * Concrete class that writes docId(maybe frq,pos,offset,payloads) list
+ * with postings format.
+ *
+ * Postings list for each term will be stored separately. 
+ *
+ * @see BlockSkipWriter for details about skipping setting and postings layout.
+ *
+ */
+final class BlockPostingsWriter extends PostingsWriterBase {
+
+  /** 
+   * Expert: The maximum number of skip levels. Smaller values result in 
+   * slightly smaller indexes, but slower skipping in big posting lists.
+   */
+  static final int maxSkipLevels = 10;
+
+  final static String TERMS_CODEC = "BlockPostingsWriterTerms";
+  final static String DOC_CODEC = "BlockPostingsWriterDoc";
+  final static String POS_CODEC = "BlockPostingsWriterPos";
+  final static String PAY_CODEC = "BlockPostingsWriterPay";
+
+  // Increment version to change it:
+  final static int VERSION_START = 0;
+  final static int VERSION_CURRENT = VERSION_START;
+
+  final IndexOutput docOut;
+  final IndexOutput posOut;
+  final IndexOutput payOut;
+
+  private IndexOutput termsOut;
+
+  // How current field indexes postings:
+  private boolean fieldHasFreqs;
+  private boolean fieldHasPositions;
+  private boolean fieldHasOffsets;
+  private boolean fieldHasPayloads;
+
+  // Holds starting file pointers for each term:
+  private long docTermStartFP;
+  private long posTermStartFP;
+  private long payTermStartFP;
+
+  final int[] docDeltaBuffer;
+  final int[] freqBuffer;
+  private int docBufferUpto;
+
+  final int[] posDeltaBuffer;
+  final int[] payloadLengthBuffer;
+  final int[] offsetStartDeltaBuffer;
+  final int[] offsetLengthBuffer;
+  private int posBufferUpto;
+
+  private byte[] payloadBytes;
+  private int payloadByteUpto;
+
+  private int lastBlockDocID;
+  private long lastBlockPosFP;
+  private long lastBlockPayFP;
+  private int lastBlockPosBufferUpto;
+  private int lastBlockStartOffset;
+  private int lastBlockPayloadByteUpto;
+
+  private int lastDocID;
+  private int lastPosition;
+  private int lastStartOffset;
+  private int docCount;
+
+  final byte[] encoded;
+
+  private final ForUtil forUtil;
+  private final BlockSkipWriter skipWriter;
+  
+  public BlockPostingsWriter(SegmentWriteState state, float acceptableOverheadRatio) throws IOException {
+    super();
+
+    docOut = state.directory.createOutput(IndexFileNames.segmentFileName(state.segmentInfo.name, state.segmentSuffix, BlockPostingsFormat.DOC_EXTENSION),
+                                          state.context);
+    IndexOutput posOut = null;
+    IndexOutput payOut = null;
+    boolean success = false;
+    try {
+      CodecUtil.writeHeader(docOut, DOC_CODEC, VERSION_CURRENT);
+      forUtil = new ForUtil(acceptableOverheadRatio, docOut);
+      if (state.fieldInfos.hasProx()) {
+        posDeltaBuffer = new int[MAX_DATA_SIZE];
+        posOut = state.directory.createOutput(IndexFileNames.segmentFileName(state.segmentInfo.name, state.segmentSuffix, BlockPostingsFormat.POS_EXTENSION),
+                                              state.context);
+        CodecUtil.writeHeader(posOut, POS_CODEC, VERSION_CURRENT);
+
+        if (state.fieldInfos.hasPayloads()) {
+          payloadBytes = new byte[128];
+          payloadLengthBuffer = new int[MAX_DATA_SIZE];
+        } else {
+          payloadBytes = null;
+          payloadLengthBuffer = null;
+        }
+
+        if (state.fieldInfos.hasOffsets()) {
+          offsetStartDeltaBuffer = new int[MAX_DATA_SIZE];
+          offsetLengthBuffer = new int[MAX_DATA_SIZE];
+        } else {
+          offsetStartDeltaBuffer = null;
+          offsetLengthBuffer = null;
+        }
+
+        if (state.fieldInfos.hasPayloads() || state.fieldInfos.hasOffsets()) {
+          payOut = state.directory.createOutput(IndexFileNames.segmentFileName(state.segmentInfo.name, state.segmentSuffix, BlockPostingsFormat.PAY_EXTENSION),
+                                                state.context);
+          CodecUtil.writeHeader(payOut, PAY_CODEC, VERSION_CURRENT);
+        }
+      } else {
+        posDeltaBuffer = null;
+        payloadLengthBuffer = null;
+        offsetStartDeltaBuffer = null;
+        offsetLengthBuffer = null;
+        payloadBytes = null;
+      }
+      this.payOut = payOut;
+      this.posOut = posOut;
+      success = true;
+    } finally {
+      if (!success) {
+        IOUtils.closeWhileHandlingException(docOut, posOut, payOut);
+      }
+    }
+
+    docDeltaBuffer = new int[MAX_DATA_SIZE];
+    freqBuffer = new int[MAX_DATA_SIZE];
+
+    // TODO: should we try skipping every 2/4 blocks...?
+    skipWriter = new BlockSkipWriter(maxSkipLevels,
+                                     BLOCK_SIZE, 
+                                     state.segmentInfo.getDocCount(),
+                                     docOut,
+                                     posOut,
+                                     payOut);
+
+    encoded = new byte[MAX_ENCODED_SIZE];
+  }
+
+  public BlockPostingsWriter(SegmentWriteState state) throws IOException {
+    this(state, PackedInts.COMPACT);
+  }
+
+  @Override
+  public void start(IndexOutput termsOut) throws IOException {
+    this.termsOut = termsOut;
+    CodecUtil.writeHeader(termsOut, TERMS_CODEC, VERSION_CURRENT);
+    termsOut.writeVInt(BLOCK_SIZE);
+  }
+
+  @Override
+  public void setField(FieldInfo fieldInfo) {
+    IndexOptions indexOptions = fieldInfo.getIndexOptions();
+    fieldHasFreqs = indexOptions.compareTo(IndexOptions.DOCS_AND_FREQS) >= 0;
+    fieldHasPositions = indexOptions.compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) >= 0;
+    fieldHasOffsets = indexOptions.compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS_AND_OFFSETS) >= 0;
+    fieldHasPayloads = fieldInfo.hasPayloads();
+    skipWriter.setField(fieldHasPositions, fieldHasOffsets, fieldHasPayloads);
+  }
+
+  @Override
+  public void startTerm() {
+    docTermStartFP = docOut.getFilePointer();
+    if (fieldHasPositions) {
+      posTermStartFP = posOut.getFilePointer();
+      if (fieldHasPayloads || fieldHasOffsets) {
+        payTermStartFP = payOut.getFilePointer();
+      }
+    }
+    lastDocID = 0;
+    lastBlockDocID = -1;
+    // if (DEBUG) {
+    //   System.out.println("FPW.startTerm startFP=" + docTermStartFP);
+    // }
+    skipWriter.resetSkip();
+  }
+
+  @Override
+  public void startDoc(int docID, int termDocFreq) throws IOException {
+    // if (DEBUG) {
+    //   System.out.println("FPW.startDoc docID["+docBufferUpto+"]=" + docID);
+    // }
+    // Have collected a block of docs, and get a new doc. 
+    // Should write skip data as well as postings list for
+    // current block.
+    if (lastBlockDocID != -1 && docBufferUpto == 0) {
+      // if (DEBUG) {
+      //   System.out.println("  bufferSkip at writeBlock: lastDocID=" + lastBlockDocID + " docCount=" + (docCount-1));
+      // }
+      skipWriter.bufferSkip(lastBlockDocID, docCount, lastBlockPosFP, lastBlockPayFP, lastBlockPosBufferUpto, lastBlockStartOffset, lastBlockPayloadByteUpto);
+    }
+
+    final int docDelta = docID - lastDocID;
+
+    if (docID < 0 || (docCount > 0 && docDelta <= 0)) {
+      throw new CorruptIndexException("docs out of order (" + docID + " <= " + lastDocID + " ) (docOut: " + docOut + ")");
+    }
+
+    docDeltaBuffer[docBufferUpto] = docDelta;
+    // if (DEBUG) {
+    //   System.out.println("  docDeltaBuffer[" + docBufferUpto + "]=" + docDelta);
+    // }
+    if (fieldHasFreqs) {
+      freqBuffer[docBufferUpto] = termDocFreq;
+    }
+    docBufferUpto++;
+    docCount++;
+
+    if (docBufferUpto == BLOCK_SIZE) {
+      // if (DEBUG) {
+      //   System.out.println("  write docDelta block @ fp=" + docOut.getFilePointer());
+      // }
+      forUtil.writeBlock(docDeltaBuffer, encoded, docOut);
+      if (fieldHasFreqs) {
+        // if (DEBUG) {
+        //   System.out.println("  write freq block @ fp=" + docOut.getFilePointer());
+        // }
+        forUtil.writeBlock(freqBuffer, encoded, docOut);
+      }
+      // NOTE: don't set docBufferUpto back to 0 here;
+      // finishDoc will do so (because it needs to see that
+      // the block was filled so it can save skip data)
+    }
+
+
+    lastDocID = docID;
+    lastPosition = 0;
+    lastStartOffset = 0;
+  }
+
+  /** Add a new position & payload */
+  @Override
+  public void addPosition(int position, BytesRef payload, int startOffset, int endOffset) throws IOException {
+    // if (DEBUG) {
+    //   System.out.println("FPW.addPosition pos=" + position + " posBufferUpto=" + posBufferUpto + (fieldHasPayloads ? " payloadByteUpto=" + payloadByteUpto: ""));
+    // }
+    posDeltaBuffer[posBufferUpto] = position - lastPosition;
+    if (fieldHasPayloads) {
+      if (payload == null || payload.length == 0) {
+        // no payload
+        payloadLengthBuffer[posBufferUpto] = 0;
+      } else {
+        payloadLengthBuffer[posBufferUpto] = payload.length;
+        if (payloadByteUpto + payload.length > payloadBytes.length) {
+          payloadBytes = ArrayUtil.grow(payloadBytes, payloadByteUpto + payload.length);
+        }
+        System.arraycopy(payload.bytes, payload.offset, payloadBytes, payloadByteUpto, payload.length);
+        payloadByteUpto += payload.length;
+      }
+    }
+
+    if (fieldHasOffsets) {
+      assert startOffset >= lastStartOffset;
+      assert endOffset >= startOffset;
+      offsetStartDeltaBuffer[posBufferUpto] = startOffset - lastStartOffset;
+      offsetLengthBuffer[posBufferUpto] = endOffset - startOffset;
+      lastStartOffset = startOffset;
+    }
+    
+    posBufferUpto++;
+    lastPosition = position;
+    if (posBufferUpto == BLOCK_SIZE) {
+      // if (DEBUG) {
+      //   System.out.println("  write pos bulk block @ fp=" + posOut.getFilePointer());
+      // }
+      forUtil.writeBlock(posDeltaBuffer, encoded, posOut);
+
+      if (fieldHasPayloads) {
+        forUtil.writeBlock(payloadLengthBuffer, encoded, payOut);
+        payOut.writeVInt(payloadByteUpto);
+        payOut.writeBytes(payloadBytes, 0, payloadByteUpto);
+        payloadByteUpto = 0;
+      }
+      if (fieldHasOffsets) {
+        forUtil.writeBlock(offsetStartDeltaBuffer, encoded, payOut);
+        forUtil.writeBlock(offsetLengthBuffer, encoded, payOut);
+      }
+      posBufferUpto = 0;
+    }
+  }
+
+  @Override
+  public void finishDoc() throws IOException {
+    // Since we don't know df for current term, we had to buffer
+    // those skip data for each block, and when a new doc comes, 
+    // write them to skip file.
+    if (docBufferUpto == BLOCK_SIZE) {
+      lastBlockDocID = lastDocID;
+      if (posOut != null) {
+        if (payOut != null) {
+          lastBlockPayFP = payOut.getFilePointer();
+        }
+        lastBlockPosFP = posOut.getFilePointer();
+        lastBlockPosBufferUpto = posBufferUpto;
+        lastBlockStartOffset = lastStartOffset;
+        lastBlockPayloadByteUpto = payloadByteUpto;
+      }
+      // if (DEBUG) {
+      //   System.out.println("  docBufferUpto="+docBufferUpto+" now get lastBlockDocID="+lastBlockDocID+" lastBlockPosFP=" + lastBlockPosFP + " lastBlockPosBufferUpto=" +  lastBlockPosBufferUpto + " lastBlockPayloadByteUpto=" + lastBlockPayloadByteUpto);
+      // }
+      docBufferUpto = 0;
+    }
+  }
+
+  private static class PendingTerm {
+    public final long docStartFP;
+    public final long posStartFP;
+    public final long payStartFP;
+    public final int skipOffset;
+    public final int lastPosBlockOffset;
+
+    public PendingTerm(long docStartFP, long posStartFP, long payStartFP, int skipOffset, int lastPosBlockOffset) {
+      this.docStartFP = docStartFP;
+      this.posStartFP = posStartFP;
+      this.payStartFP = payStartFP;
+      this.skipOffset = skipOffset;
+      this.lastPosBlockOffset = lastPosBlockOffset;
+    }
+  }
+
+  private final List<PendingTerm> pendingTerms = new ArrayList<PendingTerm>();
+
+  /** Called when we are done adding docs to this term */
+  @Override
+  public void finishTerm(TermStats stats) throws IOException {
+    assert stats.docFreq > 0;
+
+    // TODO: wasteful we are counting this (counting # docs
+    // for this term) in two places?
+    assert stats.docFreq == docCount: stats.docFreq + " vs " + docCount;
+
+    // if (DEBUG) {
+    //   System.out.println("FPW.finishTerm docFreq=" + stats.docFreq);
+    // }
+
+    // if (DEBUG) {
+    //   if (docBufferUpto > 0) {
+    //     System.out.println("  write doc/freq vInt block (count=" + docBufferUpto + ") at fp=" + docOut.getFilePointer() + " docTermStartFP=" + docTermStartFP);
+    //   }
+    // }
+
+    // vInt encode the remaining doc deltas and freqs:
+    for(int i=0;i<docBufferUpto;i++) {
+      final int docDelta = docDeltaBuffer[i];
+      final int freq = freqBuffer[i];
+      if (!fieldHasFreqs) {
+        docOut.writeVInt(docDelta);
+      } else if (freqBuffer[i] == 1) {
+        docOut.writeVInt((docDelta<<1)|1);
+      } else {
+        docOut.writeVInt(docDelta<<1);
+        docOut.writeVInt(freq);
+      }
+    }
+
+    final int lastPosBlockOffset;
+
+    if (fieldHasPositions) {
+      // if (DEBUG) {
+      //   if (posBufferUpto > 0) {
+      //     System.out.println("  write pos vInt block (count=" + posBufferUpto + ") at fp=" + posOut.getFilePointer() + " posTermStartFP=" + posTermStartFP + " hasPayloads=" + fieldHasPayloads + " hasOffsets=" + fieldHasOffsets);
+      //   }
+      // }
+
+      // totalTermFreq is just total number of positions(or payloads, or offsets)
+      // associated with current term.
+      assert stats.totalTermFreq != -1;
+      if (stats.totalTermFreq > BLOCK_SIZE) {
+        // record file offset for last pos in last block
+        lastPosBlockOffset = (int) (posOut.getFilePointer() - posTermStartFP);
+      } else {
+        lastPosBlockOffset = -1;
+      }
+      if (posBufferUpto > 0) {
+        posOut.writeVInt(posBufferUpto);
+        
+        // TODO: should we send offsets/payloads to
+        // .pay...?  seems wasteful (have to store extra
+        // vLong for low (< BLOCK_SIZE) DF terms = vast vast
+        // majority)
+
+        // vInt encode the remaining positions/payloads/offsets:
+        int lastPayloadLength = -1;
+        int payloadBytesReadUpto = 0;
+        for(int i=0;i<posBufferUpto;i++) {
+          final int posDelta = posDeltaBuffer[i];
+          if (fieldHasPayloads) {
+            final int payloadLength = payloadLengthBuffer[i];
+            if (payloadLength != lastPayloadLength) {
+              lastPayloadLength = payloadLength;
+              posOut.writeVInt((posDelta<<1)|1);
+              posOut.writeVInt(payloadLength);
+            } else {
+              posOut.writeVInt(posDelta<<1);
+            }
+
+            // if (DEBUG) {
+            //   System.out.println("        i=" + i + " payloadLen=" + payloadLength);
+            // }
+
+            if (payloadLength != 0) {
+              // if (DEBUG) {
+              //   System.out.println("          write payload @ pos.fp=" + posOut.getFilePointer());
+              // }
+              posOut.writeBytes(payloadBytes, payloadBytesReadUpto, payloadLength);
+              payloadBytesReadUpto += payloadLength;
+            }
+          } else {
+            posOut.writeVInt(posDelta);
+          }
+
+          if (fieldHasOffsets) {
+            // if (DEBUG) {
+            //   System.out.println("          write offset @ pos.fp=" + posOut.getFilePointer());
+            // }
+            posOut.writeVInt(offsetStartDeltaBuffer[i]);
+            posOut.writeVInt(offsetLengthBuffer[i]);
+          }
+        }
+
+        if (fieldHasPayloads) {
+          assert payloadBytesReadUpto == payloadByteUpto;
+          payloadByteUpto = 0;
+        }
+      }
+      // if (DEBUG) {
+      //   System.out.println("  totalTermFreq=" + stats.totalTermFreq + " lastPosBlockOffset=" + lastPosBlockOffset);
+      // }
+    } else {
+      lastPosBlockOffset = -1;
+    }
+
+    int skipOffset;
+    if (docCount > BLOCK_SIZE) {
+      skipOffset = (int) (skipWriter.writeSkip(docOut) - docTermStartFP);
+      
+      // if (DEBUG) {
+      //   System.out.println("skip packet " + (docOut.getFilePointer() - (docTermStartFP + skipOffset)) + " bytes");
+      // }
+    } else {
+      skipOffset = -1;
+      // if (DEBUG) {
+      //   System.out.println("  no skip: docCount=" + docCount);
+      // }
+    }
+
+    long payStartFP;
+    if (stats.totalTermFreq >= BLOCK_SIZE) {
+      payStartFP = payTermStartFP;
+    } else {
+      payStartFP = -1;
+    }
+
+    // if (DEBUG) {
+    //   System.out.println("  payStartFP=" + payStartFP);
+    // }
+
+    pendingTerms.add(new PendingTerm(docTermStartFP, posTermStartFP, payStartFP, skipOffset, lastPosBlockOffset));
+    docBufferUpto = 0;
+    posBufferUpto = 0;
+    lastDocID = 0;
+    docCount = 0;
+  }
+
+  private final RAMOutputStream bytesWriter = new RAMOutputStream();
+
+  @Override
+  public void flushTermsBlock(int start, int count) throws IOException {
+
+    if (count == 0) {
+      termsOut.writeByte((byte) 0);
+      return;
+    }
+
+    assert start <= pendingTerms.size();
+    assert count <= start;
+
+    final int limit = pendingTerms.size() - start + count;
+
+    long lastDocStartFP = 0;
+    long lastPosStartFP = 0;
+    long lastPayStartFP = 0;
+    for(int idx=limit-count; idx<limit; idx++) {
+      PendingTerm term = pendingTerms.get(idx);
+
+      bytesWriter.writeVLong(term.docStartFP - lastDocStartFP);
+      lastDocStartFP = term.docStartFP;
+
+      if (fieldHasPositions) {
+        bytesWriter.writeVLong(term.posStartFP - lastPosStartFP);
+        lastPosStartFP = term.posStartFP;
+        if (term.lastPosBlockOffset != -1) {
+          bytesWriter.writeVInt(term.lastPosBlockOffset);
+        }
+        if ((fieldHasPayloads || fieldHasOffsets) && term.payStartFP != -1) {
+          bytesWriter.writeVLong(term.payStartFP - lastPayStartFP);
+          lastPayStartFP = term.payStartFP;
+        }
+      }
+
+      if (term.skipOffset != -1) {
+        bytesWriter.writeVInt(term.skipOffset);
+      }
+    }
+
+    termsOut.writeVInt((int) bytesWriter.getFilePointer());
+    bytesWriter.writeTo(termsOut);
+    bytesWriter.reset();
+
+    // Remove the terms we just wrote:
+    pendingTerms.subList(limit-count, limit).clear();
+  }
+
+  @Override
+  public void close() throws IOException {
+    IOUtils.close(docOut, posOut, payOut);
+  }
+}

lucene/core/src/java/org/apache/lucene/codecs/block/BlockSkipReader.java

@@ -0,0 +1,244 @@
+package org.apache.lucene.codecs.block;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.IOException;
+import java.util.Arrays;
+
+import org.apache.lucene.codecs.MultiLevelSkipListReader;
+import org.apache.lucene.store.IndexInput;
+
+/**
+ * Implements the skip list reader for block postings format
+ * that stores positions and payloads.
+ * 
+ * Although this skipper uses MultiLevelSkipListReader as an interface, 
+ * its definition of skip position will be a little different. 
+ *
+ * For example, when skipInterval = blockSize = 3, df = 2*skipInterval = 6, 
+ * 
+ * 0 1 2 3 4 5
+ * d d d d d d    (posting list)
+ *     ^     ^    (skip point in MultiLeveSkipWriter)
+ *       ^        (skip point in BlockSkipWriter)
+ *
+ * In this case, MultiLevelSkipListReader will use the last document as a skip point, 
+ * while BlockSkipReader should assume no skip point will comes. 
+ *
+ * If we use the interface directly in BlockSkipReader, it may silly try to read 
+ * another skip data after the only skip point is loaded. 
+ *
+ * To illustrate this, we can call skipTo(d[5]), since skip point d[3] has smaller docId,
+ * and numSkipped+blockSize== df, the MultiLevelSkipListReader will assume the skip list
+ * isn't exhausted yet, and try to load a non-existed skip point
+ *
+ * Therefore, we'll trim df before passing it to the interface. see trim(int)
+ *
+ */
+final class BlockSkipReader extends MultiLevelSkipListReader {
+  // private boolean DEBUG = BlockPostingsReader.DEBUG;
+  private final int blockSize;
+
+  private long docPointer[];
+  private long posPointer[];
+  private long payPointer[];
+  private int posBufferUpto[];
+  private int startOffset[];
+  private int payloadByteUpto[];
+
+  private long lastPosPointer;
+  private long lastPayPointer;
+  private int lastStartOffset;
+  private int lastPayloadByteUpto;
+  private long lastDocPointer;
+  private int lastPosBufferUpto;
+
+  public BlockSkipReader(IndexInput skipStream, int maxSkipLevels, int blockSize, boolean hasPos, boolean hasOffsets, boolean hasPayloads) {
+    super(skipStream, maxSkipLevels, blockSize, 8);
+    this.blockSize = blockSize;
+    docPointer = new long[maxSkipLevels];
+    if (hasPos) {
+      posPointer = new long[maxSkipLevels];
+      posBufferUpto = new int[maxSkipLevels];
+      if (hasPayloads) {
+        payloadByteUpto = new int[maxSkipLevels];
+      } else {
+        payloadByteUpto = null;
+      }
+      if (hasOffsets) {
+        startOffset = new int[maxSkipLevels];
+      } else {
+        startOffset = null;
+      }
+      if (hasOffsets || hasPayloads) {
+        payPointer = new long[maxSkipLevels];
+      } else {
+        payPointer = null;
+      }
+    } else {
+      posPointer = null;
+    }
+  }
+
+  /**
+   * Trim original docFreq to tell skipReader read proper number of skip points.
+   *
+   * Since our definition in BlockSkip* is a little different from MultiLevelSkip*
+   * This trimmed docFreq will prevent skipReader from:
+   * 1. silly reading a non-existed skip point after the last block boundary
+   * 2. moving into the vInt block
+   *
+   */
+  protected int trim(int df) {
+    return df % blockSize == 0? df - 1: df;
+  }
+
+  public void init(long skipPointer, long docBasePointer, long posBasePointer, long payBasePointer, int df) {
+    super.init(skipPointer, trim(df));
+    lastDocPointer = docBasePointer;
+    lastPosPointer = posBasePointer;
+    lastPayPointer = payBasePointer;
+
+    Arrays.fill(docPointer, docBasePointer);
+    if (posPointer != null) {
+      Arrays.fill(posPointer, posBasePointer);
+      if (payPointer != null) {
+        Arrays.fill(payPointer, payBasePointer);
+      }
+    } else {
+      assert posBasePointer == 0;
+    }
+  }
+
+  /** Returns the doc pointer of the doc to which the last call of 
+   * {@link MultiLevelSkipListReader#skipTo(int)} has skipped.  */
+  public long getDocPointer() {
+    return lastDocPointer;
+  }
+
+  public long getPosPointer() {
+    return lastPosPointer;
+  }
+
+  public int getPosBufferUpto() {
+    return lastPosBufferUpto;
+  }
+
+  public long getPayPointer() {
+    return lastPayPointer;
+  }
+
+  public int getStartOffset() {
+    return lastStartOffset;
+  }
+
+  public int getPayloadByteUpto() {
+    return lastPayloadByteUpto;
+  }
+
+  public int getNextSkipDoc() {
+    return skipDoc[0];
+  }
+
+  @Override
+  protected void seekChild(int level) throws IOException {
+    super.seekChild(level);
+    // if (DEBUG) {
+    //   System.out.println("seekChild level=" + level);
+    // }
+    docPointer[level] = lastDocPointer;
+    if (posPointer != null) {
+      posPointer[level] = lastPosPointer;
+      posBufferUpto[level] = lastPosBufferUpto;
+      if (startOffset != null) {
+        startOffset[level] = lastStartOffset;
+      }
+      if (payloadByteUpto != null) {
+        payloadByteUpto[level] = lastPayloadByteUpto;
+      }
+      if (payPointer != null) {
+        payPointer[level] = lastPayPointer;
+      }
+    }
+  }
+  
+  @Override
+  protected void setLastSkipData(int level) {
+    super.setLastSkipData(level);
+    lastDocPointer = docPointer[level];
+    // if (DEBUG) {
+    //   System.out.println("setLastSkipData level=" + level);
+    //   System.out.println("  lastDocPointer=" + lastDocPointer);
+    // }
+    if (posPointer != null) {
+      lastPosPointer = posPointer[level];
+      lastPosBufferUpto = posBufferUpto[level];
+      // if (DEBUG) {
+      //   System.out.println("  lastPosPointer=" + lastPosPointer + " lastPosBUfferUpto=" + lastPosBufferUpto);
+      // }
+      if (payPointer != null) {
+        lastPayPointer = payPointer[level];
+      }
+      if (startOffset != null) {
+        lastStartOffset = startOffset[level];
+      }
+      if (payloadByteUpto != null) {
+        lastPayloadByteUpto = payloadByteUpto[level];
+      }
+    }
+  }
+
+  @Override
+  protected int readSkipData(int level, IndexInput skipStream) throws IOException {
+    // if (DEBUG) {
+    //   System.out.println("readSkipData level=" + level);
+    // }
+    int delta = skipStream.readVInt();
+    // if (DEBUG) {
+    //   System.out.println("  delta=" + delta);
+    // }
+    docPointer[level] += skipStream.readVInt();
+    // if (DEBUG) {
+    //   System.out.println("  docFP=" + docPointer[level]);
+    // }
+
+    if (posPointer != null) {
+      posPointer[level] += skipStream.readVInt();
+      // if (DEBUG) {
+      //   System.out.println("  posFP=" + posPointer[level]);
+      // }
+      posBufferUpto[level] = skipStream.readVInt();
+      // if (DEBUG) {
+      //   System.out.println("  posBufferUpto=" + posBufferUpto[level]);
+      // }
+
+      if (payloadByteUpto != null) {
+        payloadByteUpto[level] = skipStream.readVInt();
+      }
+
+      if (startOffset != null) {
+        startOffset[level] += skipStream.readVInt();
+      }
+
+      if (payPointer != null) {
+        payPointer[level] += skipStream.readVInt();
+      }
+    }
+    return delta;
+  }
+}

lucene/core/src/java/org/apache/lucene/codecs/block/BlockSkipWriter.java

@@ -0,0 +1,163 @@
+package org.apache.lucene.codecs.block;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.IOException;
+import java.util.Arrays;
+
+import org.apache.lucene.store.IndexOutput;
+import org.apache.lucene.codecs.MultiLevelSkipListWriter;
+
+/**
+ * Write skip lists with multiple levels, and support skip within block ints.
+ *
+ * Assume that docFreq = 28, skipInterval = blockSize = 12
+ *
+ *  |       block#0       | |      block#1        | |vInts|
+ *  d d d d d d d d d d d d d d d d d d d d d d d d d d d d (posting list)
+ *                          ^                       ^       (level 0 skip point)
+ *
+ * Note that skipWriter will ignore first document in block#0, since 
+ * it is useless as a skip point.  Also, we'll never skip into the vInts
+ * block, only record skip data at the start its start point(if it exist).
+ *
+ * For each skip point, we will record: 
+ * 1. docID in former position, i.e. for position 12, record docID[11], etc.
+ * 2. its related file points(position, payload), 
+ * 3. related numbers or uptos(position, payload).
+ * 4. start offset.
+ *
+ */
+final class BlockSkipWriter extends MultiLevelSkipListWriter {
+  // private boolean DEBUG = BlockPostingsReader.DEBUG;
+  
+  private int[] lastSkipDoc;
+  private long[] lastSkipDocPointer;
+  private long[] lastSkipPosPointer;
+  private long[] lastSkipPayPointer;
+  private int[] lastStartOffset;
+  private int[] lastPayloadByteUpto;
+
+  private final IndexOutput docOut;
+  private final IndexOutput posOut;
+  private final IndexOutput payOut;
+
+  private int curDoc;
+  private long curDocPointer;
+  private long curPosPointer;
+  private long curPayPointer;
+  private int curPosBufferUpto;
+  private int curStartOffset;
+  private int curPayloadByteUpto;
+  private boolean fieldHasPositions;
+  private boolean fieldHasOffsets;
+  private boolean fieldHasPayloads;
+
+  public BlockSkipWriter(int maxSkipLevels, int blockSize, int docCount, IndexOutput docOut, IndexOutput posOut, IndexOutput payOut) {
+    super(blockSize, 8, maxSkipLevels, docCount);
+    this.docOut = docOut;
+    this.posOut = posOut;
+    this.payOut = payOut;
+    
+    lastSkipDoc = new int[maxSkipLevels];
+    lastSkipDocPointer = new long[maxSkipLevels];
+    if (posOut != null) {
+      lastSkipPosPointer = new long[maxSkipLevels];
+      if (payOut != null) {
+        lastSkipPayPointer = new long[maxSkipLevels];
+      }
+      lastStartOffset = new int[maxSkipLevels];
+      lastPayloadByteUpto = new int[maxSkipLevels];
+    }
+  }
+
+  public void setField(boolean fieldHasPositions, boolean fieldHasOffsets, boolean fieldHasPayloads) {
+    this.fieldHasPositions = fieldHasPositions;
+    this.fieldHasOffsets = fieldHasOffsets;
+    this.fieldHasPayloads = fieldHasPayloads;
+  }
+
+  @Override
+  public void resetSkip() {
+    super.resetSkip();
+    Arrays.fill(lastSkipDoc, 0);
+    Arrays.fill(lastSkipDocPointer, docOut.getFilePointer());
+    if (fieldHasPositions) {
+      Arrays.fill(lastSkipPosPointer, posOut.getFilePointer());
+      if (fieldHasOffsets) {
+        Arrays.fill(lastStartOffset, 0);
+      }
+      if (fieldHasPayloads) {
+        Arrays.fill(lastPayloadByteUpto, 0);
+      }
+      if (fieldHasOffsets || fieldHasPayloads) {
+        Arrays.fill(lastSkipPayPointer, payOut.getFilePointer());
+      }
+    }
+  }
+
+  /**
+   * Sets the values for the current skip data. 
+   */
+  public void bufferSkip(int doc, int numDocs, long posFP, long payFP, int posBufferUpto, int startOffset, int payloadByteUpto) throws IOException {
+    this.curDoc = doc;
+    this.curDocPointer = docOut.getFilePointer();
+    this.curPosPointer = posFP;
+    this.curPayPointer = payFP;
+    this.curPosBufferUpto = posBufferUpto;
+    this.curPayloadByteUpto = payloadByteUpto;
+    this.curStartOffset = startOffset;
+    bufferSkip(numDocs);
+  }
+  
+  @Override
+  protected void writeSkipData(int level, IndexOutput skipBuffer) throws IOException {
+    int delta = curDoc - lastSkipDoc[level];
+    // if (DEBUG) {
+    //   System.out.println("writeSkipData level=" + level + " lastDoc=" + curDoc + " delta=" + delta + " curDocPointer=" + curDocPointer);
+    // }
+    skipBuffer.writeVInt(delta);
+    lastSkipDoc[level] = curDoc;
+
+    skipBuffer.writeVInt((int) (curDocPointer - lastSkipDocPointer[level]));
+    lastSkipDocPointer[level] = curDocPointer;
+
+    if (fieldHasPositions) {
+      // if (DEBUG) {
+      //   System.out.println("  curPosPointer=" + curPosPointer + " curPosBufferUpto=" + curPosBufferUpto);
+      // }
+      skipBuffer.writeVInt((int) (curPosPointer - lastSkipPosPointer[level]));
+      lastSkipPosPointer[level] = curPosPointer;
+      skipBuffer.writeVInt(curPosBufferUpto);
+
+      if (fieldHasPayloads) {
+        skipBuffer.writeVInt(curPayloadByteUpto);
+      }
+
+      if (fieldHasOffsets) {
+        skipBuffer.writeVInt(curStartOffset - lastStartOffset[level]);
+        lastStartOffset[level] = curStartOffset;
+      }
+
+      if (fieldHasOffsets || fieldHasPayloads) {
+        skipBuffer.writeVInt((int) (curPayPointer - lastSkipPayPointer[level]));
+        lastSkipPayPointer[level] = curPayPointer;
+      }
+    }
+  }
+}

lucene/core/src/java/org/apache/lucene/codecs/block/ForUtil.java

@@ -0,0 +1,247 @@
+package org.apache.lucene.codecs.block;
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import java.io.IOException;
+import java.util.Arrays;
+
+import org.apache.lucene.index.CorruptIndexException;
+import org.apache.lucene.store.DataInput;
+import org.apache.lucene.store.DataOutput;
+import org.apache.lucene.store.IndexInput;
+import org.apache.lucene.store.IndexOutput;
+import org.apache.lucene.util.packed.PackedInts.Decoder;
+import org.apache.lucene.util.packed.PackedInts.FormatAndBits;
+import org.apache.lucene.util.packed.PackedInts;
+
+import static org.apache.lucene.codecs.block.BlockPostingsFormat.BLOCK_SIZE;
+
+/**
+ * Encode all values in normal area with fixed bit width, 
+ * which is determined by the max value in this block.
+ */
+final class ForUtil {
+
+  /**
+   * Special number of bits per value used whenever all values to encode are equal.
+   */
+  private static final int ALL_VALUES_EQUAL = 0;
+
+  /**
+   * Upper limit of the number of bytes that might be required to stored
+   * <code>BLOCK_SIZE</code> encoded values.
+   */
+  static final int MAX_ENCODED_SIZE = BLOCK_SIZE * 4;
+
+  /**
+   * Upper limit of the number of values that might be decoded in a single call to
+   * {@link #readBlock(IndexInput, byte[], int[])}. Although values after
+   * <code>BLOCK_SIZE</code> are garbage, it is necessary to allocate value buffers
+   * whose size is >= MAX_DATA_SIZE to avoid {@link ArrayIndexOutOfBoundsException}s.
+   */
+  static final int MAX_DATA_SIZE;
+  static {
+    int maxDataSize = 0;
+    for(int version=PackedInts.VERSION_START;version<=PackedInts.VERSION_CURRENT;version++) {
+      for (PackedInts.Format format : PackedInts.Format.values()) {
+        for (int bpv = 1; bpv <= 32; ++bpv) {
+          if (!format.isSupported(bpv)) {
+            continue;
+          }
+          final PackedInts.Decoder decoder = PackedInts.getDecoder(format, version, bpv);
+          final int iterations = computeIterations(decoder);
+          maxDataSize = Math.max(maxDataSize, iterations * decoder.valueCount());
+        }
+      }
+    }
+    MAX_DATA_SIZE = maxDataSize;
+  }
+
+  /**
+   * Compute the number of iterations required to decode <code>BLOCK_SIZE</code>
+   * values with the provided {@link Decoder}.
+   */
+  private static int computeIterations(PackedInts.Decoder decoder) {
+    return (int) Math.ceil((float) BLOCK_SIZE / decoder.valueCount());
+  }
+
+  /**
+   * Compute the number of bytes required to encode a block of values that require
+   * <code>bitsPerValue</code> bits per value with format <code>format</code>.
+   */
+  private static int encodedSize(PackedInts.Format format, int bitsPerValue) {
+    return format.nblocks(bitsPerValue, BLOCK_SIZE) << 3;
+  }
+
+  private final int[] encodedSizes;
+  private final PackedInts.Encoder[] encoders;
+  private final PackedInts.Decoder[] decoders;
+  private final int[] iterations;
+
+  /**
+   * Create a new {@link ForUtil} instance and save state into <code>out</code>.
+   */
+  ForUtil(float acceptableOverheadRatio, DataOutput out) throws IOException {
+    out.writeVInt(PackedInts.VERSION_CURRENT);
+    encodedSizes = new int[33];
+    encoders = new PackedInts.Encoder[33];
+    decoders = new PackedInts.Decoder[33];
+    iterations = new int[33];
+
+    for (int bpv = 1; bpv <= 32; ++bpv) {
+      final FormatAndBits formatAndBits = PackedInts.fastestFormatAndBits(
+          BLOCK_SIZE, bpv, acceptableOverheadRatio);
+      assert formatAndBits.format.isSupported(formatAndBits.bitsPerValue);
+      assert formatAndBits.bitsPerValue <= 32;
+      encodedSizes[bpv] = encodedSize(formatAndBits.format, formatAndBits.bitsPerValue);
+      encoders[bpv] = PackedInts.getEncoder(
+          formatAndBits.format, PackedInts.VERSION_CURRENT, formatAndBits.bitsPerValue);
+      decoders[bpv] = PackedInts.getDecoder(
+          formatAndBits.format, PackedInts.VERSION_CURRENT, formatAndBits.bitsPerValue);
+      iterations[bpv] = computeIterations(decoders[bpv]);
+
+      out.writeVInt(formatAndBits.format.getId() << 5 | (formatAndBits.bitsPerValue - 1));
+    }
+  }
+
+  /**
+   * Restore a {@link ForUtil} from a {@link DataInput}.
+   */
+  ForUtil(DataInput in) throws IOException {
+    int packedIntsVersion = in.readVInt();
+    if (packedIntsVersion != PackedInts.VERSION_START) {
+      throw new CorruptIndexException("expected version=" + PackedInts.VERSION_START + " but got version=" + packedIntsVersion);
+    }
+    encodedSizes = new int[33];
+    encoders = new PackedInts.Encoder[33];
+    decoders = new PackedInts.Decoder[33];
+    iterations = new int[33];
+
+    for (int bpv = 1; bpv <= 32; ++bpv) {
+      final int code = in.readVInt();
+      final int formatId = code >>> 5;
+      final int bitsPerValue = (code & 31) + 1;
+
+      final PackedInts.Format format = PackedInts.Format.byId(formatId);
+      assert format.isSupported(bitsPerValue);
+      encodedSizes[bpv] = encodedSize(format, bitsPerValue);
+      encoders[bpv] = PackedInts.getEncoder(
+          format, packedIntsVersion, bitsPerValue);
+      decoders[bpv] = PackedInts.getDecoder(
+          format, packedIntsVersion, bitsPerValue);
+      iterations[bpv] = computeIterations(decoders[bpv]);
+    }
+  }
+
+  /**
+   * Write a block of data (<code>For</code> format).
+   *
+   * @param data     the data to write
+   * @param encoded  a buffer to use to encode data
+   * @param out      the destination output
+   * @throws IOException
+   */
+  void writeBlock(int[] data, byte[] encoded, IndexOutput out) throws IOException {
+    if (isAllEqual(data)) {
+      out.writeVInt(ALL_VALUES_EQUAL);
+      out.writeVInt(data[0]);
+      return;
+    }
+
+    final int numBits = bitsRequired(data);
+    assert numBits > 0 && numBits <= 32 : numBits;
+    final PackedInts.Encoder encoder = encoders[numBits];
+    final int iters = iterations[numBits];
+    assert iters * encoder.valueCount() >= BLOCK_SIZE;
+    final int encodedSize = encodedSizes[numBits];
+    assert (iters * encoder.blockCount()) << 3 >= encodedSize;
+
+    out.writeVInt(numBits);
+
+    encoder.encode(data, 0, encoded, 0, iters);
+    out.writeBytes(encoded, encodedSize);
+  }
+
+  /**
+   * Read the next block of data (<code>For</code> format).
+   *
+   * @param in        the input to use to read data
+   * @param encoded   a buffer that can be used to store encoded data
+   * @param decoded   where to write decoded data
+   * @throws IOException
+   */
+  void readBlock(IndexInput in, byte[] encoded, int[] decoded) throws IOException {
+    final int numBits = in.readVInt();
+    assert numBits <= 32 : numBits;
+
+    if (numBits == ALL_VALUES_EQUAL) {
+      final int value = in.readVInt();
+      Arrays.fill(decoded, 0, BLOCK_SIZE, value);
+      return;
+    }
+
+    final int encodedSize = encodedSizes[numBits];
+    in.readBytes(encoded, 0, encodedSize);
+
+    final PackedInts.Decoder decoder = decoders[numBits];
+    final int iters = iterations[numBits];
+    assert iters * decoder.valueCount() >= BLOCK_SIZE;
+
+    decoder.decode(encoded, 0, decoded, 0, iters);
+  }
+
+  /**
+   * Skip the next block of data.
+   *
+   * @param in      the input where to read data
+   * @throws IOException
+   */
+  void skipBlock(IndexInput in) throws IOException {
+    final int numBits = in.readVInt();
+    if (numBits == ALL_VALUES_EQUAL) {
+      in.readVInt();
+      return;
+    }
+    assert numBits > 0 && numBits <= 32 : numBits;
+    final int encodedSize = encodedSizes[numBits];
+    in.seek(in.getFilePointer() + encodedSize);
+  }
+
+  private static boolean isAllEqual(final int[] data) {
+    final long v = data[0];
+    for (int i = 1; i < BLOCK_SIZE; ++i) {
+      if (data[i] != v) {
+        return false;
+      }
+    }
+    return true;
+  }
+
+  /**
+   * Compute the number of bits required to serialize any of the longs in
+   * <code>data</code>.
+   */
+  private static int bitsRequired(final int[] data) {
+    long or = 0;
+    for (int i = 0; i < BLOCK_SIZE; ++i) {
+      assert data[i] >= 0;
+      or |= data[i];
+    }
+    return PackedInts.bitsRequired(or);
+  }
+
+}

lucene/core/src/java/org/apache/lucene/codecs/block/package.html

@@ -0,0 +1,25 @@
+<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
+<!--
+ Licensed to the Apache Software Foundation (ASF) under one or more
+ contributor license agreements.  See the NOTICE file distributed with
+ this work for additional information regarding copyright ownership.
+ The ASF licenses this file to You under the Apache License, Version 2.0
+ (the "License"); you may not use this file except in compliance with
+ the License.  You may obtain a copy of the License at
+
+     http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
+-->
+<html>
+<head>
+   <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
+</head>
+<body>
+BlockPostingsFormat file format.
+</body>
+</html>

lucene/core/src/java/org/apache/lucene/codecs/bloom/BloomFilteringPostingsFormat.java

@@ -207,14 +207,10 @@ public class BloomFilteringPostingsFormat extends PostingsFormat {
       }
     }
     
-    public int size() throws IOException {
+    public int size() {
       return delegateFieldsProducer.size();
     }
     
-    public long getUniqueTermCount() throws IOException {
-      return delegateFieldsProducer.getUniqueTermCount();
-    }
-    
     class BloomFilteredTerms extends Terms {
       private Terms delegateTerms;
       private FuzzySet filter;

lucene/core/src/java/org/apache/lucene/codecs/intblock/FixedIntBlockIndexInput.java

@@ -26,7 +26,6 @@ import java.io.IOException;
 import org.apache.lucene.codecs.sep.IntIndexInput;
 import org.apache.lucene.store.DataInput;
 import org.apache.lucene.store.IndexInput;
-import org.apache.lucene.util.IntsRef;
 
 /** Abstract base class that reads fixed-size blocks of ints
  *  from an IndexInput.  While this is a simple approach, a
@@ -49,7 +48,7 @@ public abstract class FixedIntBlockIndexInput extends IntIndexInput {
   @Override
   public Reader reader() throws IOException {
     final int[] buffer = new int[blockSize];
-    final IndexInput clone = (IndexInput) in.clone();
+    final IndexInput clone = in.clone();
     // TODO: can this be simplified?
     return new Reader(clone, buffer, this.getBlockReader(clone, buffer));
   }
@@ -77,76 +76,48 @@ public abstract class FixedIntBlockIndexInput extends IntIndexInput {
 
   private static class Reader extends IntIndexInput.Reader {
     private final IndexInput in;
-
-    protected final int[] pending;
-    int upto;
-
-    private boolean seekPending;
-    private long pendingFP;
-    private int pendingUpto;
-    private long lastBlockFP;
     private final BlockReader blockReader;
     private final int blockSize;
-    private final IntsRef bulkResult = new IntsRef();
+    private final int[] pending;
+
+    private int upto;
+    private boolean seekPending;
+    private long pendingFP;
+    private long lastBlockFP = -1;
 
     public Reader(final IndexInput in, final int[] pending, final BlockReader blockReader) {
       this.in = in;
       this.pending = pending;
       this.blockSize = pending.length;
-      bulkResult.ints = pending;
       this.blockReader = blockReader;
       upto = blockSize;
     }
 
     void seek(final long fp, final int upto) {
-      pendingFP = fp;
-      pendingUpto = upto;
-      seekPending = true;
-    }
-
-    private void maybeSeek() throws IOException {
-      if (seekPending) {
-        if (pendingFP != lastBlockFP) {
-          // need new block
-          in.seek(pendingFP);
-          lastBlockFP = pendingFP;
-          blockReader.readBlock();
-        }
-        upto = pendingUpto;
-        seekPending = false;
+      assert upto < blockSize;
+      if (seekPending || fp != lastBlockFP) {
+        pendingFP = fp;
+        seekPending = true;
       }
+      this.upto = upto;
     }
 
     @Override
     public int next() throws IOException {
-      this.maybeSeek();
-      if (upto == blockSize) {
+      if (seekPending) {
+        // Seek & load new block
+        in.seek(pendingFP);
+        lastBlockFP = pendingFP;
+        blockReader.readBlock();
+        seekPending = false;
+      } else if (upto == blockSize) {
+        // Load new block
         lastBlockFP = in.getFilePointer();
         blockReader.readBlock();
         upto = 0;
       }
-
       return pending[upto++];
     }
-
-    @Override
-    public IntsRef read(final int count) throws IOException {
-      this.maybeSeek();
-      if (upto == blockSize) {
-        blockReader.readBlock();
-        upto = 0;
-      }
-      bulkResult.offset = upto;
-      if (upto + count < blockSize) {
-        bulkResult.length = count;
-        upto += count;
-      } else {
-        bulkResult.length = blockSize - upto;
-        upto = blockSize;
-      }
-
-      return bulkResult;
-    }
   }
 
   private class Index extends IntIndexInput.Index {
@@ -178,7 +149,7 @@ public abstract class FixedIntBlockIndexInput extends IntIndexInput {
     }
 
     @Override
-    public void set(final IntIndexInput.Index other) {
+    public void copyFrom(final IntIndexInput.Index other) {
       final Index idx = (Index) other;
       fp = idx.fp;
       upto = idx.upto;

lucene/core/src/java/org/apache/lucene/codecs/intblock/VariableIntBlockIndexInput.java

@@ -26,7 +26,6 @@ import java.io.IOException;
 import org.apache.lucene.codecs.sep.IntIndexInput;
 import org.apache.lucene.store.DataInput;
 import org.apache.lucene.store.IndexInput;
-import org.apache.lucene.util.IntsRef;
 
 // TODO: much of this can be shared code w/ the fixed case
 
@@ -51,7 +50,7 @@ public abstract class VariableIntBlockIndexInput extends IntIndexInput {
   @Override
   public Reader reader() throws IOException {
     final int[] buffer = new int[maxBlockSize];
-    final IndexInput clone = (IndexInput) in.clone();
+    final IndexInput clone = in.clone();
     // TODO: can this be simplified?
     return new Reader(clone, buffer, this.getBlockReader(clone, buffer));
   }
@@ -90,12 +89,10 @@ public abstract class VariableIntBlockIndexInput extends IntIndexInput {
     private long lastBlockFP;
     private int blockSize;
     private final BlockReader blockReader;
-    private final IntsRef bulkResult = new IntsRef();
 
     public Reader(final IndexInput in, final int[] pending, final BlockReader blockReader) {
       this.in = in;
       this.pending = pending;
-      bulkResult.ints = pending;
       this.blockReader = blockReader;
     }
 
@@ -146,26 +143,6 @@ public abstract class VariableIntBlockIndexInput extends IntIndexInput {
 
       return pending[upto++];
     }
-
-    @Override
-    public IntsRef read(final int count) throws IOException {
-      this.maybeSeek();
-      if (upto == blockSize) {
-        lastBlockFP = in.getFilePointer();
-        blockSize = blockReader.readBlock();
-        upto = 0;
-      }
-      bulkResult.offset = upto;
-      if (upto + count < blockSize) {
-        bulkResult.length = count;
-        upto += count;
-      } else {
-        bulkResult.length = blockSize - upto;
-        upto = blockSize;
-      }
-
-      return bulkResult;
-    }
   }
 
   private class Index extends IntIndexInput.Index {
@@ -204,7 +181,7 @@ public abstract class VariableIntBlockIndexInput extends IntIndexInput {
     }
 
     @Override
-    public void set(final IntIndexInput.Index other) {
+    public void copyFrom(final IntIndexInput.Index other) {
       final Index idx = (Index) other;
       fp = idx.fp;
       upto = idx.upto;

lucene/core/src/java/org/apache/lucene/codecs/lucene40/Lucene40PostingsFormat.java

@@ -159,7 +159,7 @@ import org.apache.lucene.util.fst.FST; // javadocs
  * with the frequency of the term in that document (except when frequencies are
  * omitted: {@link IndexOptions#DOCS_ONLY}).</p>
  * <ul>
- *   <li>FreqFile (.frq) --&gt; Header, &lt;TermFreqs, SkipData&gt; <sup>TermCount</sup></li>
+ *   <li>FreqFile (.frq) --&gt; Header, &lt;TermFreqs, SkipData?&gt; <sup>TermCount</sup></li>
  *   <li>Header --&gt; {@link CodecUtil#writeHeader CodecHeader}</li>
  *   <li>TermFreqs --&gt; &lt;TermFreq&gt; <sup>DocFreq</sup></li>
  *   <li>TermFreq --&gt; DocDelta[, Freq?]</li>

lucene/core/src/java/org/apache/lucene/codecs/lucene40/Lucene40PostingsReader.java

@@ -325,7 +325,7 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
     
     SegmentDocsEnumBase(IndexInput startFreqIn, Bits liveDocs) {
       this.startFreqIn = startFreqIn;
-      this.freqIn = (IndexInput)startFreqIn.clone();
+      this.freqIn = startFreqIn.clone();
       this.liveDocs = liveDocs;
       
     }
@@ -474,7 +474,7 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
 
         if (skipper == null) {
           // This is the first time this enum has ever been used for skipping -- do lazy init
-          skipper = new Lucene40SkipListReader((IndexInput) freqIn.clone(), maxSkipLevels, skipInterval);
+          skipper = new Lucene40SkipListReader(freqIn.clone(), maxSkipLevels, skipInterval);
         }
 
         if (!skipped) {
@@ -705,8 +705,8 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
 
     public SegmentDocsAndPositionsEnum(IndexInput freqIn, IndexInput proxIn) {
       startFreqIn = freqIn;
-      this.freqIn = (IndexInput) freqIn.clone();
-      this.proxIn = (IndexInput) proxIn.clone();
+      this.freqIn = freqIn.clone();
+      this.proxIn = proxIn.clone();
     }
 
     public SegmentDocsAndPositionsEnum reset(FieldInfo fieldInfo, StandardTermState termState, Bits liveDocs) throws IOException {
@@ -795,7 +795,7 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
 
         if (skipper == null) {
           // This is the first time this enum has ever been used for skipping -- do lazy init
-          skipper = new Lucene40SkipListReader((IndexInput) freqIn.clone(), maxSkipLevels, skipInterval);
+          skipper = new Lucene40SkipListReader(freqIn.clone(), maxSkipLevels, skipInterval);
         }
 
         if (!skipped) {
@@ -913,8 +913,8 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
 
     public SegmentFullPositionsEnum(IndexInput freqIn, IndexInput proxIn) {
       startFreqIn = freqIn;
-      this.freqIn = (IndexInput) freqIn.clone();
-      this.proxIn = (IndexInput) proxIn.clone();
+      this.freqIn = freqIn.clone();
+      this.proxIn = proxIn.clone();
     }
 
     public SegmentFullPositionsEnum reset(FieldInfo fieldInfo, StandardTermState termState, Bits liveDocs) throws IOException {
@@ -1009,7 +1009,7 @@ public class Lucene40PostingsReader extends PostingsReaderBase {
 
         if (skipper == null) {
           // This is the first time this enum has ever been used for skipping -- do lazy init
-          skipper = new Lucene40SkipListReader((IndexInput) freqIn.clone(), maxSkipLevels, skipInterval);
+          skipper = new Lucene40SkipListReader(freqIn.clone(), maxSkipLevels, skipInterval);
         }
 
         if (!skipped) {

lucene/core/src/java/org/apache/lucene/codecs/lucene40/Lucene40PostingsWriter.java

@@ -185,8 +185,6 @@ public final class Lucene40PostingsWriter extends PostingsWriterBase {
   int lastDocID;
   int df;
   
-  /** Adds a new doc in this term.  If this returns null
-   *  then we just skip consuming positions/payloads. */
   @Override
   public void startDoc(int docID, int termDocFreq) throws IOException {
     // if (DEBUG) System.out.println("SPW:   startDoc seg=" + segment + " docID=" + docID + " tf=" + termDocFreq + " freqOut.fp=" + freqOut.getFilePointer());

lucene/core/src/java/org/apache/lucene/codecs/lucene40/Lucene40StoredFieldsReader.java

@@ -61,7 +61,7 @@ public final class Lucene40StoredFieldsReader extends StoredFieldsReader impleme
   @Override
   public Lucene40StoredFieldsReader clone() {
     ensureOpen();
-    return new Lucene40StoredFieldsReader(fieldInfos, numTotalDocs, size, (IndexInput)fieldsStream.clone(), (IndexInput)indexStream.clone());
+    return new Lucene40StoredFieldsReader(fieldInfos, numTotalDocs, size, fieldsStream.clone(), indexStream.clone());
   }
   
   // Used only by clone

lucene/core/src/java/org/apache/lucene/codecs/lucene40/Lucene40TermVectorsReader.java

@@ -403,7 +403,7 @@ public class Lucene40TermVectorsReader extends TermVectorsReader {
     // NOTE: tvf is pre-positioned by caller
     public TVTermsEnum() {
       this.origTVF = Lucene40TermVectorsReader.this.tvf;
-      tvf = (IndexInput) origTVF.clone();
+      tvf = origTVF.clone();
     }
 
     public boolean canReuse(IndexInput tvf) {
@@ -752,9 +752,9 @@ public class Lucene40TermVectorsReader extends TermVectorsReader {
     // These are null when a TermVectorsReader was created
     // on a segment that did not have term vectors saved
     if (tvx != null && tvd != null && tvf != null) {
-      cloneTvx = (IndexInput) tvx.clone();
-      cloneTvd = (IndexInput) tvd.clone();
-      cloneTvf = (IndexInput) tvf.clone();
+      cloneTvx = tvx.clone();
+      cloneTvd = tvd.clone();
+      cloneTvf = tvf.clone();
     }
     
     return new Lucene40TermVectorsReader(fieldInfos, cloneTvx, cloneTvd, cloneTvf, size, numTotalDocs);

lucene/core/src/java/org/apache/lucene/codecs/lucene40/values/Bytes.java

@@ -350,7 +350,7 @@ public final class Bytes {
      */
     protected final IndexInput cloneData() {
       assert datIn != null;
-      return (IndexInput) datIn.clone();
+      return datIn.clone();
     }
 
     /**
@@ -358,7 +358,7 @@ public final class Bytes {
      */
     protected final IndexInput cloneIndex() {
       assert idxIn != null;
-      return (IndexInput) idxIn.clone();
+      return idxIn.clone();
     }
 
     @Override

lucene/core/src/java/org/apache/lucene/codecs/lucene40/values/PackedIntValues.java

@@ -187,7 +187,7 @@ class PackedIntValues {
       final Source source;
       IndexInput input = null;
       try {
-        input = (IndexInput) datIn.clone();
+        input = datIn.clone();
         
         if (values == null) {
           source = new PackedIntsSource(input, false);
@@ -218,7 +218,7 @@ class PackedIntValues {
 
     @Override
     public Source getDirectSource() throws IOException {
-      return values != null ? new FixedStraightBytesImpl.DirectFixedStraightSource((IndexInput) datIn.clone(), 8, Type.FIXED_INTS_64) : new PackedIntsSource((IndexInput) datIn.clone(), true);
+      return values != null ? new FixedStraightBytesImpl.DirectFixedStraightSource(datIn.clone(), 8, Type.FIXED_INTS_64) : new PackedIntsSource(datIn.clone(), true);
     }
   }
 

lucene/core/src/java/org/apache/lucene/codecs/lucene40/values/VarSortedBytesImpl.java

@@ -211,7 +211,7 @@ final class VarSortedBytesImpl {
       valueCount = ordToOffsetIndex.size()-1; // the last value here is just a dummy value to get the length of the last value
       // advance this iterator to the end and clone the stream once it points to the docToOrdIndex header
       ordToOffsetIndex.get(valueCount);
-      docToOrdIndex = PackedInts.getDirectReader((IndexInput) idxIn.clone()); // read the ords in to prevent too many random disk seeks
+      docToOrdIndex = PackedInts.getDirectReader(idxIn.clone()); // read the ords in to prevent too many random disk seeks
       basePointer = datIn.getFilePointer();
       this.datIn = datIn;
     }

lucene/core/src/java/org/apache/lucene/codecs/memory/DirectPostingsFormat.java

@@ -18,6 +18,7 @@ package org.apache.lucene.codecs.memory;
  */
 
 import java.io.IOException;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.Iterator;
 import java.util.Map;
@@ -43,7 +44,6 @@ import org.apache.lucene.store.RAMOutputStream;
 import org.apache.lucene.util.ArrayUtil;
 import org.apache.lucene.util.Bits;
 import org.apache.lucene.util.BytesRef;
-import org.apache.lucene.util.UnmodifiableIterator;
 import org.apache.lucene.util.automaton.CompiledAutomaton;
 import org.apache.lucene.util.automaton.RunAutomaton;
 import org.apache.lucene.util.automaton.Transition;
@@ -131,7 +131,7 @@ public class DirectPostingsFormat extends PostingsFormat {
 
     @Override
     public Iterator<String> iterator() {
-      return new UnmodifiableIterator<String>(fields.keySet().iterator());
+      return Collections.unmodifiableSet(fields.keySet()).iterator();
     }
 
     @Override
@@ -144,15 +144,6 @@ public class DirectPostingsFormat extends PostingsFormat {
       return fields.size();
     }
 
-    @Override
-    public long getUniqueTermCount() {
-      long numTerms = 0;      
-      for(DirectField field : fields.values()) {
-        numTerms += field.terms.length;
-      }
-      return numTerms;
-    }
-
     @Override
     public void close() {
     }

lucene/core/src/java/org/apache/lucene/codecs/memory/MemoryPostingsFormat.java

@@ -18,6 +18,7 @@ package org.apache.lucene.codecs.memory;
  */
 
 import java.io.IOException;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.Iterator;
 import java.util.SortedMap;
@@ -48,7 +49,6 @@ import org.apache.lucene.util.ArrayUtil;
 import org.apache.lucene.util.Bits;
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.IntsRef;
-import org.apache.lucene.util.UnmodifiableIterator;
 import org.apache.lucene.util.fst.Builder;
 import org.apache.lucene.util.fst.ByteSequenceOutputs;
 import org.apache.lucene.util.fst.BytesRefFSTEnum;
@@ -863,7 +863,7 @@ public class MemoryPostingsFormat extends PostingsFormat {
     return new FieldsProducer() {
       @Override
       public Iterator<String> iterator() {
-        return new UnmodifiableIterator<String>(fields.keySet().iterator());
+        return Collections.unmodifiableSet(fields.keySet()).iterator();
       }
 
       @Override

lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldPostingsFormat.java

@@ -19,6 +19,7 @@ package org.apache.lucene.codecs.perfield;
 
 import java.io.Closeable;
 import java.io.IOException;
+import java.util.Collections;
 import java.util.HashMap;
 import java.util.Iterator;
 import java.util.Map;
@@ -34,7 +35,6 @@ import org.apache.lucene.index.SegmentReadState;
 import org.apache.lucene.index.SegmentWriteState;
 import org.apache.lucene.index.Terms;
 import org.apache.lucene.util.IOUtils;
-import org.apache.lucene.util.UnmodifiableIterator;
 
 /**
  * Enables per field format support.
@@ -199,7 +199,7 @@ public abstract class PerFieldPostingsFormat extends PostingsFormat {
 
     @Override
     public Iterator<String> iterator() {
-      return new UnmodifiableIterator<String>(fields.keySet().iterator());
+      return Collections.unmodifiableSet(fields.keySet()).iterator();
     }
 
     @Override

lucene/core/src/java/org/apache/lucene/codecs/sep/IntIndexInput.java

@@ -21,7 +21,6 @@ import java.io.Closeable;
 import java.io.IOException;
 
 import org.apache.lucene.store.DataInput;
-import org.apache.lucene.util.IntsRef;
 
 /** Defines basic API for writing ints to an IndexOutput.
  *  IntBlockCodec interacts with this API. @see
@@ -44,7 +43,7 @@ public abstract class IntIndexInput implements Closeable {
     /** Seeks primary stream to the last read offset */
     public abstract void seek(IntIndexInput.Reader stream) throws IOException;
 
-    public abstract void set(Index other);
+    public abstract void copyFrom(Index other);
     
     @Override
     public abstract Index clone();
@@ -55,23 +54,5 @@ public abstract class IntIndexInput implements Closeable {
 
     /** Reads next single int */
     public abstract int next() throws IOException;
-
-    /** Reads next chunk of ints */
-    private IntsRef bulkResult;
-
-    /** Read up to count ints. */
-    public IntsRef read(int count) throws IOException {
-      if (bulkResult == null) {
-        bulkResult = new IntsRef();
-        bulkResult.ints = new int[count];
-      } else {
-        bulkResult.grow(count);
-      }
-      for(int i=0;i<count;i++) {
-        bulkResult.ints[i] = next();
-      }
-      bulkResult.length = count;
-      return bulkResult;
-    }
   }
 }

lucene/core/src/java/org/apache/lucene/codecs/sep/SepPostingsReader.java

@@ -160,13 +160,13 @@ public class SepPostingsReader extends PostingsReaderBase {
       if (docIndex == null) {
         docIndex = other.docIndex.clone();
       } else {
-        docIndex.set(other.docIndex);
+        docIndex.copyFrom(other.docIndex);
       }
       if (other.freqIndex != null) {
         if (freqIndex == null) {
           freqIndex = other.freqIndex.clone();
         } else {
-          freqIndex.set(other.freqIndex);
+          freqIndex.copyFrom(other.freqIndex);
         }
       } else {
         freqIndex = null;
@@ -175,7 +175,7 @@ public class SepPostingsReader extends PostingsReaderBase {
         if (posIndex == null) {
           posIndex = other.posIndex.clone();
         } else {
-          posIndex.set(other.posIndex);
+          posIndex.copyFrom(other.posIndex);
         }
       } else {
         posIndex = null;
@@ -352,11 +352,11 @@ public class SepPostingsReader extends PostingsReaderBase {
 
       // TODO: can't we only do this if consumer
       // skipped consuming the previous docs?
-      docIndex.set(termState.docIndex);
+      docIndex.copyFrom(termState.docIndex);
       docIndex.seek(docReader);
 
       if (!omitTF) {
-        freqIndex.set(termState.freqIndex);
+        freqIndex.copyFrom(termState.freqIndex);
         freqIndex.seek(freqReader);
       }
 
@@ -418,7 +418,7 @@ public class SepPostingsReader extends PostingsReaderBase {
 
         if (skipper == null) {
           // This DocsEnum has never done any skipping
-          skipper = new SepSkipListReader((IndexInput) skipIn.clone(),
+          skipper = new SepSkipListReader(skipIn.clone(),
                                           freqIn,
                                           docIn,
                                           posIn,
@@ -506,7 +506,7 @@ public class SepPostingsReader extends PostingsReaderBase {
       freqIndex = freqIn.index();
       posReader = posIn.reader();
       posIndex = posIn.index();
-      payloadIn = (IndexInput) SepPostingsReader.this.payloadIn.clone();
+      payloadIn = SepPostingsReader.this.payloadIn.clone();
     }
 
     SepDocsAndPositionsEnum init(FieldInfo fieldInfo, SepTermState termState, Bits liveDocs) throws IOException {
@@ -516,15 +516,15 @@ public class SepPostingsReader extends PostingsReaderBase {
 
       // TODO: can't we only do this if consumer
       // skipped consuming the previous docs?
-      docIndex.set(termState.docIndex);
+      docIndex.copyFrom(termState.docIndex);
       docIndex.seek(docReader);
       //System.out.println("  docIndex=" + docIndex);
 
-      freqIndex.set(termState.freqIndex);
+      freqIndex.copyFrom(termState.freqIndex);
       freqIndex.seek(freqReader);
       //System.out.println("  freqIndex=" + freqIndex);
 
-      posIndex.set(termState.posIndex);
+      posIndex.copyFrom(termState.posIndex);
       //System.out.println("  posIndex=" + posIndex);
       posSeekPending = true;
       payloadPending = false;
@@ -597,7 +597,7 @@ public class SepPostingsReader extends PostingsReaderBase {
         if (skipper == null) {
           //System.out.println("  create skipper");
           // This DocsEnum has never done any skipping
-          skipper = new SepSkipListReader((IndexInput) skipIn.clone(),
+          skipper = new SepSkipListReader(skipIn.clone(),
                                           freqIn,
                                           docIn,
                                           posIn,
@@ -629,7 +629,7 @@ public class SepPostingsReader extends PostingsReaderBase {
           // NOTE: don't seek pos here; do it lazily
           // instead.  Eg a PhraseQuery may skip to many
           // docs before finally asking for positions...
-          posIndex.set(skipper.getPosIndex());
+          posIndex.copyFrom(skipper.getPosIndex());
           posSeekPending = true;
           count = newCount;
           doc = accum = skipper.getDoc();

lucene/core/src/java/org/apache/lucene/codecs/sep/SepSkipListReader.java

@@ -108,12 +108,12 @@ class SepSkipListReader extends MultiLevelSkipListReader {
     lastPayloadPointer = payloadBasePointer;
 
     for(int i=0;i<maxNumberOfSkipLevels;i++) {
-      docIndex[i].set(docBaseIndex);
+      docIndex[i].copyFrom(docBaseIndex);
       if (freqIndex != null) {
-        freqIndex[i].set(freqBaseIndex);
+        freqIndex[i].copyFrom(freqBaseIndex);
       }
       if (posBaseIndex != null) {
-        posIndex[i].set(posBaseIndex);
+        posIndex[i].copyFrom(posBaseIndex);
       }
     }
     Arrays.fill(payloadPointer, payloadBasePointer);
@@ -145,20 +145,20 @@ class SepSkipListReader extends MultiLevelSkipListReader {
     lastPayloadPointer = payloadPointer[level];
     lastPayloadLength = payloadLength[level];
     if (freqIndex != null) {
-      lastFreqIndex.set(freqIndex[level]);
+      lastFreqIndex.copyFrom(freqIndex[level]);
     }
-    lastDocIndex.set(docIndex[level]);
+    lastDocIndex.copyFrom(docIndex[level]);
     if (lastPosIndex != null) {
-      lastPosIndex.set(posIndex[level]);
+      lastPosIndex.copyFrom(posIndex[level]);
     }
 
     if (level > 0) {
       if (freqIndex != null) {
-        freqIndex[level-1].set(freqIndex[level]);
+        freqIndex[level-1].copyFrom(freqIndex[level]);
       }
-      docIndex[level-1].set(docIndex[level]);
+      docIndex[level-1].copyFrom(docIndex[level]);
       if (posIndex != null) {
-        posIndex[level-1].set(posIndex[level]);
+        posIndex[level-1].copyFrom(posIndex[level]);
       }
     }
   }

lucene/core/src/java/org/apache/lucene/codecs/simpletext/SimpleTextFieldsReader.java

@@ -18,6 +18,7 @@ package org.apache.lucene.codecs.simpletext;
  */
 
 import java.io.IOException;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.HashMap;
 import java.util.Iterator;
@@ -43,7 +44,6 @@ import org.apache.lucene.util.IntsRef;
 import org.apache.lucene.util.OpenBitSet;
 import org.apache.lucene.util.StringHelper;
 import org.apache.lucene.util.UnicodeUtil;
-import org.apache.lucene.util.UnmodifiableIterator;
 import org.apache.lucene.util.fst.Builder;
 import org.apache.lucene.util.fst.BytesRefFSTEnum;
 import org.apache.lucene.util.fst.FST;
@@ -70,7 +70,7 @@ class SimpleTextFieldsReader extends FieldsProducer {
     in = state.dir.openInput(SimpleTextPostingsFormat.getPostingsFileName(state.segmentInfo.name, state.segmentSuffix), state.context);
    
     fieldInfos = state.fieldInfos;
-    fields = readFields((IndexInput)in.clone());
+    fields = readFields(in.clone());
   }
   
   private TreeMap<String,Long> readFields(IndexInput in) throws IOException {
@@ -230,7 +230,7 @@ class SimpleTextFieldsReader extends FieldsProducer {
     
     public SimpleTextDocsEnum() {
       this.inStart = SimpleTextFieldsReader.this.in;
-      this.in = (IndexInput) this.inStart.clone();
+      this.in = this.inStart.clone();
     }
 
     public boolean canReuse(IndexInput in) {
@@ -330,7 +330,7 @@ class SimpleTextFieldsReader extends FieldsProducer {
 
     public SimpleTextDocsAndPositionsEnum() {
       this.inStart = SimpleTextFieldsReader.this.in;
-      this.in = (IndexInput) inStart.clone();
+      this.in = inStart.clone();
     }
 
     public boolean canReuse(IndexInput in) {
@@ -500,7 +500,7 @@ class SimpleTextFieldsReader extends FieldsProducer {
       final PairOutputs<Long,PairOutputs.Pair<Long,Long>> outputs = new PairOutputs<Long,PairOutputs.Pair<Long,Long>>(posIntOutputs,
                                                                                                                       outputsInner);
       b = new Builder<PairOutputs.Pair<Long,PairOutputs.Pair<Long,Long>>>(FST.INPUT_TYPE.BYTE1, outputs);
-      IndexInput in = (IndexInput) SimpleTextFieldsReader.this.in.clone();
+      IndexInput in = SimpleTextFieldsReader.this.in.clone();
       in.seek(termsStart);
       final BytesRef lastTerm = new BytesRef(10);
       long lastDocsStart = -1;
@@ -608,7 +608,7 @@ class SimpleTextFieldsReader extends FieldsProducer {
 
   @Override
   public Iterator<String> iterator() {
-    return new UnmodifiableIterator<String>(fields.keySet().iterator());
+    return Collections.unmodifiableSet(fields.keySet()).iterator();
   }
 
   private final Map<String,Terms> termsCache = new HashMap<String,Terms>();

lucene/core/src/java/org/apache/lucene/codecs/simpletext/SimpleTextPerDocProducer.java

@@ -132,7 +132,7 @@ public class SimpleTextPerDocProducer extends PerDocProducerBase {
     @Override
     public Source load() throws IOException {
       boolean success = false;
-      IndexInput in = (IndexInput) input.clone();
+      IndexInput in = input.clone();
       try {
         Source source = null;
         switch (type) {

lucene/core/src/java/org/apache/lucene/codecs/simpletext/SimpleTextStoredFieldsReader.java

@@ -163,7 +163,7 @@ public class SimpleTextStoredFieldsReader extends StoredFieldsReader {
     if (in == null) {
       throw new AlreadyClosedException("this FieldsReader is closed");
     }
-    return new SimpleTextStoredFieldsReader(offsets, (IndexInput) in.clone(), fieldInfos);
+    return new SimpleTextStoredFieldsReader(offsets, in.clone(), fieldInfos);
   }
   
   @Override

lucene/core/src/java/org/apache/lucene/codecs/simpletext/SimpleTextTermVectorsReader.java

@@ -19,6 +19,7 @@ package org.apache.lucene.codecs.simpletext;
 
 import java.io.IOException;
 import java.util.ArrayList;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.Iterator;
 import java.util.Map;
@@ -44,8 +45,6 @@ import org.apache.lucene.util.CharsRef;
 import org.apache.lucene.util.IOUtils;
 import org.apache.lucene.util.StringHelper;
 import org.apache.lucene.util.UnicodeUtil;
-import org.apache.lucene.util.UnmodifiableIterator;
-
 import static org.apache.lucene.codecs.simpletext.SimpleTextTermVectorsWriter.*;
 
 /**
@@ -205,7 +204,7 @@ public class SimpleTextTermVectorsReader extends TermVectorsReader {
     if (in == null) {
       throw new AlreadyClosedException("this TermVectorsReader is closed");
     }
-    return new SimpleTextTermVectorsReader(offsets, (IndexInput) in.clone());
+    return new SimpleTextTermVectorsReader(offsets, in.clone());
   }
   
   @Override
@@ -241,7 +240,7 @@ public class SimpleTextTermVectorsReader extends TermVectorsReader {
 
     @Override
     public Iterator<String> iterator() {
-      return new UnmodifiableIterator<String>(fields.keySet().iterator());
+      return Collections.unmodifiableSet(fields.keySet()).iterator();
     }
 
     @Override
@@ -250,7 +249,7 @@ public class SimpleTextTermVectorsReader extends TermVectorsReader {
     }
 
     @Override
-    public int size() throws IOException {
+    public int size() {
       return fields.size();
     }
   }

lucene/core/src/java/org/apache/lucene/document/DocumentStoredFieldVisitor.java

@@ -28,7 +28,8 @@ import org.apache.lucene.index.StoredFieldVisitor;
 
 /** A {@link StoredFieldVisitor} that creates a {@link
  *  Document} containing all stored fields, or only specific
- *  requested fields provided to {@link #DocumentStoredFieldVisitor(Set)}
+ *  requested fields provided to {@link #DocumentStoredFieldVisitor(Set)}.
+ *  <p>
  *  This is used by {@link IndexReader#document(int)} to load a
  *  document.
  *

lucene/core/src/java/org/apache/lucene/document/Field.java

@@ -19,7 +19,6 @@ package org.apache.lucene.document;
 
 import java.io.IOException;
 import java.io.Reader;
-import java.io.StringReader;
 
 import org.apache.lucene.analysis.Analyzer;
 import org.apache.lucene.analysis.NumericTokenStream;
@@ -74,7 +73,8 @@ public class Field implements IndexableField, StorableField {
   // customize how it's tokenized:
   protected TokenStream tokenStream;
 
-  protected transient NumericTokenStream numericTokenStream;
+  private transient TokenStream internalTokenStream;
+  private transient ReusableStringReader internalReader;
 
   protected float boost = 1.0f;
 
@@ -284,9 +284,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Byte)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Byte");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setIntValue(value);
-    }
     fieldsData = Byte.valueOf(value);
   }
 
@@ -294,9 +291,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Short)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Short");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setIntValue(value);
-    }
     fieldsData = Short.valueOf(value);
   }
 
@@ -304,9 +298,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Integer)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Integer");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setIntValue(value);
-    }
     fieldsData = Integer.valueOf(value);
   }
 
@@ -314,9 +305,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Long)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Long");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setLongValue(value);
-    }
     fieldsData = Long.valueOf(value);
   }
 
@@ -324,9 +312,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Float)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Float");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setFloatValue(value);
-    }
     fieldsData = Float.valueOf(value);
   }
 
@@ -334,9 +319,6 @@ public class Field implements IndexableField, StorableField {
     if (!(fieldsData instanceof Double)) {
       throw new IllegalArgumentException("cannot change value type from " + fieldsData.getClass().getSimpleName() + " to Double");
     }
-    if (numericTokenStream != null) {
-      numericTokenStream.setDoubleValue(value);
-    }
     fieldsData = Double.valueOf(value);
   }
 
@@ -434,62 +416,44 @@ public class Field implements IndexableField, StorableField {
 
     final NumericType numericType = fieldType().numericType();
     if (numericType != null) {
-      if (numericTokenStream == null) {
+      if (!(internalTokenStream instanceof NumericTokenStream)) {
         // lazy init the TokenStream as it is heavy to instantiate
         // (attributes,...) if not needed (stored field loading)
-        numericTokenStream = new NumericTokenStream(type.numericPrecisionStep());
-        // initialize value in TokenStream
-        final Number val = (Number) fieldsData;
-        switch (numericType) {
-        case INT:
-          numericTokenStream.setIntValue(val.intValue());
-          break;
-        case LONG:
-          numericTokenStream.setLongValue(val.longValue());
-          break;
-        case FLOAT:
-          numericTokenStream.setFloatValue(val.floatValue());
-          break;
-        case DOUBLE:
-          numericTokenStream.setDoubleValue(val.doubleValue());
-          break;
-        default:
-          assert false : "Should never get here";
-        }
-      } else {
-        // OK -- previously cached and we already updated if
-        // setters were called.
+        internalTokenStream = new NumericTokenStream(type.numericPrecisionStep());
       }
-
-      return numericTokenStream;
+      final NumericTokenStream nts = (NumericTokenStream) internalTokenStream;
+      // initialize value in TokenStream
+      final Number val = (Number) fieldsData;
+      switch (numericType) {
+      case INT:
+        nts.setIntValue(val.intValue());
+        break;
+      case LONG:
+        nts.setLongValue(val.longValue());
+        break;
+      case FLOAT:
+        nts.setFloatValue(val.floatValue());
+        break;
+      case DOUBLE:
+        nts.setDoubleValue(val.doubleValue());
+        break;
+      default:
+        assert false : "Should never get here";
+      }
+      return internalTokenStream;
     }
 
     if (!fieldType().tokenized()) {
       if (stringValue() == null) {
         throw new IllegalArgumentException("Non-Tokenized Fields must have a String value");
       }
-
-      return new TokenStream() {
-        CharTermAttribute termAttribute = addAttribute(CharTermAttribute.class);
-        OffsetAttribute offsetAttribute = addAttribute(OffsetAttribute.class);
-        boolean used;
-
-        @Override
-        public boolean incrementToken() {
-          if (used) {
-            return false;
-          }
-          termAttribute.setEmpty().append(stringValue());
-          offsetAttribute.setOffset(0, stringValue().length());
-          used = true;
-          return true;
-        }
-
-        @Override
-        public void reset() {
-          used = false;
-        }
-      };
+      if (!(internalTokenStream instanceof StringTokenStream)) {
+        // lazy init the TokenStream as it is heavy to instantiate
+        // (attributes,...) if not needed (stored field loading)
+        internalTokenStream = new StringTokenStream();
+      }
+      ((StringTokenStream) internalTokenStream).setValue(stringValue());
+      return internalTokenStream;
     }
 
     if (tokenStream != null) {
@@ -497,11 +461,102 @@ public class Field implements IndexableField, StorableField {
     } else if (readerValue() != null) {
       return analyzer.tokenStream(name(), readerValue());
     } else if (stringValue() != null) {
-      return analyzer.tokenStream(name(), new StringReader(stringValue()));
+      if (internalReader == null) {
+        internalReader = new ReusableStringReader();
+      }
+      internalReader.setValue(stringValue());
+      return analyzer.tokenStream(name(), internalReader);
     }
 
     throw new IllegalArgumentException("Field must have either TokenStream, String, Reader or Number value");
   }
+  
+  static final class ReusableStringReader extends Reader {
+    private int pos = 0, size = 0;
+    private String s = null;
+    
+    void setValue(String s) {
+      this.s = s;
+      this.size = s.length();
+      this.pos = 0;
+    }
+    
+    @Override
+    public int read() {
+      if (pos < size) {
+        return s.charAt(pos++);
+      } else {
+        s = null;
+        return -1;
+      }
+    }
+    
+    @Override
+    public int read(char[] c, int off, int len) {
+      if (pos < size) {
+        len = Math.min(len, size-pos);
+        s.getChars(pos, pos+len, c, off);
+        pos += len;
+        return len;
+      } else {
+        s = null;
+        return -1;
+      }
+    }
+    
+    @Override
+    public void close() {
+      pos = size; // this prevents NPE when reading after close!
+      s = null;
+    }
+  }
+  
+  static final class StringTokenStream extends TokenStream {
+    private final CharTermAttribute termAttribute = addAttribute(CharTermAttribute.class);
+    private final OffsetAttribute offsetAttribute = addAttribute(OffsetAttribute.class);
+    private boolean used = false;
+    private String value = null;
+    
+    /** Creates a new TokenStream that returns a String as single token.
+     * <p>Warning: Does not initialize the value, you must call
+     * {@link #setValue()} afterwards!
+     */
+    StringTokenStream() {
+    }
+    
+    /** Sets the string value. */
+    void setValue(String value) {
+      this.value = value;
+    }
+
+    @Override
+    public boolean incrementToken() {
+      if (used) {
+        return false;
+      }
+      clearAttributes();
+      termAttribute.append(value);
+      offsetAttribute.setOffset(0, value.length());
+      used = true;
+      return true;
+    }
+
+    @Override
+    public void end() {
+      final int finalOffset = value.length();
+      offsetAttribute.setOffset(finalOffset, finalOffset);
+    }
+    
+    @Override
+    public void reset() {
+      used = false;
+    }
+
+    @Override
+    public void close() {
+      value = null;
+    }
+  }
 
   /** Specifies whether and how a field should be stored. */
   public static enum Store {

lucene/core/src/java/org/apache/lucene/index/AtomicReader.java

@@ -175,17 +175,6 @@ public abstract class AtomicReader extends IndexReader {
     }
     return null;
   }
-
-  /** Returns the number of unique terms (across all fields)
-   *  in this reader.
-   */
-  public final long getUniqueTermCount() throws IOException {
-    final Fields fields = fields();
-    if (fields == null) {
-      return 0;
-    }
-    return fields.getUniqueTermCount();
-  }
   
   /**
    * Returns {@link DocValues} for this field.

lucene/core/src/java/org/apache/lucene/index/CheckIndex.java

@@ -1112,21 +1112,6 @@ public class CheckIndex {
         throw new RuntimeException("fieldCount mismatch " + fieldCount + " vs recomputed field count " + computedFieldCount);
       }
     }
-    
-    // for most implementations, this is boring (just the sum across all fields)
-    // but codecs that don't work per-field like preflex actually implement this,
-    // but don't implement it on Terms, so the check isn't redundant.
-    long uniqueTermCountAllFields = fields.getUniqueTermCount();
-    
-    // this means something is seriously screwed, e.g. we are somehow getting enclosed in PFCW!!!!!!
-    
-    if (uniqueTermCountAllFields == -1) {
-      throw new RuntimeException("invalid termCount: -1");
-    }
-    
-    if (status.termCount != uniqueTermCountAllFields) {
-      throw new RuntimeException("termCount mismatch " + uniqueTermCountAllFields + " vs " + (status.termCount));
-    }
 
     if (doPrint) {
       msg("OK [" + status.termCount + " terms; " + status.totFreq + " terms/docs pairs; " + status.totPos + " tokens]");

lucene/core/src/java/org/apache/lucene/index/CoalescedDeletes.java

@@ -47,7 +47,7 @@ class CoalescedDeletes {
 
  public Iterable<Term> termsIterable() {
    return new Iterable<Term>() {
-     @SuppressWarnings("unchecked")
+     @SuppressWarnings({"unchecked","rawtypes"})
      @Override
      public Iterator<Term> iterator() {
        Iterator<Term> subs[] = new Iterator[iterables.size()];

lucene/core/src/java/org/apache/lucene/index/FieldInfos.java

@@ -34,6 +34,8 @@ import org.apache.lucene.index.FieldInfo.IndexOptions;
 public class FieldInfos implements Iterable<FieldInfo> {
   private final boolean hasFreq;
   private final boolean hasProx;
+  private final boolean hasPayloads;
+  private final boolean hasOffsets;
   private final boolean hasVectors;
   private final boolean hasNorms;
   private final boolean hasDocValues;
@@ -45,6 +47,8 @@ public class FieldInfos implements Iterable<FieldInfo> {
   public FieldInfos(FieldInfo[] infos) {
     boolean hasVectors = false;
     boolean hasProx = false;
+    boolean hasPayloads = false;
+    boolean hasOffsets = false;
     boolean hasFreq = false;
     boolean hasNorms = false;
     boolean hasDocValues = false;
@@ -58,12 +62,16 @@ public class FieldInfos implements Iterable<FieldInfo> {
       hasVectors |= info.hasVectors();
       hasProx |= info.isIndexed() && info.getIndexOptions().compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) >= 0;
       hasFreq |= info.isIndexed() && info.getIndexOptions() != IndexOptions.DOCS_ONLY;
+      hasOffsets |= info.isIndexed() && info.getIndexOptions().compareTo(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS_AND_OFFSETS) >= 0;
       hasNorms |= info.hasNorms();
       hasDocValues |= info.hasDocValues();
+      hasPayloads |= info.hasPayloads();
     }
     
     this.hasVectors = hasVectors;
     this.hasProx = hasProx;
+    this.hasPayloads = hasPayloads;
+    this.hasOffsets = hasOffsets;
     this.hasFreq = hasFreq;
     this.hasNorms = hasNorms;
     this.hasDocValues = hasDocValues;
@@ -79,6 +87,16 @@ public class FieldInfos implements Iterable<FieldInfo> {
   public boolean hasProx() {
     return hasProx;
   }
+
+  /** Returns true if any fields have payloads */
+  public boolean hasPayloads() {
+    return hasPayloads;
+  }
+
+  /** Returns true if any fields have offsets */
+  public boolean hasOffsets() {
+    return hasOffsets;
+  }
   
   /**
    * @return true if at least one field has any vectors

lucene/core/src/java/org/apache/lucene/index/Fields.java

@@ -36,28 +36,7 @@ public abstract class Fields implements Iterable<String> {
   /** Returns the number of fields or -1 if the number of
    * distinct field names is unknown. If &gt;= 0,
    * {@link #iterator} will return as many field names. */
-  public abstract int size() throws IOException;
-  
-  /** Returns the number of terms for all fields, or -1 if this 
-   *  measure isn't stored by the codec. Note that, just like 
-   *  other term measures, this measure does not take deleted 
-   *  documents into account. */
-  // TODO: deprecate?
-  public long getUniqueTermCount() throws IOException {
-    long numTerms = 0;
-    for (String field : this) {
-      Terms terms = terms(field);
-      if (terms != null) {
-        final long termCount = terms.size();
-        if (termCount == -1) {
-          return -1;
-        }
-          
-        numTerms += termCount;
-      }
-    }
-    return numTerms;
-  }
+  public abstract int size();
   
   public final static Fields[] EMPTY_ARRAY = new Fields[0];
 }

lucene/core/src/java/org/apache/lucene/index/FilterAtomicReader.java

@@ -57,14 +57,9 @@ public class FilterAtomicReader extends AtomicReader {
     }
 
     @Override
-    public int size() throws IOException {
+    public int size() {
       return in.size();
     }
-
-    @Override
-    public long getUniqueTermCount() throws IOException {
-      return in.getUniqueTermCount();
-    }
   }
 
   /** Base class for filtering {@link Terms}

lucene/core/src/java/org/apache/lucene/index/LogMergePolicy.java

@@ -64,6 +64,13 @@ public abstract class LogMergePolicy extends MergePolicy {
    *  @see #setNoCFSRatio */
   public static final double DEFAULT_NO_CFS_RATIO = 0.1;
 
+  /** Default maxCFSSegmentSize value allows compound file
+   * for a segment of any size. The actual file format is
+   * still subject to noCFSRatio.
+   * @see #setMaxCFSSegmentSizeMB(double)
+   */
+  public static final long DEFAULT_MAX_CFS_SEGMENT_SIZE = Long.MAX_VALUE;
+
   protected int mergeFactor = DEFAULT_MERGE_FACTOR;
 
   protected long minMergeSize;
@@ -74,6 +81,7 @@ public abstract class LogMergePolicy extends MergePolicy {
   protected int maxMergeDocs = DEFAULT_MAX_MERGE_DOCS;
 
   protected double noCFSRatio = DEFAULT_NO_CFS_RATIO;
+  protected long maxCFSSegmentSize = DEFAULT_MAX_CFS_SEGMENT_SIZE;
 
   protected boolean calibrateSizeByDeletes = true;
   
@@ -136,21 +144,21 @@ public abstract class LogMergePolicy extends MergePolicy {
   // Javadoc inherited
   @Override
   public boolean useCompoundFile(SegmentInfos infos, SegmentInfoPerCommit mergedInfo) throws IOException {
-    final boolean doCFS;
-
-    if (!useCompoundFile) {
-      doCFS = false;
-    } else if (noCFSRatio == 1.0) {
-      doCFS = true;
-    } else {
-      long totalSize = 0;
-      for (SegmentInfoPerCommit info : infos) {
-        totalSize += size(info);
-      }
-
-      doCFS = size(mergedInfo) <= noCFSRatio * totalSize;
+    if (!getUseCompoundFile()) {
+      return false;
     }
-    return doCFS;
+    long mergedInfoSize = size(mergedInfo);
+    if (mergedInfoSize > maxCFSSegmentSize) {
+      return false;
+    }
+    if (getNoCFSRatio() >= 1.0) {
+      return true;
+    }
+    long totalSize = 0;
+    for (SegmentInfoPerCommit info : infos) {
+      totalSize += size(info);
+    }
+    return mergedInfoSize <= getNoCFSRatio() * totalSize;
   }
 
   /** Sets whether compound file format should be used for
@@ -674,9 +682,28 @@ public abstract class LogMergePolicy extends MergePolicy {
     sb.append("calibrateSizeByDeletes=").append(calibrateSizeByDeletes).append(", ");
     sb.append("maxMergeDocs=").append(maxMergeDocs).append(", ");
     sb.append("useCompoundFile=").append(useCompoundFile).append(", ");
+    sb.append("maxCFSSegmentSizeMB=").append(getMaxCFSSegmentSizeMB()).append(", ");
     sb.append("noCFSRatio=").append(noCFSRatio);
     sb.append("]");
     return sb.toString();
   }
-  
+
+  /** Returns the largest size allowed for a compound file segment */
+  public final double getMaxCFSSegmentSizeMB() {
+    return maxCFSSegmentSize/1024/1024.;
+  }
+
+  /** If a merged segment will be more than this value,
+   *  leave the segment as
+   *  non-compound file even if compound file is enabled.
+   *  Set this to Double.POSITIVE_INFINITY (default) and noCFSRatio to 1.0
+   *  to always use CFS regardless of merge size. */
+  public final void setMaxCFSSegmentSizeMB(double v) {
+    if (v < 0.0) {
+      throw new IllegalArgumentException("maxCFSSegmentSizeMB must be >=0 (got " + v + ")");
+    }
+    v *= 1024 * 1024;
+    this.maxCFSSegmentSize = (v > Long.MAX_VALUE) ? Long.MAX_VALUE : (long) v;
+  }
+
 }

lucene/core/src/java/org/apache/lucene/index/ParallelAtomicReader.java

@@ -27,7 +27,6 @@ import java.util.SortedMap;
 import java.util.TreeMap;
 
 import org.apache.lucene.util.Bits;
-import org.apache.lucene.util.UnmodifiableIterator;
 
 
 /** An {@link AtomicReader} which reads multiple, parallel indexes.  Each index
@@ -163,7 +162,7 @@ public final class ParallelAtomicReader extends AtomicReader {
     
     @Override
     public Iterator<String> iterator() {
-      return new UnmodifiableIterator<String>(fields.keySet().iterator());
+      return Collections.unmodifiableSet(fields.keySet()).iterator();
     }
     
     @Override

lucene/core/src/java/org/apache/lucene/index/TieredMergePolicy.java

@@ -84,6 +84,7 @@ public class TieredMergePolicy extends MergePolicy {
   private double forceMergeDeletesPctAllowed = 10.0;
   private boolean useCompoundFile = true;
   private double noCFSRatio = 0.1;
+  private long maxCFSSegmentSize = Long.MAX_VALUE;
   private double reclaimDeletesWeight = 2.0;
 
   /** Maximum number of segments to be merged at a time
@@ -127,7 +128,11 @@ public class TieredMergePolicy extends MergePolicy {
    *  sizes of to-be-merged segments (compensating for
    *  percent deleted docs).  Default is 5 GB. */
   public TieredMergePolicy setMaxMergedSegmentMB(double v) {
-    maxMergedSegmentBytes = (long) (v*1024*1024);
+    if (v < 0.0) {
+      throw new IllegalArgumentException("maxMergedSegmentMB must be >=0 (got " + v + ")");
+    }
+    v *= 1024 * 1024;
+    maxMergedSegmentBytes = (v > Long.MAX_VALUE) ? Long.MAX_VALUE : (long) v;
     return this;
   }
 
@@ -162,7 +167,8 @@ public class TieredMergePolicy extends MergePolicy {
     if (v <= 0.0) {
       throw new IllegalArgumentException("floorSegmentMB must be >= 0.0 (got " + v + ")");
     }
-    floorSegmentBytes = (long) (v*1024*1024);
+    v *= 1024 * 1024;
+    floorSegmentBytes = (v > Long.MAX_VALUE) ? Long.MAX_VALUE : (long) v;
     return this;
   }
 
@@ -602,21 +608,21 @@ public class TieredMergePolicy extends MergePolicy {
 
   @Override
   public boolean useCompoundFile(SegmentInfos infos, SegmentInfoPerCommit mergedInfo) throws IOException {
-    final boolean doCFS;
-
-    if (!useCompoundFile) {
-      doCFS = false;
-    } else if (noCFSRatio == 1.0) {
-      doCFS = true;
-    } else {
-      long totalSize = 0;
-      for (SegmentInfoPerCommit info : infos) {
-        totalSize += size(info);
-      }
-
-      doCFS = size(mergedInfo) <= noCFSRatio * totalSize;
+    if (!getUseCompoundFile()) {
+        return false;
     }
-    return doCFS;
+    long mergedInfoSize = size(mergedInfo);
+    if (mergedInfoSize > maxCFSSegmentSize) {
+        return false;
+    }
+    if (getNoCFSRatio() >= 1.0) {
+        return true;
+    }
+    long totalSize = 0;
+    for (SegmentInfoPerCommit info : infos) {
+        totalSize += size(info);
+    }
+    return mergedInfoSize <= getNoCFSRatio() * totalSize;
   }
 
   @Override
@@ -629,7 +635,7 @@ public class TieredMergePolicy extends MergePolicy {
     boolean hasDeletions = w.numDeletedDocs(info) > 0;
     return !hasDeletions &&
       info.info.dir == w.getDirectory() &&
-      (info.info.getUseCompoundFile() == useCompoundFile || noCFSRatio < 1.0);
+      (info.info.getUseCompoundFile() == useCompoundFile || noCFSRatio < 1.0 || maxCFSSegmentSize < Long.MAX_VALUE);
   }
 
   // Segment size in bytes, pro-rated by % deleted
@@ -664,7 +670,27 @@ public class TieredMergePolicy extends MergePolicy {
     sb.append("forceMergeDeletesPctAllowed=").append(forceMergeDeletesPctAllowed).append(", ");
     sb.append("segmentsPerTier=").append(segsPerTier).append(", ");
     sb.append("useCompoundFile=").append(useCompoundFile).append(", ");
+    sb.append("maxCFSSegmentSizeMB=").append(getMaxCFSSegmentSizeMB()).append(", ");
     sb.append("noCFSRatio=").append(noCFSRatio);
     return sb.toString();
   }
+
+  /** Returns the largest size allowed for a compound file segment */
+  public final double getMaxCFSSegmentSizeMB() {
+    return maxCFSSegmentSize/1024/1024.;
+  }
+
+  /** If a merged segment will be more than this value,
+   *  leave the segment as
+   *  non-compound file even if compound file is enabled.
+   *  Set this to Double.POSITIVE_INFINITY (default) and noCFSRatio to 1.0
+   *  to always use CFS regardless of merge size. */
+  public final TieredMergePolicy setMaxCFSSegmentSizeMB(double v) {
+    if (v < 0.0) {
+      throw new IllegalArgumentException("maxCFSSegmentSizeMB must be >=0 (got " + v + ")");
+    }
+    v *= 1024 * 1024;
+    this.maxCFSSegmentSize = (v > Long.MAX_VALUE) ? Long.MAX_VALUE : (long) v;
+    return this;
+  }
 }

lucene/core/src/java/org/apache/lucene/index/package.html

@@ -21,6 +21,241 @@
 </head>
 <body>
 Code to maintain and access indices.
-<!-- TODO: add a BASIC overview here, including code examples of using postings apis -->
+<!-- TODO: add IndexWriter, IndexWriterConfig, DocValues, etc etc -->
+<h2>Table Of Contents</h2>
+<p>
+    <ol>
+        <li><a href="#postings">Postings APIs</a>
+            <ul>
+                <li><a href="#fields">Fields</a></li>
+                <li><a href="#terms">Terms</a></li>
+                <li><a href="#documents">Documents</a></li>
+                <li><a href="#positions">Positions</a></li>
+            </ul>
+        </li>
+        <li><a href="#stats">Index Statistics</a>
+            <ul>
+                <li><a href="#termstats">Term-level</a></li>
+                <li><a href="#fieldstats">Field-level</a></li>
+                <li><a href="#segmentstats">Segment-level</a></li>
+                <li><a href="#documentstats">Document-level</a></li>
+            </ul>
+        </li>
+    </ol>
+</p>
+<a name="postings"></a>
+<h2>Postings APIs</h2>
+<a name="fields"></a>
+<h4>
+    Fields
+</h4>
+<p>
+{@link org.apache.lucene.index.Fields} is the initial entry point into the 
+postings APIs, this can be obtained in several ways:
+<pre class="prettyprint">
+// access indexed fields for an index segment
+Fields fields = reader.fields();
+// access term vector fields for a specified document
+Fields fields = reader.getTermVectors(docid);
+</pre>
+Fields implements Java's Iterable interface, so its easy to enumerate the
+list of fields:
+<pre class="prettyprint">
+// enumerate list of fields
+for (String field : fields) {
+  // access the terms for this field
+  Terms terms = fields.terms(field);
+}
+</pre>
+</p>
+<a name="terms"></a>
+<h4>
+    Terms
+</h4>
+<p>
+{@link org.apache.lucene.index.Terms} represents the collection of terms
+within a field, exposes some metadata and <a href="#fieldstats">statistics</a>,
+and an API for enumeration.
+<pre class="prettyprint">
+// metadata about the field
+System.out.println("positions? " + terms.hasPositions());
+System.out.println("offsets? " + terms.hasOffsets());
+System.out.println("payloads? " + terms.hasPayloads());
+// iterate through terms
+TermsEnum termsEnum = terms.iterator(null);
+BytesRef term = null;
+while ((term = termsEnum.next()) != null) {
+  doSomethingWith(termsEnum.term());
+}
+</pre>
+{@link org.apache.lucene.index.TermsEnum} provides an iterator over the list
+of terms within a field, some <a href="#termstats">statistics</a> about the term,
+and methods to access the term's <a href="#documents">documents</a> and
+<a href="#positions">positions</a>.
+<pre class="prettyprint">
+// seek to a specific term
+boolean found = termsEnum.seekExact(new BytesRef("foobar"), true);
+if (found) {
+  // get the document frequency
+  System.out.println(termsEnum.docFreq());
+  // enumerate through documents
+  DocsEnum docs = termsEnum.docs(null, null);
+  // enumerate through documents and positions
+  DocsAndPositionsEnum docsAndPositions = termsEnum.docsAndPositions(null, null);
+}
+</pre>
+</p>
+<a name="documents"></a>
+<h4>
+    Documents
+</h4>
+<p>
+{@link org.apache.lucene.index.DocsEnum} is an extension of 
+{@link org.apache.lucene.search.DocIdSetIterator}that iterates over the list of
+documents for a term, along with the term frequency within that document.
+<pre class="prettyprint">
+int docid;
+while ((docid = docsEnum.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
+  System.out.println(docid);
+  System.out.println(docsEnum.freq());
+}
+</pre>
+</p>
+<a name="positions"></a>
+<h4>
+    Positions
+</h4>
+<p>
+{@link org.apache.lucene.index.DocsAndPositionsEnum} is an extension of 
+{@link org.apache.lucene.index.DocsEnum} that additionally allows iteration
+of the positions a term occurred within the document, and any additional
+per-position information (offsets and payload)
+<pre class="prettyprint">
+int docid;
+while ((docid = docsAndPositionsEnum.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
+  System.out.println(docid);
+  int freq = docsAndPositionsEnum.freq();
+  for (int i = 0; i < freq; i++) {
+     System.out.println(docsAndPositionsEnum.nextPosition());
+     System.out.println(docsAndPositionsEnum.startOffset());
+     System.out.println(docsAndPositionsEnum.endOffset());
+     System.out.println(docsAndPositionsEnum.getPayload());
+  }
+}
+</pre>
+</p>
+<a name="stats"></a>
+<h2>Index Statistics</h2>
+<a name="termstats"></a>
+<h4>
+    Term statistics
+</h4>
+<p>
+    <ul>
+       <li>{@link org.apache.lucene.index.TermsEnum#docFreq}: Returns the number of 
+           documents that contain at least one occurrence of the term. This statistic 
+           is always available for an indexed term. Note that it will also count 
+           deleted documents, when segments are merged the statistic is updated as 
+           those deleted documents are merged away.
+       <li>{@link org.apache.lucene.index.TermsEnum#totalTermFreq}: Returns the number 
+           of occurrences of this term across all documents. Note that this statistic 
+           is unavailable (returns <code>-1</code>) if term frequencies were omitted 
+           from the index 
+           ({@link org.apache.lucene.index.FieldInfo.IndexOptions#DOCS_ONLY DOCS_ONLY}) 
+           for the field. Like docFreq(), it will also count occurrences that appear in 
+           deleted documents.
+    </ul>
+</p>
+<a name="fieldstats"></a>
+<h4>
+    Field statistics
+</h4>
+<p>
+    <ul>
+       <li>{@link org.apache.lucene.index.Terms#size}: Returns the number of 
+           unique terms in the field. This statistic may be unavailable 
+           (returns <code>-1</code>) for some Terms implementations such as
+           {@link org.apache.lucene.index.MultiTerms}, where it cannot be efficiently
+           computed.  Note that this count also includes terms that appear only
+           in deleted documents: when segments are merged such terms are also merged
+           away and the statistic is then updated.
+       <li>{@link org.apache.lucene.index.Terms#getDocCount}: Returns the number of
+           documents that contain at least one occurrence of any term for this field.
+           This can be thought of as a Field-level docFreq(). Like docFreq() it will
+           also count deleted documents.
+       <li>{@link org.apache.lucene.index.Terms#getSumDocFreq}: Returns the number of
+           postings (term-document mappings in the inverted index) for the field. This
+           can be thought of as the sum of {@link org.apache.lucene.index.TermsEnum#docFreq}
+           across all terms in the field, and like docFreq() it will also count postings
+           that appear in deleted documents.
+       <li>{@link org.apache.lucene.index.Terms#getSumTotalTermFreq}: Returns the number
+           of tokens for the field. This can be thought of as the sum of 
+           {@link org.apache.lucene.index.TermsEnum#totalTermFreq} across all terms in the
+           field, and like totalTermFreq() it will also count occurrences that appear in
+           deleted documents, and will be unavailable (returns <code>-1</code>) if term 
+           frequencies were omitted from the index 
+           ({@link org.apache.lucene.index.FieldInfo.IndexOptions#DOCS_ONLY DOCS_ONLY}) 
+           for the field.
+    </ul>
+</p>
+<a name="segmentstats"></a>
+<h4>
+    Segment statistics
+</h4>
+<p>
+    <ul>
+       <li>{@link org.apache.lucene.index.IndexReader#maxDoc}: Returns the number of 
+           documents (including deleted documents) in the index. 
+       <li>{@link org.apache.lucene.index.IndexReader#numDocs}: Returns the number 
+           of live documents (excluding deleted documents) in the index.
+       <li>{@link org.apache.lucene.index.IndexReader#numDeletedDocs}: Returns the
+           number of deleted documents in the index.
+       <li>{@link org.apache.lucene.index.Fields#size}: Returns the number of indexed
+           fields.
+    </ul>
+</p>
+<a name="documentstats"></a>
+<h4>
+    Document statistics
+</h4>
+<p>
+Document statistics are available during the indexing process for an indexed field: typically
+a {@link org.apache.lucene.search.similarities.Similarity} implementation will store some
+of these values (possibly in a lossy way), into the normalization value for the document in
+its {@link org.apache.lucene.search.similarities.Similarity#computeNorm} method.
+</p>
+<p>
+    <ul>
+       <li>{@link org.apache.lucene.index.FieldInvertState#getLength}: Returns the number of 
+           tokens for this field in the document. Note that this is just the number
+           of times that {@link org.apache.lucene.analysis.TokenStream#incrementToken} returned
+           true, and is unrelated to the values in 
+           {@link org.apache.lucene.analysis.tokenattributes.PositionIncrementAttribute}.
+       <li>{@link org.apache.lucene.index.FieldInvertState#getNumOverlap}: Returns the number
+           of tokens for this field in the document that had a position increment of zero. This
+           can be used to compute a document length that discounts artificial tokens
+           such as synonyms.
+       <li>{@link org.apache.lucene.index.FieldInvertState#getPosition}: Returns the accumulated
+           position value for this field in the document: computed from the values of
+           {@link org.apache.lucene.analysis.tokenattributes.PositionIncrementAttribute} and including
+           {@link org.apache.lucene.analysis.Analyzer#getPositionIncrementGap}s across multivalued
+           fields.
+       <li>{@link org.apache.lucene.index.FieldInvertState#getOffset}: Returns the total
+           character offset value for this field in the document: computed from the values of
+           {@link org.apache.lucene.analysis.tokenattributes.OffsetAttribute} returned by 
+           {@link org.apache.lucene.analysis.TokenStream#end}, and including
+           {@link org.apache.lucene.analysis.Analyzer#getOffsetGap}s across multivalued
+           fields.
+       <li>{@link org.apache.lucene.index.FieldInvertState#getUniqueTermCount}: Returns the number
+           of unique terms encountered for this field in the document.
+       <li>{@link org.apache.lucene.index.FieldInvertState#getMaxTermFrequency}: Returns the maximum
+           frequency across all unique terms encountered for this field in the document. 
+    </ul>
+</p>
+<p>
+Additional user-supplied statistics can be added to the document as DocValues fields and
+accessed via {@link org.apache.lucene.index.AtomicReader#docValues}.
+</p>
+<p>
 </body>
 </html>

lucene/core/src/java/org/apache/lucene/search/BooleanScorer.java

@@ -23,7 +23,6 @@ import java.util.Collection;
 import java.util.List;
 
 import org.apache.lucene.index.AtomicReaderContext;
-import org.apache.lucene.search.BooleanClause.Occur;
 import org.apache.lucene.search.BooleanQuery.BooleanWeight;
 
 /* Description from Doug Cutting (excerpted from
@@ -115,7 +114,7 @@ final class BooleanScorer extends Scorer {
   // Therefore the only methods that are implemented are score() and doc().
   private static final class BucketScorer extends Scorer {
 
-    float score;
+    double score;
     int doc = NO_MORE_DOCS;
     int freq;
     
@@ -134,13 +133,13 @@ final class BooleanScorer extends Scorer {
     public int nextDoc() { return NO_MORE_DOCS; }
     
     @Override
-    public float score() { return score; }
+    public float score() { return (float)score; }
     
   }
 
   static final class Bucket {
     int doc = -1;            // tells if bucket is valid
-    float score;             // incremental score
+    double score;             // incremental score
     // TODO: break out bool anyProhibited, int
     // numRequiredMatched; then we can remove 32 limit on
     // required clauses

lucene/core/src/java/org/apache/lucene/search/ConjunctionScorer.java

@@ -129,6 +129,7 @@ class ConjunctionScorer extends Scorer {
   
   @Override
   public float score() throws IOException {
+    // TODO: sum into a double and cast to float if we ever send required clauses to BS1
     float sum = 0.0f;
     for (int i = 0; i < scorers.length; i++) {
       sum += scorers[i].score();

lucene/core/src/java/org/apache/lucene/search/ConjunctionTermScorer.java

@@ -90,6 +90,7 @@ class ConjunctionTermScorer extends Scorer {
 
   @Override
   public float score() throws IOException {
+    // TODO: sum into a double and cast to float if we ever send required clauses to BS1
     float sum = 0.0f;
     for (DocsAndFreqs docs : docsAndFreqs) {
       sum += docs.scorer.score();

lucene/core/src/java/org/apache/lucene/search/DocIdSetIterator.java

@@ -60,13 +60,17 @@ public abstract class DocIdSetIterator {
    */
   public abstract int nextDoc() throws IOException;
 
-  /**
-   * Advances to the first beyond (see NOTE below) the current whose document
-   * number is greater than or equal to <i>target</i>. Returns the current
-   * document number or {@link #NO_MORE_DOCS} if there are no more docs in the
-   * set.
+ /**
+   * Advances to the first beyond the current whose document number is greater 
+   * than or equal to <i>target</i>, and returns the document number itself. 
+   * Exhausts the iterator and returns {@link #NO_MORE_DOCS} if <i>target</i> 
+   * is greater than the highest document number in the set.
    * <p>
-   * Behaves as if written:
+   * The behavior of this method is <b>undefined</b> when called with
+   * <code> target &le; current</code>, or after the iterator has exhausted.
+   * Both cases may result in unpredicted behavior.
+   * <p>
+   * When <code> target &gt; current</code> it behaves as if written:
    * 
    * <pre>
    * int advance(int target) {
@@ -79,18 +83,12 @@ public abstract class DocIdSetIterator {
    * 
    * Some implementations are considerably more efficient than that.
    * <p>
-   * <b>NOTE:</b> when <code> target &le; current</code> implementations may opt
-   * not to advance beyond their current {@link #docID()}.
-   * <p>
    * <b>NOTE:</b> this method may be called with {@link #NO_MORE_DOCS} for
    * efficiency by some Scorers. If your implementation cannot efficiently
    * determine that it should exhaust, it is recommended that you check for that
    * value in each call to this method.
    * <p>
-   * <b>NOTE:</b> after the iterator has exhausted you should not call this
-   * method, as it may result in unpredicted behavior.
-   * <p>
-   * 
+   *
    * @since 2.9
    */
   public abstract int advance(int target) throws IOException;

lucene/core/src/java/org/apache/lucene/search/FieldCacheImpl.java

@@ -1110,14 +1110,8 @@ class FieldCacheImpl implements FieldCache {
         // Try for coarse estimate for number of bits; this
         // should be an underestimate most of the time, which
         // is fine -- GrowableWriter will reallocate as needed
-        long numUniqueTerms = 0;
-        try {
-          numUniqueTerms = terms.size();
-        } catch (UnsupportedOperationException uoe) {
-          numUniqueTerms = -1;
-        }
-        if (numUniqueTerms != -1) {
-
+        long numUniqueTerms = terms.size();
+        if (numUniqueTerms != -1L) {
           if (numUniqueTerms > termCountHardLimit) {
             // app is misusing the API (there is more than
             // one term per doc); in this case we make best
@@ -1248,13 +1242,8 @@ class FieldCacheImpl implements FieldCache {
         // Try for coarse estimate for number of bits; this
         // should be an underestimate most of the time, which
         // is fine -- GrowableWriter will reallocate as needed
-        long numUniqueTerms = 0;
-        try {
-          numUniqueTerms = terms.size();
-        } catch (UnsupportedOperationException uoe) {
-          numUniqueTerms = -1;
-        }
-        if (numUniqueTerms != -1) {
+        long numUniqueTerms = terms.size();
+        if (numUniqueTerms != -1L) {
           if (numUniqueTerms > termCountHardLimit) {
             numUniqueTerms = termCountHardLimit;
           }

lucene/core/src/java/org/apache/lucene/search/FilteredQuery.java

@@ -83,9 +83,7 @@ public class FilteredQuery extends Query {
       
       @Override
       public boolean scoresDocsOutOfOrder() {
-        // TODO: Support out-of-order scoring!
-        // For now we return false here, as we always get the scorer in order
-        return false;
+        return true;
       }
 
       @Override
@@ -148,9 +146,7 @@ public class FilteredQuery extends Query {
 
         if (useRandomAccess) {
           // if we are using random access, we return the inner scorer, just with other acceptDocs
-          // TODO, replace this by when BooleanWeight is fixed to be consistent with its scorer implementations:
-          // return weight.scorer(context, scoreDocsInOrder, topScorer, filterAcceptDocs);
-          return weight.scorer(context, true, topScorer, filterAcceptDocs);
+          return weight.scorer(context, scoreDocsInOrder, topScorer, filterAcceptDocs);
         } else {
           assert firstFilterDoc > -1;
           // we are gonna advance() this scorer, so we set inorder=true/toplevel=false

lucene/core/src/java/org/apache/lucene/search/Query.java

@@ -28,17 +28,21 @@ import org.apache.lucene.index.Term;
     <p>Instantiable subclasses are:
     <ul>
     <li> {@link TermQuery}
-    <li> {@link MultiTermQuery}
     <li> {@link BooleanQuery}
     <li> {@link WildcardQuery}
     <li> {@link PhraseQuery}
     <li> {@link PrefixQuery}
     <li> {@link MultiPhraseQuery}
     <li> {@link FuzzyQuery}
+    <li> {@link RegexpQuery}
     <li> {@link TermRangeQuery}
     <li> {@link NumericRangeQuery}
-    <li> {@link org.apache.lucene.search.spans.SpanQuery}
+    <li> {@link ConstantScoreQuery}
+    <li> {@link DisjunctionMaxQuery}
+    <li> {@link MatchAllDocsQuery}
     </ul>
+    <p>See also the family of {@link org.apache.lucene.search.spans Span Queries}
+       and additional queries available in the <a href="{@docRoot}/../queries/overview-summary.html">Queries module</a>
 */
 public abstract class Query implements Cloneable {
   private float boost = 1.0f;                     // query boost factor

lucene/core/src/java/org/apache/lucene/search/RegexpQuery.java

@@ -97,7 +97,9 @@ public class RegexpQuery extends AutomatonQuery {
       buffer.append(term.field());
       buffer.append(":");
     }
+    buffer.append('/');
     buffer.append(term.text());
+    buffer.append('/');
     buffer.append(ToStringUtils.boost(getBoost()));
     return buffer.toString();
   }

lucene/core/src/java/org/apache/lucene/search/ReqOptSumScorer.java

@@ -69,6 +69,7 @@ class ReqOptSumScorer extends Scorer {
    */
   @Override
   public float score() throws IOException {
+    // TODO: sum into a double and cast to float if we ever send required clauses to BS1
     int curDoc = reqScorer.docID();
     float reqScore = reqScorer.score();
     if (optScorer == null) {

lucene/core/src/java/org/apache/lucene/store/DataInput.java

@@ -28,6 +28,13 @@ import org.apache.lucene.util.IOUtils;
 /**
  * Abstract base class for performing read operations of Lucene's low-level
  * data types.
+ *
+ * <p>{@code DataInput} may only be used from one thread, because it is not
+ * thread safe (it keeps internal state like file position). To allow
+ * multithreaded use, every {@code DataInput} instance must be cloned before
+ * used in another thread. Subclasses must therefore implement {@link #clone()},
+ * returning a new {@code DataInput} which operates on the same underlying
+ * resource, but positioned independently.
  */
 public abstract class DataInput implements Cloneable {
   /** Reads and returns a single byte.
@@ -195,12 +202,11 @@ public abstract class DataInput implements Cloneable {
    */
   @Override
   public DataInput clone() {
-    DataInput clone = null;
     try {
-      clone = (DataInput)super.clone();
-    } catch (CloneNotSupportedException e) {}
-
-    return clone;
+      return (DataInput) super.clone();
+    } catch (CloneNotSupportedException e) {
+      throw new Error("This cannot happen: Failing to clone DataInput");
+    }
   }
 
   /** Reads a Map&lt;String,String&gt; previously written

lucene/core/src/java/org/apache/lucene/store/DataOutput.java

@@ -27,6 +27,9 @@ import org.apache.lucene.util.UnicodeUtil;
 /**
  * Abstract base class for performing write operations of Lucene's low-level
  * data types.
+ 
+ * <p>{@code DataOutput} may only be used from one thread, because it is not
+ * thread safe (it keeps internal state like file position).
  */
 public abstract class DataOutput {
 

lucene/core/src/java/org/apache/lucene/store/Directory.java

@@ -231,7 +231,7 @@ public abstract class Directory implements Closeable {
       }
       @Override
       public IndexInput openFullSlice() {
-        return (IndexInput) base.clone();
+        return base.clone();
       }
     };
   }
@@ -278,7 +278,7 @@ public abstract class Directory implements Closeable {
     
     SlicedIndexInput(final String sliceDescription, final IndexInput base, final long fileOffset, final long length, int readBufferSize) {
       super("SlicedIndexInput(" + sliceDescription + " in " + base + " slice=" + fileOffset + ":" + (fileOffset+length) + ")", readBufferSize);
-      this.base = (IndexInput) base.clone();
+      this.base = base.clone();
       this.fileOffset = fileOffset;
       this.length = length;
     }
@@ -286,7 +286,7 @@ public abstract class Directory implements Closeable {
     @Override
     public SlicedIndexInput clone() {
       SlicedIndexInput clone = (SlicedIndexInput)super.clone();
-      clone.base = (IndexInput)base.clone();
+      clone.base = base.clone();
       clone.fileOffset = fileOffset;
       clone.length = length;
       return clone;

lucene/core/src/java/org/apache/lucene/store/IndexInput.java

@@ -22,6 +22,17 @@ import java.io.IOException;
 
 /** Abstract base class for input from a file in a {@link Directory}.  A
  * random-access input stream.  Used for all Lucene index input operations.
+ *
+ * <p>{@code IndexInput} may only be used from one thread, because it is not
+ * thread safe (it keeps internal state like file position). To allow
+ * multithreaded use, every {@code IndexInput} instance must be cloned before
+ * used in another thread. Subclasses must therefore implement {@link #clone()},
+ * returning a new {@code IndexInput} which operates on the same underlying
+ * resource, but positioned independently. Lucene never closes cloned
+ * {@code IndexInput}s, it will only do this on the original one.
+ * The original instance must take care that cloned instances throw
+ * {@link AlreadyClosedException} when the original one is closed.
+ 
  * @see Directory
  */
 public abstract class IndexInput extends DataInput implements Cloneable,Closeable {
@@ -83,4 +94,15 @@ public abstract class IndexInput extends DataInput implements Cloneable,Closeabl
   public String toString() {
     return resourceDescription;
   }
+  
+  /** {@inheritDoc}
+   * <p><b>Warning:</b> Lucene never closes cloned
+   * {@code IndexInput}s, it will only do this on the original one.
+   * The original instance must take care that cloned instances throw
+   * {@link AlreadyClosedException} when the original one is closed.
+   */
+  @Override
+  public IndexInput clone() {
+    return (IndexInput) super.clone();
+  }
 }

lucene/core/src/java/org/apache/lucene/store/IndexOutput.java

@@ -22,6 +22,10 @@ import java.io.IOException;
 
 /** Abstract base class for output to a file in a Directory.  A random-access
  * output stream.  Used for all Lucene index output operations.
+ 
+ * <p>{@code IndexOutput} may only be used from one thread, because it is not
+ * thread safe (it keeps internal state like file position).
+ 
  * @see Directory
  * @see IndexInput
  */

lucene/core/src/java/org/apache/lucene/util/UnmodifiableIterator.java

@@ -1,46 +0,0 @@
-package org.apache.lucene.util;
-
-/*
- * Licensed to the Apache Software Foundation (ASF) under one or more
- * contributor license agreements.  See the NOTICE file distributed with
- * this work for additional information regarding copyright ownership.
- * The ASF licenses this file to You under the Apache License, Version 2.0
- * (the "License"); you may not use this file except in compliance with
- * the License.  You may obtain a copy of the License at
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-import java.util.Iterator;
-
-/**
- * Wraps an iterator to ensure its unmodifiable
- */
-public class UnmodifiableIterator<T> implements Iterator<T> {
-  private final Iterator<T> in;
-  
-  public UnmodifiableIterator(Iterator<T> in) {
-    this.in = in;
-  }
-
-  @Override
-  public boolean hasNext() {
-    return in.hasNext();
-  }
-
-  @Override
-  public T next() {
-    return in.next();
-  }
-
-  @Override
-  public void remove() {
-    throw new UnsupportedOperationException();
-  }
-}

lucene/core/src/java/org/apache/lucene/util/WeakIdentityMap.java

@@ -107,6 +107,8 @@ public final class WeakIdentityMap<K,V> {
   public Iterator<K> keyIterator() {
     reap();
     final Iterator<IdentityWeakReference> iterator = backingStore.keySet().iterator();
+    // IMPORTANT: Don't use oal.util.FilterIterator here:
+    // We need *strong* reference to current key after setNext()!!!
     return new Iterator<K>() {
       // holds strong reference to next element in backing iterator:
       private Object next = null;

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked1.java

@@ -0,0 +1,360 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked1 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 1;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 63);
+        values[valuesOffset++] = (int) ((block0 >>> 62) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 61) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 60) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 59) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 58) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 57) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 56) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 55) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 54) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 53) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 52) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 51) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 50) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 49) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 48) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 47) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 46) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 45) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 44) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 43) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 42) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 41) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 40) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 39) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 38) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 37) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 36) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 35) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 34) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 33) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 32) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 31) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 30) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 29) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 28) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 27) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 26) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 25) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 24) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 23) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 22) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 21) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 20) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 19) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 18) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 17) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 16) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 15) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 14) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 13) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 12) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 11) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 10) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 9) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 8) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 7) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 6) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 5) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 4) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 3) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 2) & 1L);
+        values[valuesOffset++] = (int) ((block0 >>> 1) & 1L);
+        values[valuesOffset++] = (int) (block0 & 1L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte0 >>> 7;
+        values[valuesOffset++] = (byte0 >>> 6) & 1;
+        values[valuesOffset++] = (byte0 >>> 5) & 1;
+        values[valuesOffset++] = (byte0 >>> 4) & 1;
+        values[valuesOffset++] = (byte0 >>> 3) & 1;
+        values[valuesOffset++] = (byte0 >>> 2) & 1;
+        values[valuesOffset++] = (byte0 >>> 1) & 1;
+        values[valuesOffset++] = byte0 & 1;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte1 >>> 7;
+        values[valuesOffset++] = (byte1 >>> 6) & 1;
+        values[valuesOffset++] = (byte1 >>> 5) & 1;
+        values[valuesOffset++] = (byte1 >>> 4) & 1;
+        values[valuesOffset++] = (byte1 >>> 3) & 1;
+        values[valuesOffset++] = (byte1 >>> 2) & 1;
+        values[valuesOffset++] = (byte1 >>> 1) & 1;
+        values[valuesOffset++] = byte1 & 1;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte2 >>> 7;
+        values[valuesOffset++] = (byte2 >>> 6) & 1;
+        values[valuesOffset++] = (byte2 >>> 5) & 1;
+        values[valuesOffset++] = (byte2 >>> 4) & 1;
+        values[valuesOffset++] = (byte2 >>> 3) & 1;
+        values[valuesOffset++] = (byte2 >>> 2) & 1;
+        values[valuesOffset++] = (byte2 >>> 1) & 1;
+        values[valuesOffset++] = byte2 & 1;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte3 >>> 7;
+        values[valuesOffset++] = (byte3 >>> 6) & 1;
+        values[valuesOffset++] = (byte3 >>> 5) & 1;
+        values[valuesOffset++] = (byte3 >>> 4) & 1;
+        values[valuesOffset++] = (byte3 >>> 3) & 1;
+        values[valuesOffset++] = (byte3 >>> 2) & 1;
+        values[valuesOffset++] = (byte3 >>> 1) & 1;
+        values[valuesOffset++] = byte3 & 1;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte4 >>> 7;
+        values[valuesOffset++] = (byte4 >>> 6) & 1;
+        values[valuesOffset++] = (byte4 >>> 5) & 1;
+        values[valuesOffset++] = (byte4 >>> 4) & 1;
+        values[valuesOffset++] = (byte4 >>> 3) & 1;
+        values[valuesOffset++] = (byte4 >>> 2) & 1;
+        values[valuesOffset++] = (byte4 >>> 1) & 1;
+        values[valuesOffset++] = byte4 & 1;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte5 >>> 7;
+        values[valuesOffset++] = (byte5 >>> 6) & 1;
+        values[valuesOffset++] = (byte5 >>> 5) & 1;
+        values[valuesOffset++] = (byte5 >>> 4) & 1;
+        values[valuesOffset++] = (byte5 >>> 3) & 1;
+        values[valuesOffset++] = (byte5 >>> 2) & 1;
+        values[valuesOffset++] = (byte5 >>> 1) & 1;
+        values[valuesOffset++] = byte5 & 1;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte6 >>> 7;
+        values[valuesOffset++] = (byte6 >>> 6) & 1;
+        values[valuesOffset++] = (byte6 >>> 5) & 1;
+        values[valuesOffset++] = (byte6 >>> 4) & 1;
+        values[valuesOffset++] = (byte6 >>> 3) & 1;
+        values[valuesOffset++] = (byte6 >>> 2) & 1;
+        values[valuesOffset++] = (byte6 >>> 1) & 1;
+        values[valuesOffset++] = byte6 & 1;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte7 >>> 7;
+        values[valuesOffset++] = (byte7 >>> 6) & 1;
+        values[valuesOffset++] = (byte7 >>> 5) & 1;
+        values[valuesOffset++] = (byte7 >>> 4) & 1;
+        values[valuesOffset++] = (byte7 >>> 3) & 1;
+        values[valuesOffset++] = (byte7 >>> 2) & 1;
+        values[valuesOffset++] = (byte7 >>> 1) & 1;
+        values[valuesOffset++] = byte7 & 1;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 63;
+        values[valuesOffset++] = (block0 >>> 62) & 1L;
+        values[valuesOffset++] = (block0 >>> 61) & 1L;
+        values[valuesOffset++] = (block0 >>> 60) & 1L;
+        values[valuesOffset++] = (block0 >>> 59) & 1L;
+        values[valuesOffset++] = (block0 >>> 58) & 1L;
+        values[valuesOffset++] = (block0 >>> 57) & 1L;
+        values[valuesOffset++] = (block0 >>> 56) & 1L;
+        values[valuesOffset++] = (block0 >>> 55) & 1L;
+        values[valuesOffset++] = (block0 >>> 54) & 1L;
+        values[valuesOffset++] = (block0 >>> 53) & 1L;
+        values[valuesOffset++] = (block0 >>> 52) & 1L;
+        values[valuesOffset++] = (block0 >>> 51) & 1L;
+        values[valuesOffset++] = (block0 >>> 50) & 1L;
+        values[valuesOffset++] = (block0 >>> 49) & 1L;
+        values[valuesOffset++] = (block0 >>> 48) & 1L;
+        values[valuesOffset++] = (block0 >>> 47) & 1L;
+        values[valuesOffset++] = (block0 >>> 46) & 1L;
+        values[valuesOffset++] = (block0 >>> 45) & 1L;
+        values[valuesOffset++] = (block0 >>> 44) & 1L;
+        values[valuesOffset++] = (block0 >>> 43) & 1L;
+        values[valuesOffset++] = (block0 >>> 42) & 1L;
+        values[valuesOffset++] = (block0 >>> 41) & 1L;
+        values[valuesOffset++] = (block0 >>> 40) & 1L;
+        values[valuesOffset++] = (block0 >>> 39) & 1L;
+        values[valuesOffset++] = (block0 >>> 38) & 1L;
+        values[valuesOffset++] = (block0 >>> 37) & 1L;
+        values[valuesOffset++] = (block0 >>> 36) & 1L;
+        values[valuesOffset++] = (block0 >>> 35) & 1L;
+        values[valuesOffset++] = (block0 >>> 34) & 1L;
+        values[valuesOffset++] = (block0 >>> 33) & 1L;
+        values[valuesOffset++] = (block0 >>> 32) & 1L;
+        values[valuesOffset++] = (block0 >>> 31) & 1L;
+        values[valuesOffset++] = (block0 >>> 30) & 1L;
+        values[valuesOffset++] = (block0 >>> 29) & 1L;
+        values[valuesOffset++] = (block0 >>> 28) & 1L;
+        values[valuesOffset++] = (block0 >>> 27) & 1L;
+        values[valuesOffset++] = (block0 >>> 26) & 1L;
+        values[valuesOffset++] = (block0 >>> 25) & 1L;
+        values[valuesOffset++] = (block0 >>> 24) & 1L;
+        values[valuesOffset++] = (block0 >>> 23) & 1L;
+        values[valuesOffset++] = (block0 >>> 22) & 1L;
+        values[valuesOffset++] = (block0 >>> 21) & 1L;
+        values[valuesOffset++] = (block0 >>> 20) & 1L;
+        values[valuesOffset++] = (block0 >>> 19) & 1L;
+        values[valuesOffset++] = (block0 >>> 18) & 1L;
+        values[valuesOffset++] = (block0 >>> 17) & 1L;
+        values[valuesOffset++] = (block0 >>> 16) & 1L;
+        values[valuesOffset++] = (block0 >>> 15) & 1L;
+        values[valuesOffset++] = (block0 >>> 14) & 1L;
+        values[valuesOffset++] = (block0 >>> 13) & 1L;
+        values[valuesOffset++] = (block0 >>> 12) & 1L;
+        values[valuesOffset++] = (block0 >>> 11) & 1L;
+        values[valuesOffset++] = (block0 >>> 10) & 1L;
+        values[valuesOffset++] = (block0 >>> 9) & 1L;
+        values[valuesOffset++] = (block0 >>> 8) & 1L;
+        values[valuesOffset++] = (block0 >>> 7) & 1L;
+        values[valuesOffset++] = (block0 >>> 6) & 1L;
+        values[valuesOffset++] = (block0 >>> 5) & 1L;
+        values[valuesOffset++] = (block0 >>> 4) & 1L;
+        values[valuesOffset++] = (block0 >>> 3) & 1L;
+        values[valuesOffset++] = (block0 >>> 2) & 1L;
+        values[valuesOffset++] = (block0 >>> 1) & 1L;
+        values[valuesOffset++] = block0 & 1L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte0 >>> 7;
+        values[valuesOffset++] = (byte0 >>> 6) & 1;
+        values[valuesOffset++] = (byte0 >>> 5) & 1;
+        values[valuesOffset++] = (byte0 >>> 4) & 1;
+        values[valuesOffset++] = (byte0 >>> 3) & 1;
+        values[valuesOffset++] = (byte0 >>> 2) & 1;
+        values[valuesOffset++] = (byte0 >>> 1) & 1;
+        values[valuesOffset++] = byte0 & 1;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte1 >>> 7;
+        values[valuesOffset++] = (byte1 >>> 6) & 1;
+        values[valuesOffset++] = (byte1 >>> 5) & 1;
+        values[valuesOffset++] = (byte1 >>> 4) & 1;
+        values[valuesOffset++] = (byte1 >>> 3) & 1;
+        values[valuesOffset++] = (byte1 >>> 2) & 1;
+        values[valuesOffset++] = (byte1 >>> 1) & 1;
+        values[valuesOffset++] = byte1 & 1;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte2 >>> 7;
+        values[valuesOffset++] = (byte2 >>> 6) & 1;
+        values[valuesOffset++] = (byte2 >>> 5) & 1;
+        values[valuesOffset++] = (byte2 >>> 4) & 1;
+        values[valuesOffset++] = (byte2 >>> 3) & 1;
+        values[valuesOffset++] = (byte2 >>> 2) & 1;
+        values[valuesOffset++] = (byte2 >>> 1) & 1;
+        values[valuesOffset++] = byte2 & 1;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte3 >>> 7;
+        values[valuesOffset++] = (byte3 >>> 6) & 1;
+        values[valuesOffset++] = (byte3 >>> 5) & 1;
+        values[valuesOffset++] = (byte3 >>> 4) & 1;
+        values[valuesOffset++] = (byte3 >>> 3) & 1;
+        values[valuesOffset++] = (byte3 >>> 2) & 1;
+        values[valuesOffset++] = (byte3 >>> 1) & 1;
+        values[valuesOffset++] = byte3 & 1;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte4 >>> 7;
+        values[valuesOffset++] = (byte4 >>> 6) & 1;
+        values[valuesOffset++] = (byte4 >>> 5) & 1;
+        values[valuesOffset++] = (byte4 >>> 4) & 1;
+        values[valuesOffset++] = (byte4 >>> 3) & 1;
+        values[valuesOffset++] = (byte4 >>> 2) & 1;
+        values[valuesOffset++] = (byte4 >>> 1) & 1;
+        values[valuesOffset++] = byte4 & 1;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte5 >>> 7;
+        values[valuesOffset++] = (byte5 >>> 6) & 1;
+        values[valuesOffset++] = (byte5 >>> 5) & 1;
+        values[valuesOffset++] = (byte5 >>> 4) & 1;
+        values[valuesOffset++] = (byte5 >>> 3) & 1;
+        values[valuesOffset++] = (byte5 >>> 2) & 1;
+        values[valuesOffset++] = (byte5 >>> 1) & 1;
+        values[valuesOffset++] = byte5 & 1;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte6 >>> 7;
+        values[valuesOffset++] = (byte6 >>> 6) & 1;
+        values[valuesOffset++] = (byte6 >>> 5) & 1;
+        values[valuesOffset++] = (byte6 >>> 4) & 1;
+        values[valuesOffset++] = (byte6 >>> 3) & 1;
+        values[valuesOffset++] = (byte6 >>> 2) & 1;
+        values[valuesOffset++] = (byte6 >>> 1) & 1;
+        values[valuesOffset++] = byte6 & 1;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte7 >>> 7;
+        values[valuesOffset++] = (byte7 >>> 6) & 1;
+        values[valuesOffset++] = (byte7 >>> 5) & 1;
+        values[valuesOffset++] = (byte7 >>> 4) & 1;
+        values[valuesOffset++] = (byte7 >>> 3) & 1;
+        values[valuesOffset++] = (byte7 >>> 2) & 1;
+        values[valuesOffset++] = (byte7 >>> 1) & 1;
+        values[valuesOffset++] = byte7 & 1;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 19) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 17) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 15) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset++] & 0xffffffffL) << 13) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset++] & 0xffffffffL) << 11) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset++] & 0xffffffffL) << 1) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 62) | (values[valuesOffset++] << 61) | (values[valuesOffset++] << 60) | (values[valuesOffset++] << 59) | (values[valuesOffset++] << 58) | (values[valuesOffset++] << 57) | (values[valuesOffset++] << 56) | (values[valuesOffset++] << 55) | (values[valuesOffset++] << 54) | (values[valuesOffset++] << 53) | (values[valuesOffset++] << 52) | (values[valuesOffset++] << 51) | (values[valuesOffset++] << 50) | (values[valuesOffset++] << 49) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 47) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 45) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 19) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 17) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 15) | (values[valuesOffset++] << 14) | (values[valuesOffset++] << 13) | (values[valuesOffset++] << 12) | (values[valuesOffset++] << 11) | (values[valuesOffset++] << 10) | (values[valuesOffset++] << 9) | (values[valuesOffset++] << 8) | (values[valuesOffset++] << 7) | (values[valuesOffset++] << 6) | (values[valuesOffset++] << 5) | (values[valuesOffset++] << 4) | (values[valuesOffset++] << 3) | (values[valuesOffset++] << 2) | (values[valuesOffset++] << 1) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked10.java

@@ -0,0 +1,312 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked10 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 5;
+    }
+
+    @Override
+    public int valueCount() {
+      return 32;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 54);
+        values[valuesOffset++] = (int) ((block0 >>> 44) & 1023L);
+        values[valuesOffset++] = (int) ((block0 >>> 34) & 1023L);
+        values[valuesOffset++] = (int) ((block0 >>> 24) & 1023L);
+        values[valuesOffset++] = (int) ((block0 >>> 14) & 1023L);
+        values[valuesOffset++] = (int) ((block0 >>> 4) & 1023L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 15L) << 6) | (block1 >>> 58));
+        values[valuesOffset++] = (int) ((block1 >>> 48) & 1023L);
+        values[valuesOffset++] = (int) ((block1 >>> 38) & 1023L);
+        values[valuesOffset++] = (int) ((block1 >>> 28) & 1023L);
+        values[valuesOffset++] = (int) ((block1 >>> 18) & 1023L);
+        values[valuesOffset++] = (int) ((block1 >>> 8) & 1023L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 255L) << 2) | (block2 >>> 62));
+        values[valuesOffset++] = (int) ((block2 >>> 52) & 1023L);
+        values[valuesOffset++] = (int) ((block2 >>> 42) & 1023L);
+        values[valuesOffset++] = (int) ((block2 >>> 32) & 1023L);
+        values[valuesOffset++] = (int) ((block2 >>> 22) & 1023L);
+        values[valuesOffset++] = (int) ((block2 >>> 12) & 1023L);
+        values[valuesOffset++] = (int) ((block2 >>> 2) & 1023L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 3L) << 8) | (block3 >>> 56));
+        values[valuesOffset++] = (int) ((block3 >>> 46) & 1023L);
+        values[valuesOffset++] = (int) ((block3 >>> 36) & 1023L);
+        values[valuesOffset++] = (int) ((block3 >>> 26) & 1023L);
+        values[valuesOffset++] = (int) ((block3 >>> 16) & 1023L);
+        values[valuesOffset++] = (int) ((block3 >>> 6) & 1023L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 63L) << 4) | (block4 >>> 60));
+        values[valuesOffset++] = (int) ((block4 >>> 50) & 1023L);
+        values[valuesOffset++] = (int) ((block4 >>> 40) & 1023L);
+        values[valuesOffset++] = (int) ((block4 >>> 30) & 1023L);
+        values[valuesOffset++] = (int) ((block4 >>> 20) & 1023L);
+        values[valuesOffset++] = (int) ((block4 >>> 10) & 1023L);
+        values[valuesOffset++] = (int) (block4 & 1023L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 2) | (byte1 >>> 6);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 63) << 4) | (byte2 >>> 4);
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 15) << 6) | (byte3 >>> 2);
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 3) << 8) | byte4;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte5 << 2) | (byte6 >>> 6);
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 63) << 4) | (byte7 >>> 4);
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 6) | (byte8 >>> 2);
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 3) << 8) | byte9;
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte10 << 2) | (byte11 >>> 6);
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 4) | (byte12 >>> 4);
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 15) << 6) | (byte13 >>> 2);
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 3) << 8) | byte14;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 2) | (byte16 >>> 6);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 63) << 4) | (byte17 >>> 4);
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 15) << 6) | (byte18 >>> 2);
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte18 & 3) << 8) | byte19;
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte20 << 2) | (byte21 >>> 6);
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 63) << 4) | (byte22 >>> 4);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 6) | (byte23 >>> 2);
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 3) << 8) | byte24;
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte25 << 2) | (byte26 >>> 6);
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 4) | (byte27 >>> 4);
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 15) << 6) | (byte28 >>> 2);
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 3) << 8) | byte29;
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte30 << 2) | (byte31 >>> 6);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 63) << 4) | (byte32 >>> 4);
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte32 & 15) << 6) | (byte33 >>> 2);
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 8) | byte34;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte35 << 2) | (byte36 >>> 6);
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 63) << 4) | (byte37 >>> 4);
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 15) << 6) | (byte38 >>> 2);
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 3) << 8) | byte39;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 54;
+        values[valuesOffset++] = (block0 >>> 44) & 1023L;
+        values[valuesOffset++] = (block0 >>> 34) & 1023L;
+        values[valuesOffset++] = (block0 >>> 24) & 1023L;
+        values[valuesOffset++] = (block0 >>> 14) & 1023L;
+        values[valuesOffset++] = (block0 >>> 4) & 1023L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 15L) << 6) | (block1 >>> 58);
+        values[valuesOffset++] = (block1 >>> 48) & 1023L;
+        values[valuesOffset++] = (block1 >>> 38) & 1023L;
+        values[valuesOffset++] = (block1 >>> 28) & 1023L;
+        values[valuesOffset++] = (block1 >>> 18) & 1023L;
+        values[valuesOffset++] = (block1 >>> 8) & 1023L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 255L) << 2) | (block2 >>> 62);
+        values[valuesOffset++] = (block2 >>> 52) & 1023L;
+        values[valuesOffset++] = (block2 >>> 42) & 1023L;
+        values[valuesOffset++] = (block2 >>> 32) & 1023L;
+        values[valuesOffset++] = (block2 >>> 22) & 1023L;
+        values[valuesOffset++] = (block2 >>> 12) & 1023L;
+        values[valuesOffset++] = (block2 >>> 2) & 1023L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 3L) << 8) | (block3 >>> 56);
+        values[valuesOffset++] = (block3 >>> 46) & 1023L;
+        values[valuesOffset++] = (block3 >>> 36) & 1023L;
+        values[valuesOffset++] = (block3 >>> 26) & 1023L;
+        values[valuesOffset++] = (block3 >>> 16) & 1023L;
+        values[valuesOffset++] = (block3 >>> 6) & 1023L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 63L) << 4) | (block4 >>> 60);
+        values[valuesOffset++] = (block4 >>> 50) & 1023L;
+        values[valuesOffset++] = (block4 >>> 40) & 1023L;
+        values[valuesOffset++] = (block4 >>> 30) & 1023L;
+        values[valuesOffset++] = (block4 >>> 20) & 1023L;
+        values[valuesOffset++] = (block4 >>> 10) & 1023L;
+        values[valuesOffset++] = block4 & 1023L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 2) | (byte1 >>> 6);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 63) << 4) | (byte2 >>> 4);
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 15) << 6) | (byte3 >>> 2);
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 3) << 8) | byte4;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte5 << 2) | (byte6 >>> 6);
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 63) << 4) | (byte7 >>> 4);
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 6) | (byte8 >>> 2);
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 3) << 8) | byte9;
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte10 << 2) | (byte11 >>> 6);
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 4) | (byte12 >>> 4);
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 15) << 6) | (byte13 >>> 2);
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 3) << 8) | byte14;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 2) | (byte16 >>> 6);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 63) << 4) | (byte17 >>> 4);
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 15) << 6) | (byte18 >>> 2);
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte18 & 3) << 8) | byte19;
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte20 << 2) | (byte21 >>> 6);
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 63) << 4) | (byte22 >>> 4);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 6) | (byte23 >>> 2);
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 3) << 8) | byte24;
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte25 << 2) | (byte26 >>> 6);
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 4) | (byte27 >>> 4);
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 15) << 6) | (byte28 >>> 2);
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 3) << 8) | byte29;
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte30 << 2) | (byte31 >>> 6);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 63) << 4) | (byte32 >>> 4);
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte32 & 15) << 6) | (byte33 >>> 2);
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 8) | byte34;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte35 << 2) | (byte36 >>> 6);
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 63) << 4) | (byte37 >>> 4);
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 15) << 6) | (byte38 >>> 2);
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 3) << 8) | byte39;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 10) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 14) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 52) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 12) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 50) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 10) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked11.java

@@ -0,0 +1,560 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked11 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 11;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 53);
+        values[valuesOffset++] = (int) ((block0 >>> 42) & 2047L);
+        values[valuesOffset++] = (int) ((block0 >>> 31) & 2047L);
+        values[valuesOffset++] = (int) ((block0 >>> 20) & 2047L);
+        values[valuesOffset++] = (int) ((block0 >>> 9) & 2047L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 511L) << 2) | (block1 >>> 62));
+        values[valuesOffset++] = (int) ((block1 >>> 51) & 2047L);
+        values[valuesOffset++] = (int) ((block1 >>> 40) & 2047L);
+        values[valuesOffset++] = (int) ((block1 >>> 29) & 2047L);
+        values[valuesOffset++] = (int) ((block1 >>> 18) & 2047L);
+        values[valuesOffset++] = (int) ((block1 >>> 7) & 2047L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 127L) << 4) | (block2 >>> 60));
+        values[valuesOffset++] = (int) ((block2 >>> 49) & 2047L);
+        values[valuesOffset++] = (int) ((block2 >>> 38) & 2047L);
+        values[valuesOffset++] = (int) ((block2 >>> 27) & 2047L);
+        values[valuesOffset++] = (int) ((block2 >>> 16) & 2047L);
+        values[valuesOffset++] = (int) ((block2 >>> 5) & 2047L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 31L) << 6) | (block3 >>> 58));
+        values[valuesOffset++] = (int) ((block3 >>> 47) & 2047L);
+        values[valuesOffset++] = (int) ((block3 >>> 36) & 2047L);
+        values[valuesOffset++] = (int) ((block3 >>> 25) & 2047L);
+        values[valuesOffset++] = (int) ((block3 >>> 14) & 2047L);
+        values[valuesOffset++] = (int) ((block3 >>> 3) & 2047L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 7L) << 8) | (block4 >>> 56));
+        values[valuesOffset++] = (int) ((block4 >>> 45) & 2047L);
+        values[valuesOffset++] = (int) ((block4 >>> 34) & 2047L);
+        values[valuesOffset++] = (int) ((block4 >>> 23) & 2047L);
+        values[valuesOffset++] = (int) ((block4 >>> 12) & 2047L);
+        values[valuesOffset++] = (int) ((block4 >>> 1) & 2047L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 1L) << 10) | (block5 >>> 54));
+        values[valuesOffset++] = (int) ((block5 >>> 43) & 2047L);
+        values[valuesOffset++] = (int) ((block5 >>> 32) & 2047L);
+        values[valuesOffset++] = (int) ((block5 >>> 21) & 2047L);
+        values[valuesOffset++] = (int) ((block5 >>> 10) & 2047L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 1023L) << 1) | (block6 >>> 63));
+        values[valuesOffset++] = (int) ((block6 >>> 52) & 2047L);
+        values[valuesOffset++] = (int) ((block6 >>> 41) & 2047L);
+        values[valuesOffset++] = (int) ((block6 >>> 30) & 2047L);
+        values[valuesOffset++] = (int) ((block6 >>> 19) & 2047L);
+        values[valuesOffset++] = (int) ((block6 >>> 8) & 2047L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 255L) << 3) | (block7 >>> 61));
+        values[valuesOffset++] = (int) ((block7 >>> 50) & 2047L);
+        values[valuesOffset++] = (int) ((block7 >>> 39) & 2047L);
+        values[valuesOffset++] = (int) ((block7 >>> 28) & 2047L);
+        values[valuesOffset++] = (int) ((block7 >>> 17) & 2047L);
+        values[valuesOffset++] = (int) ((block7 >>> 6) & 2047L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 63L) << 5) | (block8 >>> 59));
+        values[valuesOffset++] = (int) ((block8 >>> 48) & 2047L);
+        values[valuesOffset++] = (int) ((block8 >>> 37) & 2047L);
+        values[valuesOffset++] = (int) ((block8 >>> 26) & 2047L);
+        values[valuesOffset++] = (int) ((block8 >>> 15) & 2047L);
+        values[valuesOffset++] = (int) ((block8 >>> 4) & 2047L);
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block8 & 15L) << 7) | (block9 >>> 57));
+        values[valuesOffset++] = (int) ((block9 >>> 46) & 2047L);
+        values[valuesOffset++] = (int) ((block9 >>> 35) & 2047L);
+        values[valuesOffset++] = (int) ((block9 >>> 24) & 2047L);
+        values[valuesOffset++] = (int) ((block9 >>> 13) & 2047L);
+        values[valuesOffset++] = (int) ((block9 >>> 2) & 2047L);
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block9 & 3L) << 9) | (block10 >>> 55));
+        values[valuesOffset++] = (int) ((block10 >>> 44) & 2047L);
+        values[valuesOffset++] = (int) ((block10 >>> 33) & 2047L);
+        values[valuesOffset++] = (int) ((block10 >>> 22) & 2047L);
+        values[valuesOffset++] = (int) ((block10 >>> 11) & 2047L);
+        values[valuesOffset++] = (int) (block10 & 2047L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 3) | (byte1 >>> 5);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 31) << 6) | (byte2 >>> 2);
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 3) << 9) | (byte3 << 1) | (byte4 >>> 7);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 127) << 4) | (byte5 >>> 4);
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 15) << 7) | (byte6 >>> 1);
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 1) << 10) | (byte7 << 2) | (byte8 >>> 6);
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 63) << 5) | (byte9 >>> 3);
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 7) << 8) | byte10;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte11 << 3) | (byte12 >>> 5);
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 31) << 6) | (byte13 >>> 2);
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 3) << 9) | (byte14 << 1) | (byte15 >>> 7);
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 127) << 4) | (byte16 >>> 4);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 15) << 7) | (byte17 >>> 1);
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 1) << 10) | (byte18 << 2) | (byte19 >>> 6);
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 63) << 5) | (byte20 >>> 3);
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 7) << 8) | byte21;
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte22 << 3) | (byte23 >>> 5);
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 31) << 6) | (byte24 >>> 2);
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 3) << 9) | (byte25 << 1) | (byte26 >>> 7);
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 127) << 4) | (byte27 >>> 4);
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 15) << 7) | (byte28 >>> 1);
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 1) << 10) | (byte29 << 2) | (byte30 >>> 6);
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 63) << 5) | (byte31 >>> 3);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 7) << 8) | byte32;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte33 << 3) | (byte34 >>> 5);
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte34 & 31) << 6) | (byte35 >>> 2);
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 3) << 9) | (byte36 << 1) | (byte37 >>> 7);
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 127) << 4) | (byte38 >>> 4);
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 15) << 7) | (byte39 >>> 1);
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte39 & 1) << 10) | (byte40 << 2) | (byte41 >>> 6);
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte41 & 63) << 5) | (byte42 >>> 3);
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 7) << 8) | byte43;
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte44 << 3) | (byte45 >>> 5);
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 31) << 6) | (byte46 >>> 2);
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 3) << 9) | (byte47 << 1) | (byte48 >>> 7);
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 127) << 4) | (byte49 >>> 4);
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 15) << 7) | (byte50 >>> 1);
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 1) << 10) | (byte51 << 2) | (byte52 >>> 6);
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 63) << 5) | (byte53 >>> 3);
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 7) << 8) | byte54;
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte55 << 3) | (byte56 >>> 5);
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 31) << 6) | (byte57 >>> 2);
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte57 & 3) << 9) | (byte58 << 1) | (byte59 >>> 7);
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 127) << 4) | (byte60 >>> 4);
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 15) << 7) | (byte61 >>> 1);
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 1) << 10) | (byte62 << 2) | (byte63 >>> 6);
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 63) << 5) | (byte64 >>> 3);
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte64 & 7) << 8) | byte65;
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte66 << 3) | (byte67 >>> 5);
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 31) << 6) | (byte68 >>> 2);
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 3) << 9) | (byte69 << 1) | (byte70 >>> 7);
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte70 & 127) << 4) | (byte71 >>> 4);
+        final int byte72 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 15) << 7) | (byte72 >>> 1);
+        final int byte73 = blocks[blocksOffset++] & 0xFF;
+        final int byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte72 & 1) << 10) | (byte73 << 2) | (byte74 >>> 6);
+        final int byte75 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 63) << 5) | (byte75 >>> 3);
+        final int byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte75 & 7) << 8) | byte76;
+        final int byte77 = blocks[blocksOffset++] & 0xFF;
+        final int byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte77 << 3) | (byte78 >>> 5);
+        final int byte79 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 31) << 6) | (byte79 >>> 2);
+        final int byte80 = blocks[blocksOffset++] & 0xFF;
+        final int byte81 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte79 & 3) << 9) | (byte80 << 1) | (byte81 >>> 7);
+        final int byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte81 & 127) << 4) | (byte82 >>> 4);
+        final int byte83 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 15) << 7) | (byte83 >>> 1);
+        final int byte84 = blocks[blocksOffset++] & 0xFF;
+        final int byte85 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte83 & 1) << 10) | (byte84 << 2) | (byte85 >>> 6);
+        final int byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte85 & 63) << 5) | (byte86 >>> 3);
+        final int byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 7) << 8) | byte87;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 53;
+        values[valuesOffset++] = (block0 >>> 42) & 2047L;
+        values[valuesOffset++] = (block0 >>> 31) & 2047L;
+        values[valuesOffset++] = (block0 >>> 20) & 2047L;
+        values[valuesOffset++] = (block0 >>> 9) & 2047L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 511L) << 2) | (block1 >>> 62);
+        values[valuesOffset++] = (block1 >>> 51) & 2047L;
+        values[valuesOffset++] = (block1 >>> 40) & 2047L;
+        values[valuesOffset++] = (block1 >>> 29) & 2047L;
+        values[valuesOffset++] = (block1 >>> 18) & 2047L;
+        values[valuesOffset++] = (block1 >>> 7) & 2047L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 127L) << 4) | (block2 >>> 60);
+        values[valuesOffset++] = (block2 >>> 49) & 2047L;
+        values[valuesOffset++] = (block2 >>> 38) & 2047L;
+        values[valuesOffset++] = (block2 >>> 27) & 2047L;
+        values[valuesOffset++] = (block2 >>> 16) & 2047L;
+        values[valuesOffset++] = (block2 >>> 5) & 2047L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 31L) << 6) | (block3 >>> 58);
+        values[valuesOffset++] = (block3 >>> 47) & 2047L;
+        values[valuesOffset++] = (block3 >>> 36) & 2047L;
+        values[valuesOffset++] = (block3 >>> 25) & 2047L;
+        values[valuesOffset++] = (block3 >>> 14) & 2047L;
+        values[valuesOffset++] = (block3 >>> 3) & 2047L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 7L) << 8) | (block4 >>> 56);
+        values[valuesOffset++] = (block4 >>> 45) & 2047L;
+        values[valuesOffset++] = (block4 >>> 34) & 2047L;
+        values[valuesOffset++] = (block4 >>> 23) & 2047L;
+        values[valuesOffset++] = (block4 >>> 12) & 2047L;
+        values[valuesOffset++] = (block4 >>> 1) & 2047L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 1L) << 10) | (block5 >>> 54);
+        values[valuesOffset++] = (block5 >>> 43) & 2047L;
+        values[valuesOffset++] = (block5 >>> 32) & 2047L;
+        values[valuesOffset++] = (block5 >>> 21) & 2047L;
+        values[valuesOffset++] = (block5 >>> 10) & 2047L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 1023L) << 1) | (block6 >>> 63);
+        values[valuesOffset++] = (block6 >>> 52) & 2047L;
+        values[valuesOffset++] = (block6 >>> 41) & 2047L;
+        values[valuesOffset++] = (block6 >>> 30) & 2047L;
+        values[valuesOffset++] = (block6 >>> 19) & 2047L;
+        values[valuesOffset++] = (block6 >>> 8) & 2047L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 255L) << 3) | (block7 >>> 61);
+        values[valuesOffset++] = (block7 >>> 50) & 2047L;
+        values[valuesOffset++] = (block7 >>> 39) & 2047L;
+        values[valuesOffset++] = (block7 >>> 28) & 2047L;
+        values[valuesOffset++] = (block7 >>> 17) & 2047L;
+        values[valuesOffset++] = (block7 >>> 6) & 2047L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 63L) << 5) | (block8 >>> 59);
+        values[valuesOffset++] = (block8 >>> 48) & 2047L;
+        values[valuesOffset++] = (block8 >>> 37) & 2047L;
+        values[valuesOffset++] = (block8 >>> 26) & 2047L;
+        values[valuesOffset++] = (block8 >>> 15) & 2047L;
+        values[valuesOffset++] = (block8 >>> 4) & 2047L;
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block8 & 15L) << 7) | (block9 >>> 57);
+        values[valuesOffset++] = (block9 >>> 46) & 2047L;
+        values[valuesOffset++] = (block9 >>> 35) & 2047L;
+        values[valuesOffset++] = (block9 >>> 24) & 2047L;
+        values[valuesOffset++] = (block9 >>> 13) & 2047L;
+        values[valuesOffset++] = (block9 >>> 2) & 2047L;
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block9 & 3L) << 9) | (block10 >>> 55);
+        values[valuesOffset++] = (block10 >>> 44) & 2047L;
+        values[valuesOffset++] = (block10 >>> 33) & 2047L;
+        values[valuesOffset++] = (block10 >>> 22) & 2047L;
+        values[valuesOffset++] = (block10 >>> 11) & 2047L;
+        values[valuesOffset++] = block10 & 2047L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 3) | (byte1 >>> 5);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 31) << 6) | (byte2 >>> 2);
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 3) << 9) | (byte3 << 1) | (byte4 >>> 7);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 127) << 4) | (byte5 >>> 4);
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 15) << 7) | (byte6 >>> 1);
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 1) << 10) | (byte7 << 2) | (byte8 >>> 6);
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 63) << 5) | (byte9 >>> 3);
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 7) << 8) | byte10;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte11 << 3) | (byte12 >>> 5);
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 31) << 6) | (byte13 >>> 2);
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 3) << 9) | (byte14 << 1) | (byte15 >>> 7);
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 127) << 4) | (byte16 >>> 4);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 15) << 7) | (byte17 >>> 1);
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 1) << 10) | (byte18 << 2) | (byte19 >>> 6);
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 63) << 5) | (byte20 >>> 3);
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 7) << 8) | byte21;
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte22 << 3) | (byte23 >>> 5);
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 31) << 6) | (byte24 >>> 2);
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 3) << 9) | (byte25 << 1) | (byte26 >>> 7);
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 127) << 4) | (byte27 >>> 4);
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 15) << 7) | (byte28 >>> 1);
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 1) << 10) | (byte29 << 2) | (byte30 >>> 6);
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 63) << 5) | (byte31 >>> 3);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 7) << 8) | byte32;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte33 << 3) | (byte34 >>> 5);
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte34 & 31) << 6) | (byte35 >>> 2);
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 3) << 9) | (byte36 << 1) | (byte37 >>> 7);
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 127) << 4) | (byte38 >>> 4);
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 15) << 7) | (byte39 >>> 1);
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte39 & 1) << 10) | (byte40 << 2) | (byte41 >>> 6);
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte41 & 63) << 5) | (byte42 >>> 3);
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 7) << 8) | byte43;
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte44 << 3) | (byte45 >>> 5);
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 31) << 6) | (byte46 >>> 2);
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 3) << 9) | (byte47 << 1) | (byte48 >>> 7);
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 127) << 4) | (byte49 >>> 4);
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 15) << 7) | (byte50 >>> 1);
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 1) << 10) | (byte51 << 2) | (byte52 >>> 6);
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 63) << 5) | (byte53 >>> 3);
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 7) << 8) | byte54;
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte55 << 3) | (byte56 >>> 5);
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 31) << 6) | (byte57 >>> 2);
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte57 & 3) << 9) | (byte58 << 1) | (byte59 >>> 7);
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 127) << 4) | (byte60 >>> 4);
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 15) << 7) | (byte61 >>> 1);
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 1) << 10) | (byte62 << 2) | (byte63 >>> 6);
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 63) << 5) | (byte64 >>> 3);
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte64 & 7) << 8) | byte65;
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte66 << 3) | (byte67 >>> 5);
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 31) << 6) | (byte68 >>> 2);
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 3) << 9) | (byte69 << 1) | (byte70 >>> 7);
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte70 & 127) << 4) | (byte71 >>> 4);
+        final long byte72 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 15) << 7) | (byte72 >>> 1);
+        final long byte73 = blocks[blocksOffset++] & 0xFF;
+        final long byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte72 & 1) << 10) | (byte73 << 2) | (byte74 >>> 6);
+        final long byte75 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 63) << 5) | (byte75 >>> 3);
+        final long byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte75 & 7) << 8) | byte76;
+        final long byte77 = blocks[blocksOffset++] & 0xFF;
+        final long byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte77 << 3) | (byte78 >>> 5);
+        final long byte79 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 31) << 6) | (byte79 >>> 2);
+        final long byte80 = blocks[blocksOffset++] & 0xFF;
+        final long byte81 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte79 & 3) << 9) | (byte80 << 1) | (byte81 >>> 7);
+        final long byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte81 & 127) << 4) | (byte82 >>> 4);
+        final long byte83 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 15) << 7) | (byte83 >>> 1);
+        final long byte84 = blocks[blocksOffset++] & 0xFF;
+        final long byte85 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte83 & 1) << 10) | (byte84 << 2) | (byte85 >>> 6);
+        final long byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte85 & 63) << 5) | (byte86 >>> 3);
+        final long byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 7) << 8) | byte87;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset++] & 0xffffffffL) << 1) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 1);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 19) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 3);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 17) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 5);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 15) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 7);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 13) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 9);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 11) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 53) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 9) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 51) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 7) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 49) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 5) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 47) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 14) | (values[valuesOffset++] << 3) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 45) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 12) | (values[valuesOffset++] << 1) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 1);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 52) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 19) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 3);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 61) | (values[valuesOffset++] << 50) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 17) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 5);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 59) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 15) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 7);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 57) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 13) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 9);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 55) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 11) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked12.java

@@ -0,0 +1,208 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked12 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 3;
+    }
+
+    @Override
+    public int valueCount() {
+      return 16;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 52);
+        values[valuesOffset++] = (int) ((block0 >>> 40) & 4095L);
+        values[valuesOffset++] = (int) ((block0 >>> 28) & 4095L);
+        values[valuesOffset++] = (int) ((block0 >>> 16) & 4095L);
+        values[valuesOffset++] = (int) ((block0 >>> 4) & 4095L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 15L) << 8) | (block1 >>> 56));
+        values[valuesOffset++] = (int) ((block1 >>> 44) & 4095L);
+        values[valuesOffset++] = (int) ((block1 >>> 32) & 4095L);
+        values[valuesOffset++] = (int) ((block1 >>> 20) & 4095L);
+        values[valuesOffset++] = (int) ((block1 >>> 8) & 4095L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 255L) << 4) | (block2 >>> 60));
+        values[valuesOffset++] = (int) ((block2 >>> 48) & 4095L);
+        values[valuesOffset++] = (int) ((block2 >>> 36) & 4095L);
+        values[valuesOffset++] = (int) ((block2 >>> 24) & 4095L);
+        values[valuesOffset++] = (int) ((block2 >>> 12) & 4095L);
+        values[valuesOffset++] = (int) (block2 & 4095L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 4) | (byte1 >>> 4);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 15) << 8) | byte2;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte3 << 4) | (byte4 >>> 4);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 15) << 8) | byte5;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte6 << 4) | (byte7 >>> 4);
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 8) | byte8;
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte9 << 4) | (byte10 >>> 4);
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 15) << 8) | byte11;
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte12 << 4) | (byte13 >>> 4);
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 15) << 8) | byte14;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 4) | (byte16 >>> 4);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 15) << 8) | byte17;
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte18 << 4) | (byte19 >>> 4);
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 15) << 8) | byte20;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte21 << 4) | (byte22 >>> 4);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 8) | byte23;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 52;
+        values[valuesOffset++] = (block0 >>> 40) & 4095L;
+        values[valuesOffset++] = (block0 >>> 28) & 4095L;
+        values[valuesOffset++] = (block0 >>> 16) & 4095L;
+        values[valuesOffset++] = (block0 >>> 4) & 4095L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 15L) << 8) | (block1 >>> 56);
+        values[valuesOffset++] = (block1 >>> 44) & 4095L;
+        values[valuesOffset++] = (block1 >>> 32) & 4095L;
+        values[valuesOffset++] = (block1 >>> 20) & 4095L;
+        values[valuesOffset++] = (block1 >>> 8) & 4095L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 255L) << 4) | (block2 >>> 60);
+        values[valuesOffset++] = (block2 >>> 48) & 4095L;
+        values[valuesOffset++] = (block2 >>> 36) & 4095L;
+        values[valuesOffset++] = (block2 >>> 24) & 4095L;
+        values[valuesOffset++] = (block2 >>> 12) & 4095L;
+        values[valuesOffset++] = block2 & 4095L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 4) | (byte1 >>> 4);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 15) << 8) | byte2;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte3 << 4) | (byte4 >>> 4);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 15) << 8) | byte5;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte6 << 4) | (byte7 >>> 4);
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 8) | byte8;
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte9 << 4) | (byte10 >>> 4);
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 15) << 8) | byte11;
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte12 << 4) | (byte13 >>> 4);
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 15) << 8) | byte14;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 4) | (byte16 >>> 4);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 15) << 8) | byte17;
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte18 << 4) | (byte19 >>> 4);
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 15) << 8) | byte20;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte21 << 4) | (byte22 >>> 4);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 8) | byte23;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 12) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 12) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked13.java

@@ -0,0 +1,600 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked13 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 13;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 51);
+        values[valuesOffset++] = (int) ((block0 >>> 38) & 8191L);
+        values[valuesOffset++] = (int) ((block0 >>> 25) & 8191L);
+        values[valuesOffset++] = (int) ((block0 >>> 12) & 8191L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 4095L) << 1) | (block1 >>> 63));
+        values[valuesOffset++] = (int) ((block1 >>> 50) & 8191L);
+        values[valuesOffset++] = (int) ((block1 >>> 37) & 8191L);
+        values[valuesOffset++] = (int) ((block1 >>> 24) & 8191L);
+        values[valuesOffset++] = (int) ((block1 >>> 11) & 8191L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 2047L) << 2) | (block2 >>> 62));
+        values[valuesOffset++] = (int) ((block2 >>> 49) & 8191L);
+        values[valuesOffset++] = (int) ((block2 >>> 36) & 8191L);
+        values[valuesOffset++] = (int) ((block2 >>> 23) & 8191L);
+        values[valuesOffset++] = (int) ((block2 >>> 10) & 8191L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 1023L) << 3) | (block3 >>> 61));
+        values[valuesOffset++] = (int) ((block3 >>> 48) & 8191L);
+        values[valuesOffset++] = (int) ((block3 >>> 35) & 8191L);
+        values[valuesOffset++] = (int) ((block3 >>> 22) & 8191L);
+        values[valuesOffset++] = (int) ((block3 >>> 9) & 8191L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 511L) << 4) | (block4 >>> 60));
+        values[valuesOffset++] = (int) ((block4 >>> 47) & 8191L);
+        values[valuesOffset++] = (int) ((block4 >>> 34) & 8191L);
+        values[valuesOffset++] = (int) ((block4 >>> 21) & 8191L);
+        values[valuesOffset++] = (int) ((block4 >>> 8) & 8191L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 255L) << 5) | (block5 >>> 59));
+        values[valuesOffset++] = (int) ((block5 >>> 46) & 8191L);
+        values[valuesOffset++] = (int) ((block5 >>> 33) & 8191L);
+        values[valuesOffset++] = (int) ((block5 >>> 20) & 8191L);
+        values[valuesOffset++] = (int) ((block5 >>> 7) & 8191L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 127L) << 6) | (block6 >>> 58));
+        values[valuesOffset++] = (int) ((block6 >>> 45) & 8191L);
+        values[valuesOffset++] = (int) ((block6 >>> 32) & 8191L);
+        values[valuesOffset++] = (int) ((block6 >>> 19) & 8191L);
+        values[valuesOffset++] = (int) ((block6 >>> 6) & 8191L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 63L) << 7) | (block7 >>> 57));
+        values[valuesOffset++] = (int) ((block7 >>> 44) & 8191L);
+        values[valuesOffset++] = (int) ((block7 >>> 31) & 8191L);
+        values[valuesOffset++] = (int) ((block7 >>> 18) & 8191L);
+        values[valuesOffset++] = (int) ((block7 >>> 5) & 8191L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 31L) << 8) | (block8 >>> 56));
+        values[valuesOffset++] = (int) ((block8 >>> 43) & 8191L);
+        values[valuesOffset++] = (int) ((block8 >>> 30) & 8191L);
+        values[valuesOffset++] = (int) ((block8 >>> 17) & 8191L);
+        values[valuesOffset++] = (int) ((block8 >>> 4) & 8191L);
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block8 & 15L) << 9) | (block9 >>> 55));
+        values[valuesOffset++] = (int) ((block9 >>> 42) & 8191L);
+        values[valuesOffset++] = (int) ((block9 >>> 29) & 8191L);
+        values[valuesOffset++] = (int) ((block9 >>> 16) & 8191L);
+        values[valuesOffset++] = (int) ((block9 >>> 3) & 8191L);
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block9 & 7L) << 10) | (block10 >>> 54));
+        values[valuesOffset++] = (int) ((block10 >>> 41) & 8191L);
+        values[valuesOffset++] = (int) ((block10 >>> 28) & 8191L);
+        values[valuesOffset++] = (int) ((block10 >>> 15) & 8191L);
+        values[valuesOffset++] = (int) ((block10 >>> 2) & 8191L);
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block10 & 3L) << 11) | (block11 >>> 53));
+        values[valuesOffset++] = (int) ((block11 >>> 40) & 8191L);
+        values[valuesOffset++] = (int) ((block11 >>> 27) & 8191L);
+        values[valuesOffset++] = (int) ((block11 >>> 14) & 8191L);
+        values[valuesOffset++] = (int) ((block11 >>> 1) & 8191L);
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block11 & 1L) << 12) | (block12 >>> 52));
+        values[valuesOffset++] = (int) ((block12 >>> 39) & 8191L);
+        values[valuesOffset++] = (int) ((block12 >>> 26) & 8191L);
+        values[valuesOffset++] = (int) ((block12 >>> 13) & 8191L);
+        values[valuesOffset++] = (int) (block12 & 8191L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 5) | (byte1 >>> 3);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 7) << 10) | (byte2 << 2) | (byte3 >>> 6);
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 63) << 7) | (byte4 >>> 1);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 1) << 12) | (byte5 << 4) | (byte6 >>> 4);
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 15) << 9) | (byte7 << 1) | (byte8 >>> 7);
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 127) << 6) | (byte9 >>> 2);
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 3) << 11) | (byte10 << 3) | (byte11 >>> 5);
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 31) << 8) | byte12;
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte13 << 5) | (byte14 >>> 3);
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 7) << 10) | (byte15 << 2) | (byte16 >>> 6);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 63) << 7) | (byte17 >>> 1);
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 1) << 12) | (byte18 << 4) | (byte19 >>> 4);
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 15) << 9) | (byte20 << 1) | (byte21 >>> 7);
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 127) << 6) | (byte22 >>> 2);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 3) << 11) | (byte23 << 3) | (byte24 >>> 5);
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 31) << 8) | byte25;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte26 << 5) | (byte27 >>> 3);
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 7) << 10) | (byte28 << 2) | (byte29 >>> 6);
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 63) << 7) | (byte30 >>> 1);
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 1) << 12) | (byte31 << 4) | (byte32 >>> 4);
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte32 & 15) << 9) | (byte33 << 1) | (byte34 >>> 7);
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte34 & 127) << 6) | (byte35 >>> 2);
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 3) << 11) | (byte36 << 3) | (byte37 >>> 5);
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 31) << 8) | byte38;
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte39 << 5) | (byte40 >>> 3);
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 7) << 10) | (byte41 << 2) | (byte42 >>> 6);
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 63) << 7) | (byte43 >>> 1);
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 1) << 12) | (byte44 << 4) | (byte45 >>> 4);
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 15) << 9) | (byte46 << 1) | (byte47 >>> 7);
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 127) << 6) | (byte48 >>> 2);
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 3) << 11) | (byte49 << 3) | (byte50 >>> 5);
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 31) << 8) | byte51;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte52 << 5) | (byte53 >>> 3);
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 7) << 10) | (byte54 << 2) | (byte55 >>> 6);
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte55 & 63) << 7) | (byte56 >>> 1);
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 1) << 12) | (byte57 << 4) | (byte58 >>> 4);
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 15) << 9) | (byte59 << 1) | (byte60 >>> 7);
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 127) << 6) | (byte61 >>> 2);
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 3) << 11) | (byte62 << 3) | (byte63 >>> 5);
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 31) << 8) | byte64;
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte65 << 5) | (byte66 >>> 3);
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte66 & 7) << 10) | (byte67 << 2) | (byte68 >>> 6);
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 63) << 7) | (byte69 >>> 1);
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 1) << 12) | (byte70 << 4) | (byte71 >>> 4);
+        final int byte72 = blocks[blocksOffset++] & 0xFF;
+        final int byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 15) << 9) | (byte72 << 1) | (byte73 >>> 7);
+        final int byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 127) << 6) | (byte74 >>> 2);
+        final int byte75 = blocks[blocksOffset++] & 0xFF;
+        final int byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 3) << 11) | (byte75 << 3) | (byte76 >>> 5);
+        final int byte77 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 31) << 8) | byte77;
+        final int byte78 = blocks[blocksOffset++] & 0xFF;
+        final int byte79 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte78 << 5) | (byte79 >>> 3);
+        final int byte80 = blocks[blocksOffset++] & 0xFF;
+        final int byte81 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte79 & 7) << 10) | (byte80 << 2) | (byte81 >>> 6);
+        final int byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte81 & 63) << 7) | (byte82 >>> 1);
+        final int byte83 = blocks[blocksOffset++] & 0xFF;
+        final int byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 1) << 12) | (byte83 << 4) | (byte84 >>> 4);
+        final int byte85 = blocks[blocksOffset++] & 0xFF;
+        final int byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte84 & 15) << 9) | (byte85 << 1) | (byte86 >>> 7);
+        final int byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 127) << 6) | (byte87 >>> 2);
+        final int byte88 = blocks[blocksOffset++] & 0xFF;
+        final int byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 3) << 11) | (byte88 << 3) | (byte89 >>> 5);
+        final int byte90 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte89 & 31) << 8) | byte90;
+        final int byte91 = blocks[blocksOffset++] & 0xFF;
+        final int byte92 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte91 << 5) | (byte92 >>> 3);
+        final int byte93 = blocks[blocksOffset++] & 0xFF;
+        final int byte94 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte92 & 7) << 10) | (byte93 << 2) | (byte94 >>> 6);
+        final int byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte94 & 63) << 7) | (byte95 >>> 1);
+        final int byte96 = blocks[blocksOffset++] & 0xFF;
+        final int byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 1) << 12) | (byte96 << 4) | (byte97 >>> 4);
+        final int byte98 = blocks[blocksOffset++] & 0xFF;
+        final int byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 15) << 9) | (byte98 << 1) | (byte99 >>> 7);
+        final int byte100 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 127) << 6) | (byte100 >>> 2);
+        final int byte101 = blocks[blocksOffset++] & 0xFF;
+        final int byte102 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte100 & 3) << 11) | (byte101 << 3) | (byte102 >>> 5);
+        final int byte103 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte102 & 31) << 8) | byte103;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 51;
+        values[valuesOffset++] = (block0 >>> 38) & 8191L;
+        values[valuesOffset++] = (block0 >>> 25) & 8191L;
+        values[valuesOffset++] = (block0 >>> 12) & 8191L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 4095L) << 1) | (block1 >>> 63);
+        values[valuesOffset++] = (block1 >>> 50) & 8191L;
+        values[valuesOffset++] = (block1 >>> 37) & 8191L;
+        values[valuesOffset++] = (block1 >>> 24) & 8191L;
+        values[valuesOffset++] = (block1 >>> 11) & 8191L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 2047L) << 2) | (block2 >>> 62);
+        values[valuesOffset++] = (block2 >>> 49) & 8191L;
+        values[valuesOffset++] = (block2 >>> 36) & 8191L;
+        values[valuesOffset++] = (block2 >>> 23) & 8191L;
+        values[valuesOffset++] = (block2 >>> 10) & 8191L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 1023L) << 3) | (block3 >>> 61);
+        values[valuesOffset++] = (block3 >>> 48) & 8191L;
+        values[valuesOffset++] = (block3 >>> 35) & 8191L;
+        values[valuesOffset++] = (block3 >>> 22) & 8191L;
+        values[valuesOffset++] = (block3 >>> 9) & 8191L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 511L) << 4) | (block4 >>> 60);
+        values[valuesOffset++] = (block4 >>> 47) & 8191L;
+        values[valuesOffset++] = (block4 >>> 34) & 8191L;
+        values[valuesOffset++] = (block4 >>> 21) & 8191L;
+        values[valuesOffset++] = (block4 >>> 8) & 8191L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 255L) << 5) | (block5 >>> 59);
+        values[valuesOffset++] = (block5 >>> 46) & 8191L;
+        values[valuesOffset++] = (block5 >>> 33) & 8191L;
+        values[valuesOffset++] = (block5 >>> 20) & 8191L;
+        values[valuesOffset++] = (block5 >>> 7) & 8191L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 127L) << 6) | (block6 >>> 58);
+        values[valuesOffset++] = (block6 >>> 45) & 8191L;
+        values[valuesOffset++] = (block6 >>> 32) & 8191L;
+        values[valuesOffset++] = (block6 >>> 19) & 8191L;
+        values[valuesOffset++] = (block6 >>> 6) & 8191L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 63L) << 7) | (block7 >>> 57);
+        values[valuesOffset++] = (block7 >>> 44) & 8191L;
+        values[valuesOffset++] = (block7 >>> 31) & 8191L;
+        values[valuesOffset++] = (block7 >>> 18) & 8191L;
+        values[valuesOffset++] = (block7 >>> 5) & 8191L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 31L) << 8) | (block8 >>> 56);
+        values[valuesOffset++] = (block8 >>> 43) & 8191L;
+        values[valuesOffset++] = (block8 >>> 30) & 8191L;
+        values[valuesOffset++] = (block8 >>> 17) & 8191L;
+        values[valuesOffset++] = (block8 >>> 4) & 8191L;
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block8 & 15L) << 9) | (block9 >>> 55);
+        values[valuesOffset++] = (block9 >>> 42) & 8191L;
+        values[valuesOffset++] = (block9 >>> 29) & 8191L;
+        values[valuesOffset++] = (block9 >>> 16) & 8191L;
+        values[valuesOffset++] = (block9 >>> 3) & 8191L;
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block9 & 7L) << 10) | (block10 >>> 54);
+        values[valuesOffset++] = (block10 >>> 41) & 8191L;
+        values[valuesOffset++] = (block10 >>> 28) & 8191L;
+        values[valuesOffset++] = (block10 >>> 15) & 8191L;
+        values[valuesOffset++] = (block10 >>> 2) & 8191L;
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block10 & 3L) << 11) | (block11 >>> 53);
+        values[valuesOffset++] = (block11 >>> 40) & 8191L;
+        values[valuesOffset++] = (block11 >>> 27) & 8191L;
+        values[valuesOffset++] = (block11 >>> 14) & 8191L;
+        values[valuesOffset++] = (block11 >>> 1) & 8191L;
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block11 & 1L) << 12) | (block12 >>> 52);
+        values[valuesOffset++] = (block12 >>> 39) & 8191L;
+        values[valuesOffset++] = (block12 >>> 26) & 8191L;
+        values[valuesOffset++] = (block12 >>> 13) & 8191L;
+        values[valuesOffset++] = block12 & 8191L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 5) | (byte1 >>> 3);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 7) << 10) | (byte2 << 2) | (byte3 >>> 6);
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 63) << 7) | (byte4 >>> 1);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 1) << 12) | (byte5 << 4) | (byte6 >>> 4);
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 15) << 9) | (byte7 << 1) | (byte8 >>> 7);
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 127) << 6) | (byte9 >>> 2);
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 3) << 11) | (byte10 << 3) | (byte11 >>> 5);
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 31) << 8) | byte12;
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte13 << 5) | (byte14 >>> 3);
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 7) << 10) | (byte15 << 2) | (byte16 >>> 6);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 63) << 7) | (byte17 >>> 1);
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 1) << 12) | (byte18 << 4) | (byte19 >>> 4);
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 15) << 9) | (byte20 << 1) | (byte21 >>> 7);
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 127) << 6) | (byte22 >>> 2);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 3) << 11) | (byte23 << 3) | (byte24 >>> 5);
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 31) << 8) | byte25;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte26 << 5) | (byte27 >>> 3);
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 7) << 10) | (byte28 << 2) | (byte29 >>> 6);
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 63) << 7) | (byte30 >>> 1);
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 1) << 12) | (byte31 << 4) | (byte32 >>> 4);
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte32 & 15) << 9) | (byte33 << 1) | (byte34 >>> 7);
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte34 & 127) << 6) | (byte35 >>> 2);
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 3) << 11) | (byte36 << 3) | (byte37 >>> 5);
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 31) << 8) | byte38;
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte39 << 5) | (byte40 >>> 3);
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 7) << 10) | (byte41 << 2) | (byte42 >>> 6);
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 63) << 7) | (byte43 >>> 1);
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 1) << 12) | (byte44 << 4) | (byte45 >>> 4);
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 15) << 9) | (byte46 << 1) | (byte47 >>> 7);
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 127) << 6) | (byte48 >>> 2);
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 3) << 11) | (byte49 << 3) | (byte50 >>> 5);
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 31) << 8) | byte51;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte52 << 5) | (byte53 >>> 3);
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 7) << 10) | (byte54 << 2) | (byte55 >>> 6);
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte55 & 63) << 7) | (byte56 >>> 1);
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 1) << 12) | (byte57 << 4) | (byte58 >>> 4);
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 15) << 9) | (byte59 << 1) | (byte60 >>> 7);
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 127) << 6) | (byte61 >>> 2);
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 3) << 11) | (byte62 << 3) | (byte63 >>> 5);
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 31) << 8) | byte64;
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte65 << 5) | (byte66 >>> 3);
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte66 & 7) << 10) | (byte67 << 2) | (byte68 >>> 6);
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 63) << 7) | (byte69 >>> 1);
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 1) << 12) | (byte70 << 4) | (byte71 >>> 4);
+        final long byte72 = blocks[blocksOffset++] & 0xFF;
+        final long byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 15) << 9) | (byte72 << 1) | (byte73 >>> 7);
+        final long byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 127) << 6) | (byte74 >>> 2);
+        final long byte75 = blocks[blocksOffset++] & 0xFF;
+        final long byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 3) << 11) | (byte75 << 3) | (byte76 >>> 5);
+        final long byte77 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 31) << 8) | byte77;
+        final long byte78 = blocks[blocksOffset++] & 0xFF;
+        final long byte79 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte78 << 5) | (byte79 >>> 3);
+        final long byte80 = blocks[blocksOffset++] & 0xFF;
+        final long byte81 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte79 & 7) << 10) | (byte80 << 2) | (byte81 >>> 6);
+        final long byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte81 & 63) << 7) | (byte82 >>> 1);
+        final long byte83 = blocks[blocksOffset++] & 0xFF;
+        final long byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 1) << 12) | (byte83 << 4) | (byte84 >>> 4);
+        final long byte85 = blocks[blocksOffset++] & 0xFF;
+        final long byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte84 & 15) << 9) | (byte85 << 1) | (byte86 >>> 7);
+        final long byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 127) << 6) | (byte87 >>> 2);
+        final long byte88 = blocks[blocksOffset++] & 0xFF;
+        final long byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 3) << 11) | (byte88 << 3) | (byte89 >>> 5);
+        final long byte90 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte89 & 31) << 8) | byte90;
+        final long byte91 = blocks[blocksOffset++] & 0xFF;
+        final long byte92 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte91 << 5) | (byte92 >>> 3);
+        final long byte93 = blocks[blocksOffset++] & 0xFF;
+        final long byte94 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte92 & 7) << 10) | (byte93 << 2) | (byte94 >>> 6);
+        final long byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte94 & 63) << 7) | (byte95 >>> 1);
+        final long byte96 = blocks[blocksOffset++] & 0xFF;
+        final long byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 1) << 12) | (byte96 << 4) | (byte97 >>> 4);
+        final long byte98 = blocks[blocksOffset++] & 0xFF;
+        final long byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 15) << 9) | (byte98 << 1) | (byte99 >>> 7);
+        final long byte100 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 127) << 6) | (byte100 >>> 2);
+        final long byte101 = blocks[blocksOffset++] & 0xFF;
+        final long byte102 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte100 & 3) << 11) | (byte101 << 3) | (byte102 >>> 5);
+        final long byte103 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte102 & 31) << 8) | byte103;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 1);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 11) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 3);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 5);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 19) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 7);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 17) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 9);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 15) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 11);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset++] & 0xffffffffL) << 1) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 13) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 51) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 1);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 50) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 11) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 49) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 3);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 61) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 9) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 47) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 5);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 59) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 7) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 45) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 19) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 7);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 57) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 5) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 17) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 9);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 55) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 3) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 15) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 11);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 53) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 14) | (values[valuesOffset++] << 1) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 13) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked14.java

@@ -0,0 +1,352 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked14 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 7;
+    }
+
+    @Override
+    public int valueCount() {
+      return 32;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 50);
+        values[valuesOffset++] = (int) ((block0 >>> 36) & 16383L);
+        values[valuesOffset++] = (int) ((block0 >>> 22) & 16383L);
+        values[valuesOffset++] = (int) ((block0 >>> 8) & 16383L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 255L) << 6) | (block1 >>> 58));
+        values[valuesOffset++] = (int) ((block1 >>> 44) & 16383L);
+        values[valuesOffset++] = (int) ((block1 >>> 30) & 16383L);
+        values[valuesOffset++] = (int) ((block1 >>> 16) & 16383L);
+        values[valuesOffset++] = (int) ((block1 >>> 2) & 16383L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 3L) << 12) | (block2 >>> 52));
+        values[valuesOffset++] = (int) ((block2 >>> 38) & 16383L);
+        values[valuesOffset++] = (int) ((block2 >>> 24) & 16383L);
+        values[valuesOffset++] = (int) ((block2 >>> 10) & 16383L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 1023L) << 4) | (block3 >>> 60));
+        values[valuesOffset++] = (int) ((block3 >>> 46) & 16383L);
+        values[valuesOffset++] = (int) ((block3 >>> 32) & 16383L);
+        values[valuesOffset++] = (int) ((block3 >>> 18) & 16383L);
+        values[valuesOffset++] = (int) ((block3 >>> 4) & 16383L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 15L) << 10) | (block4 >>> 54));
+        values[valuesOffset++] = (int) ((block4 >>> 40) & 16383L);
+        values[valuesOffset++] = (int) ((block4 >>> 26) & 16383L);
+        values[valuesOffset++] = (int) ((block4 >>> 12) & 16383L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 4095L) << 2) | (block5 >>> 62));
+        values[valuesOffset++] = (int) ((block5 >>> 48) & 16383L);
+        values[valuesOffset++] = (int) ((block5 >>> 34) & 16383L);
+        values[valuesOffset++] = (int) ((block5 >>> 20) & 16383L);
+        values[valuesOffset++] = (int) ((block5 >>> 6) & 16383L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 63L) << 8) | (block6 >>> 56));
+        values[valuesOffset++] = (int) ((block6 >>> 42) & 16383L);
+        values[valuesOffset++] = (int) ((block6 >>> 28) & 16383L);
+        values[valuesOffset++] = (int) ((block6 >>> 14) & 16383L);
+        values[valuesOffset++] = (int) (block6 & 16383L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 6) | (byte1 >>> 2);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 3) << 12) | (byte2 << 4) | (byte3 >>> 4);
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 15) << 10) | (byte4 << 2) | (byte5 >>> 6);
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 63) << 8) | byte6;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte7 << 6) | (byte8 >>> 2);
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 3) << 12) | (byte9 << 4) | (byte10 >>> 4);
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 15) << 10) | (byte11 << 2) | (byte12 >>> 6);
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 63) << 8) | byte13;
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte14 << 6) | (byte15 >>> 2);
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 3) << 12) | (byte16 << 4) | (byte17 >>> 4);
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 15) << 10) | (byte18 << 2) | (byte19 >>> 6);
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 63) << 8) | byte20;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte21 << 6) | (byte22 >>> 2);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 3) << 12) | (byte23 << 4) | (byte24 >>> 4);
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 15) << 10) | (byte25 << 2) | (byte26 >>> 6);
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 8) | byte27;
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte28 << 6) | (byte29 >>> 2);
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 3) << 12) | (byte30 << 4) | (byte31 >>> 4);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 15) << 10) | (byte32 << 2) | (byte33 >>> 6);
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 63) << 8) | byte34;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte35 << 6) | (byte36 >>> 2);
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 3) << 12) | (byte37 << 4) | (byte38 >>> 4);
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 15) << 10) | (byte39 << 2) | (byte40 >>> 6);
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 63) << 8) | byte41;
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte42 << 6) | (byte43 >>> 2);
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 3) << 12) | (byte44 << 4) | (byte45 >>> 4);
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 15) << 10) | (byte46 << 2) | (byte47 >>> 6);
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 63) << 8) | byte48;
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte49 << 6) | (byte50 >>> 2);
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 3) << 12) | (byte51 << 4) | (byte52 >>> 4);
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 15) << 10) | (byte53 << 2) | (byte54 >>> 6);
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 63) << 8) | byte55;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 50;
+        values[valuesOffset++] = (block0 >>> 36) & 16383L;
+        values[valuesOffset++] = (block0 >>> 22) & 16383L;
+        values[valuesOffset++] = (block0 >>> 8) & 16383L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 255L) << 6) | (block1 >>> 58);
+        values[valuesOffset++] = (block1 >>> 44) & 16383L;
+        values[valuesOffset++] = (block1 >>> 30) & 16383L;
+        values[valuesOffset++] = (block1 >>> 16) & 16383L;
+        values[valuesOffset++] = (block1 >>> 2) & 16383L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 3L) << 12) | (block2 >>> 52);
+        values[valuesOffset++] = (block2 >>> 38) & 16383L;
+        values[valuesOffset++] = (block2 >>> 24) & 16383L;
+        values[valuesOffset++] = (block2 >>> 10) & 16383L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 1023L) << 4) | (block3 >>> 60);
+        values[valuesOffset++] = (block3 >>> 46) & 16383L;
+        values[valuesOffset++] = (block3 >>> 32) & 16383L;
+        values[valuesOffset++] = (block3 >>> 18) & 16383L;
+        values[valuesOffset++] = (block3 >>> 4) & 16383L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 15L) << 10) | (block4 >>> 54);
+        values[valuesOffset++] = (block4 >>> 40) & 16383L;
+        values[valuesOffset++] = (block4 >>> 26) & 16383L;
+        values[valuesOffset++] = (block4 >>> 12) & 16383L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 4095L) << 2) | (block5 >>> 62);
+        values[valuesOffset++] = (block5 >>> 48) & 16383L;
+        values[valuesOffset++] = (block5 >>> 34) & 16383L;
+        values[valuesOffset++] = (block5 >>> 20) & 16383L;
+        values[valuesOffset++] = (block5 >>> 6) & 16383L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 63L) << 8) | (block6 >>> 56);
+        values[valuesOffset++] = (block6 >>> 42) & 16383L;
+        values[valuesOffset++] = (block6 >>> 28) & 16383L;
+        values[valuesOffset++] = (block6 >>> 14) & 16383L;
+        values[valuesOffset++] = block6 & 16383L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 6) | (byte1 >>> 2);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 3) << 12) | (byte2 << 4) | (byte3 >>> 4);
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 15) << 10) | (byte4 << 2) | (byte5 >>> 6);
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 63) << 8) | byte6;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte7 << 6) | (byte8 >>> 2);
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 3) << 12) | (byte9 << 4) | (byte10 >>> 4);
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 15) << 10) | (byte11 << 2) | (byte12 >>> 6);
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 63) << 8) | byte13;
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte14 << 6) | (byte15 >>> 2);
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 3) << 12) | (byte16 << 4) | (byte17 >>> 4);
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte17 & 15) << 10) | (byte18 << 2) | (byte19 >>> 6);
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 63) << 8) | byte20;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte21 << 6) | (byte22 >>> 2);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 3) << 12) | (byte23 << 4) | (byte24 >>> 4);
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 15) << 10) | (byte25 << 2) | (byte26 >>> 6);
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 8) | byte27;
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte28 << 6) | (byte29 >>> 2);
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 3) << 12) | (byte30 << 4) | (byte31 >>> 4);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 15) << 10) | (byte32 << 2) | (byte33 >>> 6);
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 63) << 8) | byte34;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte35 << 6) | (byte36 >>> 2);
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 3) << 12) | (byte37 << 4) | (byte38 >>> 4);
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 15) << 10) | (byte39 << 2) | (byte40 >>> 6);
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 63) << 8) | byte41;
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte42 << 6) | (byte43 >>> 2);
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 3) << 12) | (byte44 << 4) | (byte45 >>> 4);
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 15) << 10) | (byte46 << 2) | (byte47 >>> 6);
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 63) << 8) | byte48;
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte49 << 6) | (byte50 >>> 2);
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 3) << 12) | (byte51 << 4) | (byte52 >>> 4);
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 15) << 10) | (byte53 << 2) | (byte54 >>> 6);
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 63) << 8) | byte55;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 14) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 50) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 14) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked15.java

@@ -0,0 +1,640 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked15 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 15;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 49);
+        values[valuesOffset++] = (int) ((block0 >>> 34) & 32767L);
+        values[valuesOffset++] = (int) ((block0 >>> 19) & 32767L);
+        values[valuesOffset++] = (int) ((block0 >>> 4) & 32767L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 15L) << 11) | (block1 >>> 53));
+        values[valuesOffset++] = (int) ((block1 >>> 38) & 32767L);
+        values[valuesOffset++] = (int) ((block1 >>> 23) & 32767L);
+        values[valuesOffset++] = (int) ((block1 >>> 8) & 32767L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 255L) << 7) | (block2 >>> 57));
+        values[valuesOffset++] = (int) ((block2 >>> 42) & 32767L);
+        values[valuesOffset++] = (int) ((block2 >>> 27) & 32767L);
+        values[valuesOffset++] = (int) ((block2 >>> 12) & 32767L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 4095L) << 3) | (block3 >>> 61));
+        values[valuesOffset++] = (int) ((block3 >>> 46) & 32767L);
+        values[valuesOffset++] = (int) ((block3 >>> 31) & 32767L);
+        values[valuesOffset++] = (int) ((block3 >>> 16) & 32767L);
+        values[valuesOffset++] = (int) ((block3 >>> 1) & 32767L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 1L) << 14) | (block4 >>> 50));
+        values[valuesOffset++] = (int) ((block4 >>> 35) & 32767L);
+        values[valuesOffset++] = (int) ((block4 >>> 20) & 32767L);
+        values[valuesOffset++] = (int) ((block4 >>> 5) & 32767L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 31L) << 10) | (block5 >>> 54));
+        values[valuesOffset++] = (int) ((block5 >>> 39) & 32767L);
+        values[valuesOffset++] = (int) ((block5 >>> 24) & 32767L);
+        values[valuesOffset++] = (int) ((block5 >>> 9) & 32767L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 511L) << 6) | (block6 >>> 58));
+        values[valuesOffset++] = (int) ((block6 >>> 43) & 32767L);
+        values[valuesOffset++] = (int) ((block6 >>> 28) & 32767L);
+        values[valuesOffset++] = (int) ((block6 >>> 13) & 32767L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 8191L) << 2) | (block7 >>> 62));
+        values[valuesOffset++] = (int) ((block7 >>> 47) & 32767L);
+        values[valuesOffset++] = (int) ((block7 >>> 32) & 32767L);
+        values[valuesOffset++] = (int) ((block7 >>> 17) & 32767L);
+        values[valuesOffset++] = (int) ((block7 >>> 2) & 32767L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 3L) << 13) | (block8 >>> 51));
+        values[valuesOffset++] = (int) ((block8 >>> 36) & 32767L);
+        values[valuesOffset++] = (int) ((block8 >>> 21) & 32767L);
+        values[valuesOffset++] = (int) ((block8 >>> 6) & 32767L);
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block8 & 63L) << 9) | (block9 >>> 55));
+        values[valuesOffset++] = (int) ((block9 >>> 40) & 32767L);
+        values[valuesOffset++] = (int) ((block9 >>> 25) & 32767L);
+        values[valuesOffset++] = (int) ((block9 >>> 10) & 32767L);
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block9 & 1023L) << 5) | (block10 >>> 59));
+        values[valuesOffset++] = (int) ((block10 >>> 44) & 32767L);
+        values[valuesOffset++] = (int) ((block10 >>> 29) & 32767L);
+        values[valuesOffset++] = (int) ((block10 >>> 14) & 32767L);
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block10 & 16383L) << 1) | (block11 >>> 63));
+        values[valuesOffset++] = (int) ((block11 >>> 48) & 32767L);
+        values[valuesOffset++] = (int) ((block11 >>> 33) & 32767L);
+        values[valuesOffset++] = (int) ((block11 >>> 18) & 32767L);
+        values[valuesOffset++] = (int) ((block11 >>> 3) & 32767L);
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block11 & 7L) << 12) | (block12 >>> 52));
+        values[valuesOffset++] = (int) ((block12 >>> 37) & 32767L);
+        values[valuesOffset++] = (int) ((block12 >>> 22) & 32767L);
+        values[valuesOffset++] = (int) ((block12 >>> 7) & 32767L);
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block12 & 127L) << 8) | (block13 >>> 56));
+        values[valuesOffset++] = (int) ((block13 >>> 41) & 32767L);
+        values[valuesOffset++] = (int) ((block13 >>> 26) & 32767L);
+        values[valuesOffset++] = (int) ((block13 >>> 11) & 32767L);
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block13 & 2047L) << 4) | (block14 >>> 60));
+        values[valuesOffset++] = (int) ((block14 >>> 45) & 32767L);
+        values[valuesOffset++] = (int) ((block14 >>> 30) & 32767L);
+        values[valuesOffset++] = (int) ((block14 >>> 15) & 32767L);
+        values[valuesOffset++] = (int) (block14 & 32767L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 7) | (byte1 >>> 1);
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 1) << 14) | (byte2 << 6) | (byte3 >>> 2);
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 3) << 13) | (byte4 << 5) | (byte5 >>> 3);
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 7) << 12) | (byte6 << 4) | (byte7 >>> 4);
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 11) | (byte8 << 3) | (byte9 >>> 5);
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 31) << 10) | (byte10 << 2) | (byte11 >>> 6);
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 9) | (byte12 << 1) | (byte13 >>> 7);
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 127) << 8) | byte14;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 7) | (byte16 >>> 1);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 1) << 14) | (byte17 << 6) | (byte18 >>> 2);
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte18 & 3) << 13) | (byte19 << 5) | (byte20 >>> 3);
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 7) << 12) | (byte21 << 4) | (byte22 >>> 4);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 11) | (byte23 << 3) | (byte24 >>> 5);
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 31) << 10) | (byte25 << 2) | (byte26 >>> 6);
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 9) | (byte27 << 1) | (byte28 >>> 7);
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 127) << 8) | byte29;
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte30 << 7) | (byte31 >>> 1);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 1) << 14) | (byte32 << 6) | (byte33 >>> 2);
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 13) | (byte34 << 5) | (byte35 >>> 3);
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 7) << 12) | (byte36 << 4) | (byte37 >>> 4);
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 15) << 11) | (byte38 << 3) | (byte39 >>> 5);
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte39 & 31) << 10) | (byte40 << 2) | (byte41 >>> 6);
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte41 & 63) << 9) | (byte42 << 1) | (byte43 >>> 7);
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 127) << 8) | byte44;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte45 << 7) | (byte46 >>> 1);
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 1) << 14) | (byte47 << 6) | (byte48 >>> 2);
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 3) << 13) | (byte49 << 5) | (byte50 >>> 3);
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 7) << 12) | (byte51 << 4) | (byte52 >>> 4);
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 15) << 11) | (byte53 << 3) | (byte54 >>> 5);
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 31) << 10) | (byte55 << 2) | (byte56 >>> 6);
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 63) << 9) | (byte57 << 1) | (byte58 >>> 7);
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 127) << 8) | byte59;
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte60 << 7) | (byte61 >>> 1);
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 1) << 14) | (byte62 << 6) | (byte63 >>> 2);
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 3) << 13) | (byte64 << 5) | (byte65 >>> 3);
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 7) << 12) | (byte66 << 4) | (byte67 >>> 4);
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 15) << 11) | (byte68 << 3) | (byte69 >>> 5);
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 31) << 10) | (byte70 << 2) | (byte71 >>> 6);
+        final int byte72 = blocks[blocksOffset++] & 0xFF;
+        final int byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 63) << 9) | (byte72 << 1) | (byte73 >>> 7);
+        final int byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 127) << 8) | byte74;
+        final int byte75 = blocks[blocksOffset++] & 0xFF;
+        final int byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte75 << 7) | (byte76 >>> 1);
+        final int byte77 = blocks[blocksOffset++] & 0xFF;
+        final int byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 1) << 14) | (byte77 << 6) | (byte78 >>> 2);
+        final int byte79 = blocks[blocksOffset++] & 0xFF;
+        final int byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 3) << 13) | (byte79 << 5) | (byte80 >>> 3);
+        final int byte81 = blocks[blocksOffset++] & 0xFF;
+        final int byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 7) << 12) | (byte81 << 4) | (byte82 >>> 4);
+        final int byte83 = blocks[blocksOffset++] & 0xFF;
+        final int byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 15) << 11) | (byte83 << 3) | (byte84 >>> 5);
+        final int byte85 = blocks[blocksOffset++] & 0xFF;
+        final int byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte84 & 31) << 10) | (byte85 << 2) | (byte86 >>> 6);
+        final int byte87 = blocks[blocksOffset++] & 0xFF;
+        final int byte88 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 63) << 9) | (byte87 << 1) | (byte88 >>> 7);
+        final int byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte88 & 127) << 8) | byte89;
+        final int byte90 = blocks[blocksOffset++] & 0xFF;
+        final int byte91 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte90 << 7) | (byte91 >>> 1);
+        final int byte92 = blocks[blocksOffset++] & 0xFF;
+        final int byte93 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte91 & 1) << 14) | (byte92 << 6) | (byte93 >>> 2);
+        final int byte94 = blocks[blocksOffset++] & 0xFF;
+        final int byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte93 & 3) << 13) | (byte94 << 5) | (byte95 >>> 3);
+        final int byte96 = blocks[blocksOffset++] & 0xFF;
+        final int byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 7) << 12) | (byte96 << 4) | (byte97 >>> 4);
+        final int byte98 = blocks[blocksOffset++] & 0xFF;
+        final int byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 15) << 11) | (byte98 << 3) | (byte99 >>> 5);
+        final int byte100 = blocks[blocksOffset++] & 0xFF;
+        final int byte101 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 31) << 10) | (byte100 << 2) | (byte101 >>> 6);
+        final int byte102 = blocks[blocksOffset++] & 0xFF;
+        final int byte103 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte101 & 63) << 9) | (byte102 << 1) | (byte103 >>> 7);
+        final int byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte103 & 127) << 8) | byte104;
+        final int byte105 = blocks[blocksOffset++] & 0xFF;
+        final int byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte105 << 7) | (byte106 >>> 1);
+        final int byte107 = blocks[blocksOffset++] & 0xFF;
+        final int byte108 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 1) << 14) | (byte107 << 6) | (byte108 >>> 2);
+        final int byte109 = blocks[blocksOffset++] & 0xFF;
+        final int byte110 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte108 & 3) << 13) | (byte109 << 5) | (byte110 >>> 3);
+        final int byte111 = blocks[blocksOffset++] & 0xFF;
+        final int byte112 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte110 & 7) << 12) | (byte111 << 4) | (byte112 >>> 4);
+        final int byte113 = blocks[blocksOffset++] & 0xFF;
+        final int byte114 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte112 & 15) << 11) | (byte113 << 3) | (byte114 >>> 5);
+        final int byte115 = blocks[blocksOffset++] & 0xFF;
+        final int byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte114 & 31) << 10) | (byte115 << 2) | (byte116 >>> 6);
+        final int byte117 = blocks[blocksOffset++] & 0xFF;
+        final int byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 63) << 9) | (byte117 << 1) | (byte118 >>> 7);
+        final int byte119 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte118 & 127) << 8) | byte119;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 49;
+        values[valuesOffset++] = (block0 >>> 34) & 32767L;
+        values[valuesOffset++] = (block0 >>> 19) & 32767L;
+        values[valuesOffset++] = (block0 >>> 4) & 32767L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 15L) << 11) | (block1 >>> 53);
+        values[valuesOffset++] = (block1 >>> 38) & 32767L;
+        values[valuesOffset++] = (block1 >>> 23) & 32767L;
+        values[valuesOffset++] = (block1 >>> 8) & 32767L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 255L) << 7) | (block2 >>> 57);
+        values[valuesOffset++] = (block2 >>> 42) & 32767L;
+        values[valuesOffset++] = (block2 >>> 27) & 32767L;
+        values[valuesOffset++] = (block2 >>> 12) & 32767L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 4095L) << 3) | (block3 >>> 61);
+        values[valuesOffset++] = (block3 >>> 46) & 32767L;
+        values[valuesOffset++] = (block3 >>> 31) & 32767L;
+        values[valuesOffset++] = (block3 >>> 16) & 32767L;
+        values[valuesOffset++] = (block3 >>> 1) & 32767L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 1L) << 14) | (block4 >>> 50);
+        values[valuesOffset++] = (block4 >>> 35) & 32767L;
+        values[valuesOffset++] = (block4 >>> 20) & 32767L;
+        values[valuesOffset++] = (block4 >>> 5) & 32767L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 31L) << 10) | (block5 >>> 54);
+        values[valuesOffset++] = (block5 >>> 39) & 32767L;
+        values[valuesOffset++] = (block5 >>> 24) & 32767L;
+        values[valuesOffset++] = (block5 >>> 9) & 32767L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 511L) << 6) | (block6 >>> 58);
+        values[valuesOffset++] = (block6 >>> 43) & 32767L;
+        values[valuesOffset++] = (block6 >>> 28) & 32767L;
+        values[valuesOffset++] = (block6 >>> 13) & 32767L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 8191L) << 2) | (block7 >>> 62);
+        values[valuesOffset++] = (block7 >>> 47) & 32767L;
+        values[valuesOffset++] = (block7 >>> 32) & 32767L;
+        values[valuesOffset++] = (block7 >>> 17) & 32767L;
+        values[valuesOffset++] = (block7 >>> 2) & 32767L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 3L) << 13) | (block8 >>> 51);
+        values[valuesOffset++] = (block8 >>> 36) & 32767L;
+        values[valuesOffset++] = (block8 >>> 21) & 32767L;
+        values[valuesOffset++] = (block8 >>> 6) & 32767L;
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block8 & 63L) << 9) | (block9 >>> 55);
+        values[valuesOffset++] = (block9 >>> 40) & 32767L;
+        values[valuesOffset++] = (block9 >>> 25) & 32767L;
+        values[valuesOffset++] = (block9 >>> 10) & 32767L;
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block9 & 1023L) << 5) | (block10 >>> 59);
+        values[valuesOffset++] = (block10 >>> 44) & 32767L;
+        values[valuesOffset++] = (block10 >>> 29) & 32767L;
+        values[valuesOffset++] = (block10 >>> 14) & 32767L;
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block10 & 16383L) << 1) | (block11 >>> 63);
+        values[valuesOffset++] = (block11 >>> 48) & 32767L;
+        values[valuesOffset++] = (block11 >>> 33) & 32767L;
+        values[valuesOffset++] = (block11 >>> 18) & 32767L;
+        values[valuesOffset++] = (block11 >>> 3) & 32767L;
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block11 & 7L) << 12) | (block12 >>> 52);
+        values[valuesOffset++] = (block12 >>> 37) & 32767L;
+        values[valuesOffset++] = (block12 >>> 22) & 32767L;
+        values[valuesOffset++] = (block12 >>> 7) & 32767L;
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block12 & 127L) << 8) | (block13 >>> 56);
+        values[valuesOffset++] = (block13 >>> 41) & 32767L;
+        values[valuesOffset++] = (block13 >>> 26) & 32767L;
+        values[valuesOffset++] = (block13 >>> 11) & 32767L;
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block13 & 2047L) << 4) | (block14 >>> 60);
+        values[valuesOffset++] = (block14 >>> 45) & 32767L;
+        values[valuesOffset++] = (block14 >>> 30) & 32767L;
+        values[valuesOffset++] = (block14 >>> 15) & 32767L;
+        values[valuesOffset++] = block14 & 32767L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 7) | (byte1 >>> 1);
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte1 & 1) << 14) | (byte2 << 6) | (byte3 >>> 2);
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte3 & 3) << 13) | (byte4 << 5) | (byte5 >>> 3);
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte5 & 7) << 12) | (byte6 << 4) | (byte7 >>> 4);
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 15) << 11) | (byte8 << 3) | (byte9 >>> 5);
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 31) << 10) | (byte10 << 2) | (byte11 >>> 6);
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 9) | (byte12 << 1) | (byte13 >>> 7);
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 127) << 8) | byte14;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte15 << 7) | (byte16 >>> 1);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 1) << 14) | (byte17 << 6) | (byte18 >>> 2);
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte18 & 3) << 13) | (byte19 << 5) | (byte20 >>> 3);
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 7) << 12) | (byte21 << 4) | (byte22 >>> 4);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 11) | (byte23 << 3) | (byte24 >>> 5);
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 31) << 10) | (byte25 << 2) | (byte26 >>> 6);
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 63) << 9) | (byte27 << 1) | (byte28 >>> 7);
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 127) << 8) | byte29;
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte30 << 7) | (byte31 >>> 1);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 1) << 14) | (byte32 << 6) | (byte33 >>> 2);
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 13) | (byte34 << 5) | (byte35 >>> 3);
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 7) << 12) | (byte36 << 4) | (byte37 >>> 4);
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte37 & 15) << 11) | (byte38 << 3) | (byte39 >>> 5);
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte39 & 31) << 10) | (byte40 << 2) | (byte41 >>> 6);
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte41 & 63) << 9) | (byte42 << 1) | (byte43 >>> 7);
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte43 & 127) << 8) | byte44;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte45 << 7) | (byte46 >>> 1);
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 1) << 14) | (byte47 << 6) | (byte48 >>> 2);
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 3) << 13) | (byte49 << 5) | (byte50 >>> 3);
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte50 & 7) << 12) | (byte51 << 4) | (byte52 >>> 4);
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 15) << 11) | (byte53 << 3) | (byte54 >>> 5);
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 31) << 10) | (byte55 << 2) | (byte56 >>> 6);
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 63) << 9) | (byte57 << 1) | (byte58 >>> 7);
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 127) << 8) | byte59;
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte60 << 7) | (byte61 >>> 1);
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 1) << 14) | (byte62 << 6) | (byte63 >>> 2);
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 3) << 13) | (byte64 << 5) | (byte65 >>> 3);
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 7) << 12) | (byte66 << 4) | (byte67 >>> 4);
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 15) << 11) | (byte68 << 3) | (byte69 >>> 5);
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 31) << 10) | (byte70 << 2) | (byte71 >>> 6);
+        final long byte72 = blocks[blocksOffset++] & 0xFF;
+        final long byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 63) << 9) | (byte72 << 1) | (byte73 >>> 7);
+        final long byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 127) << 8) | byte74;
+        final long byte75 = blocks[blocksOffset++] & 0xFF;
+        final long byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte75 << 7) | (byte76 >>> 1);
+        final long byte77 = blocks[blocksOffset++] & 0xFF;
+        final long byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 1) << 14) | (byte77 << 6) | (byte78 >>> 2);
+        final long byte79 = blocks[blocksOffset++] & 0xFF;
+        final long byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 3) << 13) | (byte79 << 5) | (byte80 >>> 3);
+        final long byte81 = blocks[blocksOffset++] & 0xFF;
+        final long byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 7) << 12) | (byte81 << 4) | (byte82 >>> 4);
+        final long byte83 = blocks[blocksOffset++] & 0xFF;
+        final long byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 15) << 11) | (byte83 << 3) | (byte84 >>> 5);
+        final long byte85 = blocks[blocksOffset++] & 0xFF;
+        final long byte86 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte84 & 31) << 10) | (byte85 << 2) | (byte86 >>> 6);
+        final long byte87 = blocks[blocksOffset++] & 0xFF;
+        final long byte88 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte86 & 63) << 9) | (byte87 << 1) | (byte88 >>> 7);
+        final long byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte88 & 127) << 8) | byte89;
+        final long byte90 = blocks[blocksOffset++] & 0xFF;
+        final long byte91 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte90 << 7) | (byte91 >>> 1);
+        final long byte92 = blocks[blocksOffset++] & 0xFF;
+        final long byte93 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte91 & 1) << 14) | (byte92 << 6) | (byte93 >>> 2);
+        final long byte94 = blocks[blocksOffset++] & 0xFF;
+        final long byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte93 & 3) << 13) | (byte94 << 5) | (byte95 >>> 3);
+        final long byte96 = blocks[blocksOffset++] & 0xFF;
+        final long byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 7) << 12) | (byte96 << 4) | (byte97 >>> 4);
+        final long byte98 = blocks[blocksOffset++] & 0xFF;
+        final long byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 15) << 11) | (byte98 << 3) | (byte99 >>> 5);
+        final long byte100 = blocks[blocksOffset++] & 0xFF;
+        final long byte101 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 31) << 10) | (byte100 << 2) | (byte101 >>> 6);
+        final long byte102 = blocks[blocksOffset++] & 0xFF;
+        final long byte103 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte101 & 63) << 9) | (byte102 << 1) | (byte103 >>> 7);
+        final long byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte103 & 127) << 8) | byte104;
+        final long byte105 = blocks[blocksOffset++] & 0xFF;
+        final long byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte105 << 7) | (byte106 >>> 1);
+        final long byte107 = blocks[blocksOffset++] & 0xFF;
+        final long byte108 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 1) << 14) | (byte107 << 6) | (byte108 >>> 2);
+        final long byte109 = blocks[blocksOffset++] & 0xFF;
+        final long byte110 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte108 & 3) << 13) | (byte109 << 5) | (byte110 >>> 3);
+        final long byte111 = blocks[blocksOffset++] & 0xFF;
+        final long byte112 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte110 & 7) << 12) | (byte111 << 4) | (byte112 >>> 4);
+        final long byte113 = blocks[blocksOffset++] & 0xFF;
+        final long byte114 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte112 & 15) << 11) | (byte113 << 3) | (byte114 >>> 5);
+        final long byte115 = blocks[blocksOffset++] & 0xFF;
+        final long byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte114 & 31) << 10) | (byte115 << 2) | (byte116 >>> 6);
+        final long byte117 = blocks[blocksOffset++] & 0xFF;
+        final long byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 63) << 9) | (byte117 << 1) | (byte118 >>> 7);
+        final long byte119 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte118 & 127) << 8) | byte119;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 19) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 11);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 7);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 3);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 1) | ((values[valuesOffset] & 0xffffffffL) >>> 14);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 13) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 17) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 13);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 9);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 5);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset] & 0xffffffffL) >>> 1);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 11) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 15) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 49) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 19) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 11);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 53) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 7);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 57) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 3);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 61) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 1) | (values[valuesOffset] >>> 14);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 50) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 5) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 9) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 13) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 47) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 17) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 13);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 51) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 9);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 55) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 5);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 59) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 14) | (values[valuesOffset] >>> 1);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 3) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 7) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 11) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 45) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 15) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked16.java

@@ -0,0 +1,120 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked16 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 1;
+    }
+
+    @Override
+    public int valueCount() {
+      return 4;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 48);
+        values[valuesOffset++] = (int) ((block0 >>> 32) & 65535L);
+        values[valuesOffset++] = (int) ((block0 >>> 16) & 65535L);
+        values[valuesOffset++] = (int) (block0 & 65535L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 8) | byte1;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte2 << 8) | byte3;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte4 << 8) | byte5;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte6 << 8) | byte7;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 48;
+        values[valuesOffset++] = (block0 >>> 32) & 65535L;
+        values[valuesOffset++] = (block0 >>> 16) & 65535L;
+        values[valuesOffset++] = block0 & 65535L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 8) | byte1;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte2 << 8) | byte3;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte4 << 8) | byte5;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte6 << 8) | byte7;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 16) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 48) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 16) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked17.java

@@ -0,0 +1,680 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked17 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 17;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 47);
+        values[valuesOffset++] = (int) ((block0 >>> 30) & 131071L);
+        values[valuesOffset++] = (int) ((block0 >>> 13) & 131071L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 8191L) << 4) | (block1 >>> 60));
+        values[valuesOffset++] = (int) ((block1 >>> 43) & 131071L);
+        values[valuesOffset++] = (int) ((block1 >>> 26) & 131071L);
+        values[valuesOffset++] = (int) ((block1 >>> 9) & 131071L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 511L) << 8) | (block2 >>> 56));
+        values[valuesOffset++] = (int) ((block2 >>> 39) & 131071L);
+        values[valuesOffset++] = (int) ((block2 >>> 22) & 131071L);
+        values[valuesOffset++] = (int) ((block2 >>> 5) & 131071L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 31L) << 12) | (block3 >>> 52));
+        values[valuesOffset++] = (int) ((block3 >>> 35) & 131071L);
+        values[valuesOffset++] = (int) ((block3 >>> 18) & 131071L);
+        values[valuesOffset++] = (int) ((block3 >>> 1) & 131071L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 1L) << 16) | (block4 >>> 48));
+        values[valuesOffset++] = (int) ((block4 >>> 31) & 131071L);
+        values[valuesOffset++] = (int) ((block4 >>> 14) & 131071L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 16383L) << 3) | (block5 >>> 61));
+        values[valuesOffset++] = (int) ((block5 >>> 44) & 131071L);
+        values[valuesOffset++] = (int) ((block5 >>> 27) & 131071L);
+        values[valuesOffset++] = (int) ((block5 >>> 10) & 131071L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 1023L) << 7) | (block6 >>> 57));
+        values[valuesOffset++] = (int) ((block6 >>> 40) & 131071L);
+        values[valuesOffset++] = (int) ((block6 >>> 23) & 131071L);
+        values[valuesOffset++] = (int) ((block6 >>> 6) & 131071L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 63L) << 11) | (block7 >>> 53));
+        values[valuesOffset++] = (int) ((block7 >>> 36) & 131071L);
+        values[valuesOffset++] = (int) ((block7 >>> 19) & 131071L);
+        values[valuesOffset++] = (int) ((block7 >>> 2) & 131071L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 3L) << 15) | (block8 >>> 49));
+        values[valuesOffset++] = (int) ((block8 >>> 32) & 131071L);
+        values[valuesOffset++] = (int) ((block8 >>> 15) & 131071L);
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block8 & 32767L) << 2) | (block9 >>> 62));
+        values[valuesOffset++] = (int) ((block9 >>> 45) & 131071L);
+        values[valuesOffset++] = (int) ((block9 >>> 28) & 131071L);
+        values[valuesOffset++] = (int) ((block9 >>> 11) & 131071L);
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block9 & 2047L) << 6) | (block10 >>> 58));
+        values[valuesOffset++] = (int) ((block10 >>> 41) & 131071L);
+        values[valuesOffset++] = (int) ((block10 >>> 24) & 131071L);
+        values[valuesOffset++] = (int) ((block10 >>> 7) & 131071L);
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block10 & 127L) << 10) | (block11 >>> 54));
+        values[valuesOffset++] = (int) ((block11 >>> 37) & 131071L);
+        values[valuesOffset++] = (int) ((block11 >>> 20) & 131071L);
+        values[valuesOffset++] = (int) ((block11 >>> 3) & 131071L);
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block11 & 7L) << 14) | (block12 >>> 50));
+        values[valuesOffset++] = (int) ((block12 >>> 33) & 131071L);
+        values[valuesOffset++] = (int) ((block12 >>> 16) & 131071L);
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block12 & 65535L) << 1) | (block13 >>> 63));
+        values[valuesOffset++] = (int) ((block13 >>> 46) & 131071L);
+        values[valuesOffset++] = (int) ((block13 >>> 29) & 131071L);
+        values[valuesOffset++] = (int) ((block13 >>> 12) & 131071L);
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block13 & 4095L) << 5) | (block14 >>> 59));
+        values[valuesOffset++] = (int) ((block14 >>> 42) & 131071L);
+        values[valuesOffset++] = (int) ((block14 >>> 25) & 131071L);
+        values[valuesOffset++] = (int) ((block14 >>> 8) & 131071L);
+        final long block15 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block14 & 255L) << 9) | (block15 >>> 55));
+        values[valuesOffset++] = (int) ((block15 >>> 38) & 131071L);
+        values[valuesOffset++] = (int) ((block15 >>> 21) & 131071L);
+        values[valuesOffset++] = (int) ((block15 >>> 4) & 131071L);
+        final long block16 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block15 & 15L) << 13) | (block16 >>> 51));
+        values[valuesOffset++] = (int) ((block16 >>> 34) & 131071L);
+        values[valuesOffset++] = (int) ((block16 >>> 17) & 131071L);
+        values[valuesOffset++] = (int) (block16 & 131071L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 9) | (byte1 << 1) | (byte2 >>> 7);
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 127) << 10) | (byte3 << 2) | (byte4 >>> 6);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 63) << 11) | (byte5 << 3) | (byte6 >>> 5);
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 31) << 12) | (byte7 << 4) | (byte8 >>> 4);
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 15) << 13) | (byte9 << 5) | (byte10 >>> 3);
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 7) << 14) | (byte11 << 6) | (byte12 >>> 2);
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 3) << 15) | (byte13 << 7) | (byte14 >>> 1);
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 1) << 16) | (byte15 << 8) | byte16;
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte17 << 9) | (byte18 << 1) | (byte19 >>> 7);
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 127) << 10) | (byte20 << 2) | (byte21 >>> 6);
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 63) << 11) | (byte22 << 3) | (byte23 >>> 5);
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 31) << 12) | (byte24 << 4) | (byte25 >>> 4);
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte25 & 15) << 13) | (byte26 << 5) | (byte27 >>> 3);
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 7) << 14) | (byte28 << 6) | (byte29 >>> 2);
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 3) << 15) | (byte30 << 7) | (byte31 >>> 1);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 1) << 16) | (byte32 << 8) | byte33;
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte34 << 9) | (byte35 << 1) | (byte36 >>> 7);
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 127) << 10) | (byte37 << 2) | (byte38 >>> 6);
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 63) << 11) | (byte39 << 3) | (byte40 >>> 5);
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 31) << 12) | (byte41 << 4) | (byte42 >>> 4);
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 15) << 13) | (byte43 << 5) | (byte44 >>> 3);
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte44 & 7) << 14) | (byte45 << 6) | (byte46 >>> 2);
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 3) << 15) | (byte47 << 7) | (byte48 >>> 1);
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 1) << 16) | (byte49 << 8) | byte50;
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte51 << 9) | (byte52 << 1) | (byte53 >>> 7);
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 127) << 10) | (byte54 << 2) | (byte55 >>> 6);
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte55 & 63) << 11) | (byte56 << 3) | (byte57 >>> 5);
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte57 & 31) << 12) | (byte58 << 4) | (byte59 >>> 4);
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 15) << 13) | (byte60 << 5) | (byte61 >>> 3);
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 7) << 14) | (byte62 << 6) | (byte63 >>> 2);
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 3) << 15) | (byte64 << 7) | (byte65 >>> 1);
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 1) << 16) | (byte66 << 8) | byte67;
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte68 << 9) | (byte69 << 1) | (byte70 >>> 7);
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        final int byte72 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte70 & 127) << 10) | (byte71 << 2) | (byte72 >>> 6);
+        final int byte73 = blocks[blocksOffset++] & 0xFF;
+        final int byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte72 & 63) << 11) | (byte73 << 3) | (byte74 >>> 5);
+        final int byte75 = blocks[blocksOffset++] & 0xFF;
+        final int byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 31) << 12) | (byte75 << 4) | (byte76 >>> 4);
+        final int byte77 = blocks[blocksOffset++] & 0xFF;
+        final int byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 15) << 13) | (byte77 << 5) | (byte78 >>> 3);
+        final int byte79 = blocks[blocksOffset++] & 0xFF;
+        final int byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 7) << 14) | (byte79 << 6) | (byte80 >>> 2);
+        final int byte81 = blocks[blocksOffset++] & 0xFF;
+        final int byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 3) << 15) | (byte81 << 7) | (byte82 >>> 1);
+        final int byte83 = blocks[blocksOffset++] & 0xFF;
+        final int byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 1) << 16) | (byte83 << 8) | byte84;
+        final int byte85 = blocks[blocksOffset++] & 0xFF;
+        final int byte86 = blocks[blocksOffset++] & 0xFF;
+        final int byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte85 << 9) | (byte86 << 1) | (byte87 >>> 7);
+        final int byte88 = blocks[blocksOffset++] & 0xFF;
+        final int byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 127) << 10) | (byte88 << 2) | (byte89 >>> 6);
+        final int byte90 = blocks[blocksOffset++] & 0xFF;
+        final int byte91 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte89 & 63) << 11) | (byte90 << 3) | (byte91 >>> 5);
+        final int byte92 = blocks[blocksOffset++] & 0xFF;
+        final int byte93 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte91 & 31) << 12) | (byte92 << 4) | (byte93 >>> 4);
+        final int byte94 = blocks[blocksOffset++] & 0xFF;
+        final int byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte93 & 15) << 13) | (byte94 << 5) | (byte95 >>> 3);
+        final int byte96 = blocks[blocksOffset++] & 0xFF;
+        final int byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 7) << 14) | (byte96 << 6) | (byte97 >>> 2);
+        final int byte98 = blocks[blocksOffset++] & 0xFF;
+        final int byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 3) << 15) | (byte98 << 7) | (byte99 >>> 1);
+        final int byte100 = blocks[blocksOffset++] & 0xFF;
+        final int byte101 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 1) << 16) | (byte100 << 8) | byte101;
+        final int byte102 = blocks[blocksOffset++] & 0xFF;
+        final int byte103 = blocks[blocksOffset++] & 0xFF;
+        final int byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte102 << 9) | (byte103 << 1) | (byte104 >>> 7);
+        final int byte105 = blocks[blocksOffset++] & 0xFF;
+        final int byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte104 & 127) << 10) | (byte105 << 2) | (byte106 >>> 6);
+        final int byte107 = blocks[blocksOffset++] & 0xFF;
+        final int byte108 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 63) << 11) | (byte107 << 3) | (byte108 >>> 5);
+        final int byte109 = blocks[blocksOffset++] & 0xFF;
+        final int byte110 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte108 & 31) << 12) | (byte109 << 4) | (byte110 >>> 4);
+        final int byte111 = blocks[blocksOffset++] & 0xFF;
+        final int byte112 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte110 & 15) << 13) | (byte111 << 5) | (byte112 >>> 3);
+        final int byte113 = blocks[blocksOffset++] & 0xFF;
+        final int byte114 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte112 & 7) << 14) | (byte113 << 6) | (byte114 >>> 2);
+        final int byte115 = blocks[blocksOffset++] & 0xFF;
+        final int byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte114 & 3) << 15) | (byte115 << 7) | (byte116 >>> 1);
+        final int byte117 = blocks[blocksOffset++] & 0xFF;
+        final int byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 1) << 16) | (byte117 << 8) | byte118;
+        final int byte119 = blocks[blocksOffset++] & 0xFF;
+        final int byte120 = blocks[blocksOffset++] & 0xFF;
+        final int byte121 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte119 << 9) | (byte120 << 1) | (byte121 >>> 7);
+        final int byte122 = blocks[blocksOffset++] & 0xFF;
+        final int byte123 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte121 & 127) << 10) | (byte122 << 2) | (byte123 >>> 6);
+        final int byte124 = blocks[blocksOffset++] & 0xFF;
+        final int byte125 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte123 & 63) << 11) | (byte124 << 3) | (byte125 >>> 5);
+        final int byte126 = blocks[blocksOffset++] & 0xFF;
+        final int byte127 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte125 & 31) << 12) | (byte126 << 4) | (byte127 >>> 4);
+        final int byte128 = blocks[blocksOffset++] & 0xFF;
+        final int byte129 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte127 & 15) << 13) | (byte128 << 5) | (byte129 >>> 3);
+        final int byte130 = blocks[blocksOffset++] & 0xFF;
+        final int byte131 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte129 & 7) << 14) | (byte130 << 6) | (byte131 >>> 2);
+        final int byte132 = blocks[blocksOffset++] & 0xFF;
+        final int byte133 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte131 & 3) << 15) | (byte132 << 7) | (byte133 >>> 1);
+        final int byte134 = blocks[blocksOffset++] & 0xFF;
+        final int byte135 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte133 & 1) << 16) | (byte134 << 8) | byte135;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 47;
+        values[valuesOffset++] = (block0 >>> 30) & 131071L;
+        values[valuesOffset++] = (block0 >>> 13) & 131071L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 8191L) << 4) | (block1 >>> 60);
+        values[valuesOffset++] = (block1 >>> 43) & 131071L;
+        values[valuesOffset++] = (block1 >>> 26) & 131071L;
+        values[valuesOffset++] = (block1 >>> 9) & 131071L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 511L) << 8) | (block2 >>> 56);
+        values[valuesOffset++] = (block2 >>> 39) & 131071L;
+        values[valuesOffset++] = (block2 >>> 22) & 131071L;
+        values[valuesOffset++] = (block2 >>> 5) & 131071L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 31L) << 12) | (block3 >>> 52);
+        values[valuesOffset++] = (block3 >>> 35) & 131071L;
+        values[valuesOffset++] = (block3 >>> 18) & 131071L;
+        values[valuesOffset++] = (block3 >>> 1) & 131071L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 1L) << 16) | (block4 >>> 48);
+        values[valuesOffset++] = (block4 >>> 31) & 131071L;
+        values[valuesOffset++] = (block4 >>> 14) & 131071L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 16383L) << 3) | (block5 >>> 61);
+        values[valuesOffset++] = (block5 >>> 44) & 131071L;
+        values[valuesOffset++] = (block5 >>> 27) & 131071L;
+        values[valuesOffset++] = (block5 >>> 10) & 131071L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 1023L) << 7) | (block6 >>> 57);
+        values[valuesOffset++] = (block6 >>> 40) & 131071L;
+        values[valuesOffset++] = (block6 >>> 23) & 131071L;
+        values[valuesOffset++] = (block6 >>> 6) & 131071L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 63L) << 11) | (block7 >>> 53);
+        values[valuesOffset++] = (block7 >>> 36) & 131071L;
+        values[valuesOffset++] = (block7 >>> 19) & 131071L;
+        values[valuesOffset++] = (block7 >>> 2) & 131071L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 3L) << 15) | (block8 >>> 49);
+        values[valuesOffset++] = (block8 >>> 32) & 131071L;
+        values[valuesOffset++] = (block8 >>> 15) & 131071L;
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block8 & 32767L) << 2) | (block9 >>> 62);
+        values[valuesOffset++] = (block9 >>> 45) & 131071L;
+        values[valuesOffset++] = (block9 >>> 28) & 131071L;
+        values[valuesOffset++] = (block9 >>> 11) & 131071L;
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block9 & 2047L) << 6) | (block10 >>> 58);
+        values[valuesOffset++] = (block10 >>> 41) & 131071L;
+        values[valuesOffset++] = (block10 >>> 24) & 131071L;
+        values[valuesOffset++] = (block10 >>> 7) & 131071L;
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block10 & 127L) << 10) | (block11 >>> 54);
+        values[valuesOffset++] = (block11 >>> 37) & 131071L;
+        values[valuesOffset++] = (block11 >>> 20) & 131071L;
+        values[valuesOffset++] = (block11 >>> 3) & 131071L;
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block11 & 7L) << 14) | (block12 >>> 50);
+        values[valuesOffset++] = (block12 >>> 33) & 131071L;
+        values[valuesOffset++] = (block12 >>> 16) & 131071L;
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block12 & 65535L) << 1) | (block13 >>> 63);
+        values[valuesOffset++] = (block13 >>> 46) & 131071L;
+        values[valuesOffset++] = (block13 >>> 29) & 131071L;
+        values[valuesOffset++] = (block13 >>> 12) & 131071L;
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block13 & 4095L) << 5) | (block14 >>> 59);
+        values[valuesOffset++] = (block14 >>> 42) & 131071L;
+        values[valuesOffset++] = (block14 >>> 25) & 131071L;
+        values[valuesOffset++] = (block14 >>> 8) & 131071L;
+        final long block15 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block14 & 255L) << 9) | (block15 >>> 55);
+        values[valuesOffset++] = (block15 >>> 38) & 131071L;
+        values[valuesOffset++] = (block15 >>> 21) & 131071L;
+        values[valuesOffset++] = (block15 >>> 4) & 131071L;
+        final long block16 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block15 & 15L) << 13) | (block16 >>> 51);
+        values[valuesOffset++] = (block16 >>> 34) & 131071L;
+        values[valuesOffset++] = (block16 >>> 17) & 131071L;
+        values[valuesOffset++] = block16 & 131071L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 9) | (byte1 << 1) | (byte2 >>> 7);
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 127) << 10) | (byte3 << 2) | (byte4 >>> 6);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 63) << 11) | (byte5 << 3) | (byte6 >>> 5);
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 31) << 12) | (byte7 << 4) | (byte8 >>> 4);
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte8 & 15) << 13) | (byte9 << 5) | (byte10 >>> 3);
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte10 & 7) << 14) | (byte11 << 6) | (byte12 >>> 2);
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte12 & 3) << 15) | (byte13 << 7) | (byte14 >>> 1);
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 1) << 16) | (byte15 << 8) | byte16;
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte17 << 9) | (byte18 << 1) | (byte19 >>> 7);
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte19 & 127) << 10) | (byte20 << 2) | (byte21 >>> 6);
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 63) << 11) | (byte22 << 3) | (byte23 >>> 5);
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 31) << 12) | (byte24 << 4) | (byte25 >>> 4);
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte25 & 15) << 13) | (byte26 << 5) | (byte27 >>> 3);
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte27 & 7) << 14) | (byte28 << 6) | (byte29 >>> 2);
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 3) << 15) | (byte30 << 7) | (byte31 >>> 1);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 1) << 16) | (byte32 << 8) | byte33;
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte34 << 9) | (byte35 << 1) | (byte36 >>> 7);
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte36 & 127) << 10) | (byte37 << 2) | (byte38 >>> 6);
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 63) << 11) | (byte39 << 3) | (byte40 >>> 5);
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 31) << 12) | (byte41 << 4) | (byte42 >>> 4);
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 15) << 13) | (byte43 << 5) | (byte44 >>> 3);
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte44 & 7) << 14) | (byte45 << 6) | (byte46 >>> 2);
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte46 & 3) << 15) | (byte47 << 7) | (byte48 >>> 1);
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte48 & 1) << 16) | (byte49 << 8) | byte50;
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte51 << 9) | (byte52 << 1) | (byte53 >>> 7);
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte53 & 127) << 10) | (byte54 << 2) | (byte55 >>> 6);
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte55 & 63) << 11) | (byte56 << 3) | (byte57 >>> 5);
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte57 & 31) << 12) | (byte58 << 4) | (byte59 >>> 4);
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 15) << 13) | (byte60 << 5) | (byte61 >>> 3);
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 7) << 14) | (byte62 << 6) | (byte63 >>> 2);
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte63 & 3) << 15) | (byte64 << 7) | (byte65 >>> 1);
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 1) << 16) | (byte66 << 8) | byte67;
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte68 << 9) | (byte69 << 1) | (byte70 >>> 7);
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        final long byte72 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte70 & 127) << 10) | (byte71 << 2) | (byte72 >>> 6);
+        final long byte73 = blocks[blocksOffset++] & 0xFF;
+        final long byte74 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte72 & 63) << 11) | (byte73 << 3) | (byte74 >>> 5);
+        final long byte75 = blocks[blocksOffset++] & 0xFF;
+        final long byte76 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte74 & 31) << 12) | (byte75 << 4) | (byte76 >>> 4);
+        final long byte77 = blocks[blocksOffset++] & 0xFF;
+        final long byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte76 & 15) << 13) | (byte77 << 5) | (byte78 >>> 3);
+        final long byte79 = blocks[blocksOffset++] & 0xFF;
+        final long byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 7) << 14) | (byte79 << 6) | (byte80 >>> 2);
+        final long byte81 = blocks[blocksOffset++] & 0xFF;
+        final long byte82 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 3) << 15) | (byte81 << 7) | (byte82 >>> 1);
+        final long byte83 = blocks[blocksOffset++] & 0xFF;
+        final long byte84 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte82 & 1) << 16) | (byte83 << 8) | byte84;
+        final long byte85 = blocks[blocksOffset++] & 0xFF;
+        final long byte86 = blocks[blocksOffset++] & 0xFF;
+        final long byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte85 << 9) | (byte86 << 1) | (byte87 >>> 7);
+        final long byte88 = blocks[blocksOffset++] & 0xFF;
+        final long byte89 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 127) << 10) | (byte88 << 2) | (byte89 >>> 6);
+        final long byte90 = blocks[blocksOffset++] & 0xFF;
+        final long byte91 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte89 & 63) << 11) | (byte90 << 3) | (byte91 >>> 5);
+        final long byte92 = blocks[blocksOffset++] & 0xFF;
+        final long byte93 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte91 & 31) << 12) | (byte92 << 4) | (byte93 >>> 4);
+        final long byte94 = blocks[blocksOffset++] & 0xFF;
+        final long byte95 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte93 & 15) << 13) | (byte94 << 5) | (byte95 >>> 3);
+        final long byte96 = blocks[blocksOffset++] & 0xFF;
+        final long byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte95 & 7) << 14) | (byte96 << 6) | (byte97 >>> 2);
+        final long byte98 = blocks[blocksOffset++] & 0xFF;
+        final long byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 3) << 15) | (byte98 << 7) | (byte99 >>> 1);
+        final long byte100 = blocks[blocksOffset++] & 0xFF;
+        final long byte101 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 1) << 16) | (byte100 << 8) | byte101;
+        final long byte102 = blocks[blocksOffset++] & 0xFF;
+        final long byte103 = blocks[blocksOffset++] & 0xFF;
+        final long byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte102 << 9) | (byte103 << 1) | (byte104 >>> 7);
+        final long byte105 = blocks[blocksOffset++] & 0xFF;
+        final long byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte104 & 127) << 10) | (byte105 << 2) | (byte106 >>> 6);
+        final long byte107 = blocks[blocksOffset++] & 0xFF;
+        final long byte108 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 63) << 11) | (byte107 << 3) | (byte108 >>> 5);
+        final long byte109 = blocks[blocksOffset++] & 0xFF;
+        final long byte110 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte108 & 31) << 12) | (byte109 << 4) | (byte110 >>> 4);
+        final long byte111 = blocks[blocksOffset++] & 0xFF;
+        final long byte112 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte110 & 15) << 13) | (byte111 << 5) | (byte112 >>> 3);
+        final long byte113 = blocks[blocksOffset++] & 0xFF;
+        final long byte114 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte112 & 7) << 14) | (byte113 << 6) | (byte114 >>> 2);
+        final long byte115 = blocks[blocksOffset++] & 0xFF;
+        final long byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte114 & 3) << 15) | (byte115 << 7) | (byte116 >>> 1);
+        final long byte117 = blocks[blocksOffset++] & 0xFF;
+        final long byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 1) << 16) | (byte117 << 8) | byte118;
+        final long byte119 = blocks[blocksOffset++] & 0xFF;
+        final long byte120 = blocks[blocksOffset++] & 0xFF;
+        final long byte121 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte119 << 9) | (byte120 << 1) | (byte121 >>> 7);
+        final long byte122 = blocks[blocksOffset++] & 0xFF;
+        final long byte123 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte121 & 127) << 10) | (byte122 << 2) | (byte123 >>> 6);
+        final long byte124 = blocks[blocksOffset++] & 0xFF;
+        final long byte125 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte123 & 63) << 11) | (byte124 << 3) | (byte125 >>> 5);
+        final long byte126 = blocks[blocksOffset++] & 0xFF;
+        final long byte127 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte125 & 31) << 12) | (byte126 << 4) | (byte127 >>> 4);
+        final long byte128 = blocks[blocksOffset++] & 0xFF;
+        final long byte129 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte127 & 15) << 13) | (byte128 << 5) | (byte129 >>> 3);
+        final long byte130 = blocks[blocksOffset++] & 0xFF;
+        final long byte131 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte129 & 7) << 14) | (byte130 << 6) | (byte131 >>> 2);
+        final long byte132 = blocks[blocksOffset++] & 0xFF;
+        final long byte133 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte131 & 3) << 15) | (byte132 << 7) | (byte133 >>> 1);
+        final long byte134 = blocks[blocksOffset++] & 0xFF;
+        final long byte135 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte133 & 1) << 16) | (byte134 << 8) | byte135;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 13) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 1) | ((values[valuesOffset] & 0xffffffffL) >>> 16);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset] & 0xffffffffL) >>> 3);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 7);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 11);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 19) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 15);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 15) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 11) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset] & 0xffffffffL) >>> 14);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset] & 0xffffffffL) >>> 1);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 5);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 9);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 13);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 17) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 47) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 13) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 9) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 5) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 1) | (values[valuesOffset] >>> 16);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 48) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 14) | (values[valuesOffset] >>> 3);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 61) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 7);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 57) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 11);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 53) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 19) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 15);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 49) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 15) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 45) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 11) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 7) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 3) | (values[valuesOffset] >>> 14);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 50) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 16) | (values[valuesOffset] >>> 1);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 5);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 59) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 9);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 55) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 13);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 51) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 17) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked18.java

@@ -0,0 +1,392 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked18 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 9;
+    }
+
+    @Override
+    public int valueCount() {
+      return 32;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 46);
+        values[valuesOffset++] = (int) ((block0 >>> 28) & 262143L);
+        values[valuesOffset++] = (int) ((block0 >>> 10) & 262143L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 1023L) << 8) | (block1 >>> 56));
+        values[valuesOffset++] = (int) ((block1 >>> 38) & 262143L);
+        values[valuesOffset++] = (int) ((block1 >>> 20) & 262143L);
+        values[valuesOffset++] = (int) ((block1 >>> 2) & 262143L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 3L) << 16) | (block2 >>> 48));
+        values[valuesOffset++] = (int) ((block2 >>> 30) & 262143L);
+        values[valuesOffset++] = (int) ((block2 >>> 12) & 262143L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 4095L) << 6) | (block3 >>> 58));
+        values[valuesOffset++] = (int) ((block3 >>> 40) & 262143L);
+        values[valuesOffset++] = (int) ((block3 >>> 22) & 262143L);
+        values[valuesOffset++] = (int) ((block3 >>> 4) & 262143L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 15L) << 14) | (block4 >>> 50));
+        values[valuesOffset++] = (int) ((block4 >>> 32) & 262143L);
+        values[valuesOffset++] = (int) ((block4 >>> 14) & 262143L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 16383L) << 4) | (block5 >>> 60));
+        values[valuesOffset++] = (int) ((block5 >>> 42) & 262143L);
+        values[valuesOffset++] = (int) ((block5 >>> 24) & 262143L);
+        values[valuesOffset++] = (int) ((block5 >>> 6) & 262143L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 63L) << 12) | (block6 >>> 52));
+        values[valuesOffset++] = (int) ((block6 >>> 34) & 262143L);
+        values[valuesOffset++] = (int) ((block6 >>> 16) & 262143L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 65535L) << 2) | (block7 >>> 62));
+        values[valuesOffset++] = (int) ((block7 >>> 44) & 262143L);
+        values[valuesOffset++] = (int) ((block7 >>> 26) & 262143L);
+        values[valuesOffset++] = (int) ((block7 >>> 8) & 262143L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 255L) << 10) | (block8 >>> 54));
+        values[valuesOffset++] = (int) ((block8 >>> 36) & 262143L);
+        values[valuesOffset++] = (int) ((block8 >>> 18) & 262143L);
+        values[valuesOffset++] = (int) (block8 & 262143L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 10) | (byte1 << 2) | (byte2 >>> 6);
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 63) << 12) | (byte3 << 4) | (byte4 >>> 4);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 15) << 14) | (byte5 << 6) | (byte6 >>> 2);
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 3) << 16) | (byte7 << 8) | byte8;
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte9 << 10) | (byte10 << 2) | (byte11 >>> 6);
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 12) | (byte12 << 4) | (byte13 >>> 4);
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 15) << 14) | (byte14 << 6) | (byte15 >>> 2);
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 3) << 16) | (byte16 << 8) | byte17;
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte18 << 10) | (byte19 << 2) | (byte20 >>> 6);
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 63) << 12) | (byte21 << 4) | (byte22 >>> 4);
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 14) | (byte23 << 6) | (byte24 >>> 2);
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 3) << 16) | (byte25 << 8) | byte26;
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte27 << 10) | (byte28 << 2) | (byte29 >>> 6);
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 63) << 12) | (byte30 << 4) | (byte31 >>> 4);
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 15) << 14) | (byte32 << 6) | (byte33 >>> 2);
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 16) | (byte34 << 8) | byte35;
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte36 << 10) | (byte37 << 2) | (byte38 >>> 6);
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 63) << 12) | (byte39 << 4) | (byte40 >>> 4);
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 15) << 14) | (byte41 << 6) | (byte42 >>> 2);
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 3) << 16) | (byte43 << 8) | byte44;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte45 << 10) | (byte46 << 2) | (byte47 >>> 6);
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 63) << 12) | (byte48 << 4) | (byte49 >>> 4);
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 15) << 14) | (byte50 << 6) | (byte51 >>> 2);
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte51 & 3) << 16) | (byte52 << 8) | byte53;
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte54 << 10) | (byte55 << 2) | (byte56 >>> 6);
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 63) << 12) | (byte57 << 4) | (byte58 >>> 4);
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 15) << 14) | (byte59 << 6) | (byte60 >>> 2);
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 3) << 16) | (byte61 << 8) | byte62;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte63 << 10) | (byte64 << 2) | (byte65 >>> 6);
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 63) << 12) | (byte66 << 4) | (byte67 >>> 4);
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 15) << 14) | (byte68 << 6) | (byte69 >>> 2);
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 3) << 16) | (byte70 << 8) | byte71;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 46;
+        values[valuesOffset++] = (block0 >>> 28) & 262143L;
+        values[valuesOffset++] = (block0 >>> 10) & 262143L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 1023L) << 8) | (block1 >>> 56);
+        values[valuesOffset++] = (block1 >>> 38) & 262143L;
+        values[valuesOffset++] = (block1 >>> 20) & 262143L;
+        values[valuesOffset++] = (block1 >>> 2) & 262143L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 3L) << 16) | (block2 >>> 48);
+        values[valuesOffset++] = (block2 >>> 30) & 262143L;
+        values[valuesOffset++] = (block2 >>> 12) & 262143L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 4095L) << 6) | (block3 >>> 58);
+        values[valuesOffset++] = (block3 >>> 40) & 262143L;
+        values[valuesOffset++] = (block3 >>> 22) & 262143L;
+        values[valuesOffset++] = (block3 >>> 4) & 262143L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 15L) << 14) | (block4 >>> 50);
+        values[valuesOffset++] = (block4 >>> 32) & 262143L;
+        values[valuesOffset++] = (block4 >>> 14) & 262143L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 16383L) << 4) | (block5 >>> 60);
+        values[valuesOffset++] = (block5 >>> 42) & 262143L;
+        values[valuesOffset++] = (block5 >>> 24) & 262143L;
+        values[valuesOffset++] = (block5 >>> 6) & 262143L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 63L) << 12) | (block6 >>> 52);
+        values[valuesOffset++] = (block6 >>> 34) & 262143L;
+        values[valuesOffset++] = (block6 >>> 16) & 262143L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 65535L) << 2) | (block7 >>> 62);
+        values[valuesOffset++] = (block7 >>> 44) & 262143L;
+        values[valuesOffset++] = (block7 >>> 26) & 262143L;
+        values[valuesOffset++] = (block7 >>> 8) & 262143L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 255L) << 10) | (block8 >>> 54);
+        values[valuesOffset++] = (block8 >>> 36) & 262143L;
+        values[valuesOffset++] = (block8 >>> 18) & 262143L;
+        values[valuesOffset++] = block8 & 262143L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 10) | (byte1 << 2) | (byte2 >>> 6);
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 63) << 12) | (byte3 << 4) | (byte4 >>> 4);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 15) << 14) | (byte5 << 6) | (byte6 >>> 2);
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte6 & 3) << 16) | (byte7 << 8) | byte8;
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte9 << 10) | (byte10 << 2) | (byte11 >>> 6);
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 63) << 12) | (byte12 << 4) | (byte13 >>> 4);
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte13 & 15) << 14) | (byte14 << 6) | (byte15 >>> 2);
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte15 & 3) << 16) | (byte16 << 8) | byte17;
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte18 << 10) | (byte19 << 2) | (byte20 >>> 6);
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte20 & 63) << 12) | (byte21 << 4) | (byte22 >>> 4);
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte22 & 15) << 14) | (byte23 << 6) | (byte24 >>> 2);
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte24 & 3) << 16) | (byte25 << 8) | byte26;
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte27 << 10) | (byte28 << 2) | (byte29 >>> 6);
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte29 & 63) << 12) | (byte30 << 4) | (byte31 >>> 4);
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte31 & 15) << 14) | (byte32 << 6) | (byte33 >>> 2);
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 3) << 16) | (byte34 << 8) | byte35;
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte36 << 10) | (byte37 << 2) | (byte38 >>> 6);
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte38 & 63) << 12) | (byte39 << 4) | (byte40 >>> 4);
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 15) << 14) | (byte41 << 6) | (byte42 >>> 2);
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 3) << 16) | (byte43 << 8) | byte44;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte45 << 10) | (byte46 << 2) | (byte47 >>> 6);
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 63) << 12) | (byte48 << 4) | (byte49 >>> 4);
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 15) << 14) | (byte50 << 6) | (byte51 >>> 2);
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte51 & 3) << 16) | (byte52 << 8) | byte53;
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte54 << 10) | (byte55 << 2) | (byte56 >>> 6);
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte56 & 63) << 12) | (byte57 << 4) | (byte58 >>> 4);
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte58 & 15) << 14) | (byte59 << 6) | (byte60 >>> 2);
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte60 & 3) << 16) | (byte61 << 8) | byte62;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte63 << 10) | (byte64 << 2) | (byte65 >>> 6);
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte65 & 63) << 12) | (byte66 << 4) | (byte67 >>> 4);
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte67 & 15) << 14) | (byte68 << 6) | (byte69 >>> 2);
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte69 & 3) << 16) | (byte70 << 8) | byte71;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 16);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 14);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 18) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 46) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 16);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 48) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 14);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 50) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 14) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 16) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 18) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked19.java

@@ -0,0 +1,720 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked19 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 19;
+    }
+
+    @Override
+    public int valueCount() {
+      return 64;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 45);
+        values[valuesOffset++] = (int) ((block0 >>> 26) & 524287L);
+        values[valuesOffset++] = (int) ((block0 >>> 7) & 524287L);
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block0 & 127L) << 12) | (block1 >>> 52));
+        values[valuesOffset++] = (int) ((block1 >>> 33) & 524287L);
+        values[valuesOffset++] = (int) ((block1 >>> 14) & 524287L);
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block1 & 16383L) << 5) | (block2 >>> 59));
+        values[valuesOffset++] = (int) ((block2 >>> 40) & 524287L);
+        values[valuesOffset++] = (int) ((block2 >>> 21) & 524287L);
+        values[valuesOffset++] = (int) ((block2 >>> 2) & 524287L);
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block2 & 3L) << 17) | (block3 >>> 47));
+        values[valuesOffset++] = (int) ((block3 >>> 28) & 524287L);
+        values[valuesOffset++] = (int) ((block3 >>> 9) & 524287L);
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block3 & 511L) << 10) | (block4 >>> 54));
+        values[valuesOffset++] = (int) ((block4 >>> 35) & 524287L);
+        values[valuesOffset++] = (int) ((block4 >>> 16) & 524287L);
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block4 & 65535L) << 3) | (block5 >>> 61));
+        values[valuesOffset++] = (int) ((block5 >>> 42) & 524287L);
+        values[valuesOffset++] = (int) ((block5 >>> 23) & 524287L);
+        values[valuesOffset++] = (int) ((block5 >>> 4) & 524287L);
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block5 & 15L) << 15) | (block6 >>> 49));
+        values[valuesOffset++] = (int) ((block6 >>> 30) & 524287L);
+        values[valuesOffset++] = (int) ((block6 >>> 11) & 524287L);
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block6 & 2047L) << 8) | (block7 >>> 56));
+        values[valuesOffset++] = (int) ((block7 >>> 37) & 524287L);
+        values[valuesOffset++] = (int) ((block7 >>> 18) & 524287L);
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block7 & 262143L) << 1) | (block8 >>> 63));
+        values[valuesOffset++] = (int) ((block8 >>> 44) & 524287L);
+        values[valuesOffset++] = (int) ((block8 >>> 25) & 524287L);
+        values[valuesOffset++] = (int) ((block8 >>> 6) & 524287L);
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block8 & 63L) << 13) | (block9 >>> 51));
+        values[valuesOffset++] = (int) ((block9 >>> 32) & 524287L);
+        values[valuesOffset++] = (int) ((block9 >>> 13) & 524287L);
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block9 & 8191L) << 6) | (block10 >>> 58));
+        values[valuesOffset++] = (int) ((block10 >>> 39) & 524287L);
+        values[valuesOffset++] = (int) ((block10 >>> 20) & 524287L);
+        values[valuesOffset++] = (int) ((block10 >>> 1) & 524287L);
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block10 & 1L) << 18) | (block11 >>> 46));
+        values[valuesOffset++] = (int) ((block11 >>> 27) & 524287L);
+        values[valuesOffset++] = (int) ((block11 >>> 8) & 524287L);
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block11 & 255L) << 11) | (block12 >>> 53));
+        values[valuesOffset++] = (int) ((block12 >>> 34) & 524287L);
+        values[valuesOffset++] = (int) ((block12 >>> 15) & 524287L);
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block12 & 32767L) << 4) | (block13 >>> 60));
+        values[valuesOffset++] = (int) ((block13 >>> 41) & 524287L);
+        values[valuesOffset++] = (int) ((block13 >>> 22) & 524287L);
+        values[valuesOffset++] = (int) ((block13 >>> 3) & 524287L);
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block13 & 7L) << 16) | (block14 >>> 48));
+        values[valuesOffset++] = (int) ((block14 >>> 29) & 524287L);
+        values[valuesOffset++] = (int) ((block14 >>> 10) & 524287L);
+        final long block15 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block14 & 1023L) << 9) | (block15 >>> 55));
+        values[valuesOffset++] = (int) ((block15 >>> 36) & 524287L);
+        values[valuesOffset++] = (int) ((block15 >>> 17) & 524287L);
+        final long block16 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block15 & 131071L) << 2) | (block16 >>> 62));
+        values[valuesOffset++] = (int) ((block16 >>> 43) & 524287L);
+        values[valuesOffset++] = (int) ((block16 >>> 24) & 524287L);
+        values[valuesOffset++] = (int) ((block16 >>> 5) & 524287L);
+        final long block17 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block16 & 31L) << 14) | (block17 >>> 50));
+        values[valuesOffset++] = (int) ((block17 >>> 31) & 524287L);
+        values[valuesOffset++] = (int) ((block17 >>> 12) & 524287L);
+        final long block18 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (((block17 & 4095L) << 7) | (block18 >>> 57));
+        values[valuesOffset++] = (int) ((block18 >>> 38) & 524287L);
+        values[valuesOffset++] = (int) ((block18 >>> 19) & 524287L);
+        values[valuesOffset++] = (int) (block18 & 524287L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 11) | (byte1 << 3) | (byte2 >>> 5);
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 31) << 14) | (byte3 << 6) | (byte4 >>> 2);
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 3) << 17) | (byte5 << 9) | (byte6 << 1) | (byte7 >>> 7);
+        final int byte8 = blocks[blocksOffset++] & 0xFF;
+        final int byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 127) << 12) | (byte8 << 4) | (byte9 >>> 4);
+        final int byte10 = blocks[blocksOffset++] & 0xFF;
+        final int byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 15) << 15) | (byte10 << 7) | (byte11 >>> 1);
+        final int byte12 = blocks[blocksOffset++] & 0xFF;
+        final int byte13 = blocks[blocksOffset++] & 0xFF;
+        final int byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 1) << 18) | (byte12 << 10) | (byte13 << 2) | (byte14 >>> 6);
+        final int byte15 = blocks[blocksOffset++] & 0xFF;
+        final int byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 63) << 13) | (byte15 << 5) | (byte16 >>> 3);
+        final int byte17 = blocks[blocksOffset++] & 0xFF;
+        final int byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 7) << 16) | (byte17 << 8) | byte18;
+        final int byte19 = blocks[blocksOffset++] & 0xFF;
+        final int byte20 = blocks[blocksOffset++] & 0xFF;
+        final int byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte19 << 11) | (byte20 << 3) | (byte21 >>> 5);
+        final int byte22 = blocks[blocksOffset++] & 0xFF;
+        final int byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 31) << 14) | (byte22 << 6) | (byte23 >>> 2);
+        final int byte24 = blocks[blocksOffset++] & 0xFF;
+        final int byte25 = blocks[blocksOffset++] & 0xFF;
+        final int byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 3) << 17) | (byte24 << 9) | (byte25 << 1) | (byte26 >>> 7);
+        final int byte27 = blocks[blocksOffset++] & 0xFF;
+        final int byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 127) << 12) | (byte27 << 4) | (byte28 >>> 4);
+        final int byte29 = blocks[blocksOffset++] & 0xFF;
+        final int byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 15) << 15) | (byte29 << 7) | (byte30 >>> 1);
+        final int byte31 = blocks[blocksOffset++] & 0xFF;
+        final int byte32 = blocks[blocksOffset++] & 0xFF;
+        final int byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 1) << 18) | (byte31 << 10) | (byte32 << 2) | (byte33 >>> 6);
+        final int byte34 = blocks[blocksOffset++] & 0xFF;
+        final int byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 63) << 13) | (byte34 << 5) | (byte35 >>> 3);
+        final int byte36 = blocks[blocksOffset++] & 0xFF;
+        final int byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 7) << 16) | (byte36 << 8) | byte37;
+        final int byte38 = blocks[blocksOffset++] & 0xFF;
+        final int byte39 = blocks[blocksOffset++] & 0xFF;
+        final int byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte38 << 11) | (byte39 << 3) | (byte40 >>> 5);
+        final int byte41 = blocks[blocksOffset++] & 0xFF;
+        final int byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 31) << 14) | (byte41 << 6) | (byte42 >>> 2);
+        final int byte43 = blocks[blocksOffset++] & 0xFF;
+        final int byte44 = blocks[blocksOffset++] & 0xFF;
+        final int byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 3) << 17) | (byte43 << 9) | (byte44 << 1) | (byte45 >>> 7);
+        final int byte46 = blocks[blocksOffset++] & 0xFF;
+        final int byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 127) << 12) | (byte46 << 4) | (byte47 >>> 4);
+        final int byte48 = blocks[blocksOffset++] & 0xFF;
+        final int byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 15) << 15) | (byte48 << 7) | (byte49 >>> 1);
+        final int byte50 = blocks[blocksOffset++] & 0xFF;
+        final int byte51 = blocks[blocksOffset++] & 0xFF;
+        final int byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 1) << 18) | (byte50 << 10) | (byte51 << 2) | (byte52 >>> 6);
+        final int byte53 = blocks[blocksOffset++] & 0xFF;
+        final int byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 63) << 13) | (byte53 << 5) | (byte54 >>> 3);
+        final int byte55 = blocks[blocksOffset++] & 0xFF;
+        final int byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 7) << 16) | (byte55 << 8) | byte56;
+        final int byte57 = blocks[blocksOffset++] & 0xFF;
+        final int byte58 = blocks[blocksOffset++] & 0xFF;
+        final int byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte57 << 11) | (byte58 << 3) | (byte59 >>> 5);
+        final int byte60 = blocks[blocksOffset++] & 0xFF;
+        final int byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 31) << 14) | (byte60 << 6) | (byte61 >>> 2);
+        final int byte62 = blocks[blocksOffset++] & 0xFF;
+        final int byte63 = blocks[blocksOffset++] & 0xFF;
+        final int byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 3) << 17) | (byte62 << 9) | (byte63 << 1) | (byte64 >>> 7);
+        final int byte65 = blocks[blocksOffset++] & 0xFF;
+        final int byte66 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte64 & 127) << 12) | (byte65 << 4) | (byte66 >>> 4);
+        final int byte67 = blocks[blocksOffset++] & 0xFF;
+        final int byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte66 & 15) << 15) | (byte67 << 7) | (byte68 >>> 1);
+        final int byte69 = blocks[blocksOffset++] & 0xFF;
+        final int byte70 = blocks[blocksOffset++] & 0xFF;
+        final int byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 1) << 18) | (byte69 << 10) | (byte70 << 2) | (byte71 >>> 6);
+        final int byte72 = blocks[blocksOffset++] & 0xFF;
+        final int byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 63) << 13) | (byte72 << 5) | (byte73 >>> 3);
+        final int byte74 = blocks[blocksOffset++] & 0xFF;
+        final int byte75 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 7) << 16) | (byte74 << 8) | byte75;
+        final int byte76 = blocks[blocksOffset++] & 0xFF;
+        final int byte77 = blocks[blocksOffset++] & 0xFF;
+        final int byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte76 << 11) | (byte77 << 3) | (byte78 >>> 5);
+        final int byte79 = blocks[blocksOffset++] & 0xFF;
+        final int byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 31) << 14) | (byte79 << 6) | (byte80 >>> 2);
+        final int byte81 = blocks[blocksOffset++] & 0xFF;
+        final int byte82 = blocks[blocksOffset++] & 0xFF;
+        final int byte83 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 3) << 17) | (byte81 << 9) | (byte82 << 1) | (byte83 >>> 7);
+        final int byte84 = blocks[blocksOffset++] & 0xFF;
+        final int byte85 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte83 & 127) << 12) | (byte84 << 4) | (byte85 >>> 4);
+        final int byte86 = blocks[blocksOffset++] & 0xFF;
+        final int byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte85 & 15) << 15) | (byte86 << 7) | (byte87 >>> 1);
+        final int byte88 = blocks[blocksOffset++] & 0xFF;
+        final int byte89 = blocks[blocksOffset++] & 0xFF;
+        final int byte90 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 1) << 18) | (byte88 << 10) | (byte89 << 2) | (byte90 >>> 6);
+        final int byte91 = blocks[blocksOffset++] & 0xFF;
+        final int byte92 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte90 & 63) << 13) | (byte91 << 5) | (byte92 >>> 3);
+        final int byte93 = blocks[blocksOffset++] & 0xFF;
+        final int byte94 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte92 & 7) << 16) | (byte93 << 8) | byte94;
+        final int byte95 = blocks[blocksOffset++] & 0xFF;
+        final int byte96 = blocks[blocksOffset++] & 0xFF;
+        final int byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte95 << 11) | (byte96 << 3) | (byte97 >>> 5);
+        final int byte98 = blocks[blocksOffset++] & 0xFF;
+        final int byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 31) << 14) | (byte98 << 6) | (byte99 >>> 2);
+        final int byte100 = blocks[blocksOffset++] & 0xFF;
+        final int byte101 = blocks[blocksOffset++] & 0xFF;
+        final int byte102 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 3) << 17) | (byte100 << 9) | (byte101 << 1) | (byte102 >>> 7);
+        final int byte103 = blocks[blocksOffset++] & 0xFF;
+        final int byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte102 & 127) << 12) | (byte103 << 4) | (byte104 >>> 4);
+        final int byte105 = blocks[blocksOffset++] & 0xFF;
+        final int byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte104 & 15) << 15) | (byte105 << 7) | (byte106 >>> 1);
+        final int byte107 = blocks[blocksOffset++] & 0xFF;
+        final int byte108 = blocks[blocksOffset++] & 0xFF;
+        final int byte109 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 1) << 18) | (byte107 << 10) | (byte108 << 2) | (byte109 >>> 6);
+        final int byte110 = blocks[blocksOffset++] & 0xFF;
+        final int byte111 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte109 & 63) << 13) | (byte110 << 5) | (byte111 >>> 3);
+        final int byte112 = blocks[blocksOffset++] & 0xFF;
+        final int byte113 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte111 & 7) << 16) | (byte112 << 8) | byte113;
+        final int byte114 = blocks[blocksOffset++] & 0xFF;
+        final int byte115 = blocks[blocksOffset++] & 0xFF;
+        final int byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte114 << 11) | (byte115 << 3) | (byte116 >>> 5);
+        final int byte117 = blocks[blocksOffset++] & 0xFF;
+        final int byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 31) << 14) | (byte117 << 6) | (byte118 >>> 2);
+        final int byte119 = blocks[blocksOffset++] & 0xFF;
+        final int byte120 = blocks[blocksOffset++] & 0xFF;
+        final int byte121 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte118 & 3) << 17) | (byte119 << 9) | (byte120 << 1) | (byte121 >>> 7);
+        final int byte122 = blocks[blocksOffset++] & 0xFF;
+        final int byte123 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte121 & 127) << 12) | (byte122 << 4) | (byte123 >>> 4);
+        final int byte124 = blocks[blocksOffset++] & 0xFF;
+        final int byte125 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte123 & 15) << 15) | (byte124 << 7) | (byte125 >>> 1);
+        final int byte126 = blocks[blocksOffset++] & 0xFF;
+        final int byte127 = blocks[blocksOffset++] & 0xFF;
+        final int byte128 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte125 & 1) << 18) | (byte126 << 10) | (byte127 << 2) | (byte128 >>> 6);
+        final int byte129 = blocks[blocksOffset++] & 0xFF;
+        final int byte130 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte128 & 63) << 13) | (byte129 << 5) | (byte130 >>> 3);
+        final int byte131 = blocks[blocksOffset++] & 0xFF;
+        final int byte132 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte130 & 7) << 16) | (byte131 << 8) | byte132;
+        final int byte133 = blocks[blocksOffset++] & 0xFF;
+        final int byte134 = blocks[blocksOffset++] & 0xFF;
+        final int byte135 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte133 << 11) | (byte134 << 3) | (byte135 >>> 5);
+        final int byte136 = blocks[blocksOffset++] & 0xFF;
+        final int byte137 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte135 & 31) << 14) | (byte136 << 6) | (byte137 >>> 2);
+        final int byte138 = blocks[blocksOffset++] & 0xFF;
+        final int byte139 = blocks[blocksOffset++] & 0xFF;
+        final int byte140 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte137 & 3) << 17) | (byte138 << 9) | (byte139 << 1) | (byte140 >>> 7);
+        final int byte141 = blocks[blocksOffset++] & 0xFF;
+        final int byte142 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte140 & 127) << 12) | (byte141 << 4) | (byte142 >>> 4);
+        final int byte143 = blocks[blocksOffset++] & 0xFF;
+        final int byte144 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte142 & 15) << 15) | (byte143 << 7) | (byte144 >>> 1);
+        final int byte145 = blocks[blocksOffset++] & 0xFF;
+        final int byte146 = blocks[blocksOffset++] & 0xFF;
+        final int byte147 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte144 & 1) << 18) | (byte145 << 10) | (byte146 << 2) | (byte147 >>> 6);
+        final int byte148 = blocks[blocksOffset++] & 0xFF;
+        final int byte149 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte147 & 63) << 13) | (byte148 << 5) | (byte149 >>> 3);
+        final int byte150 = blocks[blocksOffset++] & 0xFF;
+        final int byte151 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte149 & 7) << 16) | (byte150 << 8) | byte151;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 45;
+        values[valuesOffset++] = (block0 >>> 26) & 524287L;
+        values[valuesOffset++] = (block0 >>> 7) & 524287L;
+        final long block1 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block0 & 127L) << 12) | (block1 >>> 52);
+        values[valuesOffset++] = (block1 >>> 33) & 524287L;
+        values[valuesOffset++] = (block1 >>> 14) & 524287L;
+        final long block2 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block1 & 16383L) << 5) | (block2 >>> 59);
+        values[valuesOffset++] = (block2 >>> 40) & 524287L;
+        values[valuesOffset++] = (block2 >>> 21) & 524287L;
+        values[valuesOffset++] = (block2 >>> 2) & 524287L;
+        final long block3 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block2 & 3L) << 17) | (block3 >>> 47);
+        values[valuesOffset++] = (block3 >>> 28) & 524287L;
+        values[valuesOffset++] = (block3 >>> 9) & 524287L;
+        final long block4 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block3 & 511L) << 10) | (block4 >>> 54);
+        values[valuesOffset++] = (block4 >>> 35) & 524287L;
+        values[valuesOffset++] = (block4 >>> 16) & 524287L;
+        final long block5 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block4 & 65535L) << 3) | (block5 >>> 61);
+        values[valuesOffset++] = (block5 >>> 42) & 524287L;
+        values[valuesOffset++] = (block5 >>> 23) & 524287L;
+        values[valuesOffset++] = (block5 >>> 4) & 524287L;
+        final long block6 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block5 & 15L) << 15) | (block6 >>> 49);
+        values[valuesOffset++] = (block6 >>> 30) & 524287L;
+        values[valuesOffset++] = (block6 >>> 11) & 524287L;
+        final long block7 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block6 & 2047L) << 8) | (block7 >>> 56);
+        values[valuesOffset++] = (block7 >>> 37) & 524287L;
+        values[valuesOffset++] = (block7 >>> 18) & 524287L;
+        final long block8 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block7 & 262143L) << 1) | (block8 >>> 63);
+        values[valuesOffset++] = (block8 >>> 44) & 524287L;
+        values[valuesOffset++] = (block8 >>> 25) & 524287L;
+        values[valuesOffset++] = (block8 >>> 6) & 524287L;
+        final long block9 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block8 & 63L) << 13) | (block9 >>> 51);
+        values[valuesOffset++] = (block9 >>> 32) & 524287L;
+        values[valuesOffset++] = (block9 >>> 13) & 524287L;
+        final long block10 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block9 & 8191L) << 6) | (block10 >>> 58);
+        values[valuesOffset++] = (block10 >>> 39) & 524287L;
+        values[valuesOffset++] = (block10 >>> 20) & 524287L;
+        values[valuesOffset++] = (block10 >>> 1) & 524287L;
+        final long block11 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block10 & 1L) << 18) | (block11 >>> 46);
+        values[valuesOffset++] = (block11 >>> 27) & 524287L;
+        values[valuesOffset++] = (block11 >>> 8) & 524287L;
+        final long block12 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block11 & 255L) << 11) | (block12 >>> 53);
+        values[valuesOffset++] = (block12 >>> 34) & 524287L;
+        values[valuesOffset++] = (block12 >>> 15) & 524287L;
+        final long block13 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block12 & 32767L) << 4) | (block13 >>> 60);
+        values[valuesOffset++] = (block13 >>> 41) & 524287L;
+        values[valuesOffset++] = (block13 >>> 22) & 524287L;
+        values[valuesOffset++] = (block13 >>> 3) & 524287L;
+        final long block14 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block13 & 7L) << 16) | (block14 >>> 48);
+        values[valuesOffset++] = (block14 >>> 29) & 524287L;
+        values[valuesOffset++] = (block14 >>> 10) & 524287L;
+        final long block15 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block14 & 1023L) << 9) | (block15 >>> 55);
+        values[valuesOffset++] = (block15 >>> 36) & 524287L;
+        values[valuesOffset++] = (block15 >>> 17) & 524287L;
+        final long block16 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block15 & 131071L) << 2) | (block16 >>> 62);
+        values[valuesOffset++] = (block16 >>> 43) & 524287L;
+        values[valuesOffset++] = (block16 >>> 24) & 524287L;
+        values[valuesOffset++] = (block16 >>> 5) & 524287L;
+        final long block17 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block16 & 31L) << 14) | (block17 >>> 50);
+        values[valuesOffset++] = (block17 >>> 31) & 524287L;
+        values[valuesOffset++] = (block17 >>> 12) & 524287L;
+        final long block18 = blocks[blocksOffset++];
+        values[valuesOffset++] = ((block17 & 4095L) << 7) | (block18 >>> 57);
+        values[valuesOffset++] = (block18 >>> 38) & 524287L;
+        values[valuesOffset++] = (block18 >>> 19) & 524287L;
+        values[valuesOffset++] = block18 & 524287L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte0 << 11) | (byte1 << 3) | (byte2 >>> 5);
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte2 & 31) << 14) | (byte3 << 6) | (byte4 >>> 2);
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte4 & 3) << 17) | (byte5 << 9) | (byte6 << 1) | (byte7 >>> 7);
+        final long byte8 = blocks[blocksOffset++] & 0xFF;
+        final long byte9 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte7 & 127) << 12) | (byte8 << 4) | (byte9 >>> 4);
+        final long byte10 = blocks[blocksOffset++] & 0xFF;
+        final long byte11 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte9 & 15) << 15) | (byte10 << 7) | (byte11 >>> 1);
+        final long byte12 = blocks[blocksOffset++] & 0xFF;
+        final long byte13 = blocks[blocksOffset++] & 0xFF;
+        final long byte14 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte11 & 1) << 18) | (byte12 << 10) | (byte13 << 2) | (byte14 >>> 6);
+        final long byte15 = blocks[blocksOffset++] & 0xFF;
+        final long byte16 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte14 & 63) << 13) | (byte15 << 5) | (byte16 >>> 3);
+        final long byte17 = blocks[blocksOffset++] & 0xFF;
+        final long byte18 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte16 & 7) << 16) | (byte17 << 8) | byte18;
+        final long byte19 = blocks[blocksOffset++] & 0xFF;
+        final long byte20 = blocks[blocksOffset++] & 0xFF;
+        final long byte21 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte19 << 11) | (byte20 << 3) | (byte21 >>> 5);
+        final long byte22 = blocks[blocksOffset++] & 0xFF;
+        final long byte23 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte21 & 31) << 14) | (byte22 << 6) | (byte23 >>> 2);
+        final long byte24 = blocks[blocksOffset++] & 0xFF;
+        final long byte25 = blocks[blocksOffset++] & 0xFF;
+        final long byte26 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte23 & 3) << 17) | (byte24 << 9) | (byte25 << 1) | (byte26 >>> 7);
+        final long byte27 = blocks[blocksOffset++] & 0xFF;
+        final long byte28 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte26 & 127) << 12) | (byte27 << 4) | (byte28 >>> 4);
+        final long byte29 = blocks[blocksOffset++] & 0xFF;
+        final long byte30 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte28 & 15) << 15) | (byte29 << 7) | (byte30 >>> 1);
+        final long byte31 = blocks[blocksOffset++] & 0xFF;
+        final long byte32 = blocks[blocksOffset++] & 0xFF;
+        final long byte33 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte30 & 1) << 18) | (byte31 << 10) | (byte32 << 2) | (byte33 >>> 6);
+        final long byte34 = blocks[blocksOffset++] & 0xFF;
+        final long byte35 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte33 & 63) << 13) | (byte34 << 5) | (byte35 >>> 3);
+        final long byte36 = blocks[blocksOffset++] & 0xFF;
+        final long byte37 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte35 & 7) << 16) | (byte36 << 8) | byte37;
+        final long byte38 = blocks[blocksOffset++] & 0xFF;
+        final long byte39 = blocks[blocksOffset++] & 0xFF;
+        final long byte40 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte38 << 11) | (byte39 << 3) | (byte40 >>> 5);
+        final long byte41 = blocks[blocksOffset++] & 0xFF;
+        final long byte42 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte40 & 31) << 14) | (byte41 << 6) | (byte42 >>> 2);
+        final long byte43 = blocks[blocksOffset++] & 0xFF;
+        final long byte44 = blocks[blocksOffset++] & 0xFF;
+        final long byte45 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte42 & 3) << 17) | (byte43 << 9) | (byte44 << 1) | (byte45 >>> 7);
+        final long byte46 = blocks[blocksOffset++] & 0xFF;
+        final long byte47 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte45 & 127) << 12) | (byte46 << 4) | (byte47 >>> 4);
+        final long byte48 = blocks[blocksOffset++] & 0xFF;
+        final long byte49 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte47 & 15) << 15) | (byte48 << 7) | (byte49 >>> 1);
+        final long byte50 = blocks[blocksOffset++] & 0xFF;
+        final long byte51 = blocks[blocksOffset++] & 0xFF;
+        final long byte52 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte49 & 1) << 18) | (byte50 << 10) | (byte51 << 2) | (byte52 >>> 6);
+        final long byte53 = blocks[blocksOffset++] & 0xFF;
+        final long byte54 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte52 & 63) << 13) | (byte53 << 5) | (byte54 >>> 3);
+        final long byte55 = blocks[blocksOffset++] & 0xFF;
+        final long byte56 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte54 & 7) << 16) | (byte55 << 8) | byte56;
+        final long byte57 = blocks[blocksOffset++] & 0xFF;
+        final long byte58 = blocks[blocksOffset++] & 0xFF;
+        final long byte59 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte57 << 11) | (byte58 << 3) | (byte59 >>> 5);
+        final long byte60 = blocks[blocksOffset++] & 0xFF;
+        final long byte61 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte59 & 31) << 14) | (byte60 << 6) | (byte61 >>> 2);
+        final long byte62 = blocks[blocksOffset++] & 0xFF;
+        final long byte63 = blocks[blocksOffset++] & 0xFF;
+        final long byte64 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte61 & 3) << 17) | (byte62 << 9) | (byte63 << 1) | (byte64 >>> 7);
+        final long byte65 = blocks[blocksOffset++] & 0xFF;
+        final long byte66 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte64 & 127) << 12) | (byte65 << 4) | (byte66 >>> 4);
+        final long byte67 = blocks[blocksOffset++] & 0xFF;
+        final long byte68 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte66 & 15) << 15) | (byte67 << 7) | (byte68 >>> 1);
+        final long byte69 = blocks[blocksOffset++] & 0xFF;
+        final long byte70 = blocks[blocksOffset++] & 0xFF;
+        final long byte71 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte68 & 1) << 18) | (byte69 << 10) | (byte70 << 2) | (byte71 >>> 6);
+        final long byte72 = blocks[blocksOffset++] & 0xFF;
+        final long byte73 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte71 & 63) << 13) | (byte72 << 5) | (byte73 >>> 3);
+        final long byte74 = blocks[blocksOffset++] & 0xFF;
+        final long byte75 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte73 & 7) << 16) | (byte74 << 8) | byte75;
+        final long byte76 = blocks[blocksOffset++] & 0xFF;
+        final long byte77 = blocks[blocksOffset++] & 0xFF;
+        final long byte78 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte76 << 11) | (byte77 << 3) | (byte78 >>> 5);
+        final long byte79 = blocks[blocksOffset++] & 0xFF;
+        final long byte80 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte78 & 31) << 14) | (byte79 << 6) | (byte80 >>> 2);
+        final long byte81 = blocks[blocksOffset++] & 0xFF;
+        final long byte82 = blocks[blocksOffset++] & 0xFF;
+        final long byte83 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte80 & 3) << 17) | (byte81 << 9) | (byte82 << 1) | (byte83 >>> 7);
+        final long byte84 = blocks[blocksOffset++] & 0xFF;
+        final long byte85 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte83 & 127) << 12) | (byte84 << 4) | (byte85 >>> 4);
+        final long byte86 = blocks[blocksOffset++] & 0xFF;
+        final long byte87 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte85 & 15) << 15) | (byte86 << 7) | (byte87 >>> 1);
+        final long byte88 = blocks[blocksOffset++] & 0xFF;
+        final long byte89 = blocks[blocksOffset++] & 0xFF;
+        final long byte90 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte87 & 1) << 18) | (byte88 << 10) | (byte89 << 2) | (byte90 >>> 6);
+        final long byte91 = blocks[blocksOffset++] & 0xFF;
+        final long byte92 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte90 & 63) << 13) | (byte91 << 5) | (byte92 >>> 3);
+        final long byte93 = blocks[blocksOffset++] & 0xFF;
+        final long byte94 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte92 & 7) << 16) | (byte93 << 8) | byte94;
+        final long byte95 = blocks[blocksOffset++] & 0xFF;
+        final long byte96 = blocks[blocksOffset++] & 0xFF;
+        final long byte97 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte95 << 11) | (byte96 << 3) | (byte97 >>> 5);
+        final long byte98 = blocks[blocksOffset++] & 0xFF;
+        final long byte99 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte97 & 31) << 14) | (byte98 << 6) | (byte99 >>> 2);
+        final long byte100 = blocks[blocksOffset++] & 0xFF;
+        final long byte101 = blocks[blocksOffset++] & 0xFF;
+        final long byte102 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte99 & 3) << 17) | (byte100 << 9) | (byte101 << 1) | (byte102 >>> 7);
+        final long byte103 = blocks[blocksOffset++] & 0xFF;
+        final long byte104 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte102 & 127) << 12) | (byte103 << 4) | (byte104 >>> 4);
+        final long byte105 = blocks[blocksOffset++] & 0xFF;
+        final long byte106 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte104 & 15) << 15) | (byte105 << 7) | (byte106 >>> 1);
+        final long byte107 = blocks[blocksOffset++] & 0xFF;
+        final long byte108 = blocks[blocksOffset++] & 0xFF;
+        final long byte109 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte106 & 1) << 18) | (byte107 << 10) | (byte108 << 2) | (byte109 >>> 6);
+        final long byte110 = blocks[blocksOffset++] & 0xFF;
+        final long byte111 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte109 & 63) << 13) | (byte110 << 5) | (byte111 >>> 3);
+        final long byte112 = blocks[blocksOffset++] & 0xFF;
+        final long byte113 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte111 & 7) << 16) | (byte112 << 8) | byte113;
+        final long byte114 = blocks[blocksOffset++] & 0xFF;
+        final long byte115 = blocks[blocksOffset++] & 0xFF;
+        final long byte116 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte114 << 11) | (byte115 << 3) | (byte116 >>> 5);
+        final long byte117 = blocks[blocksOffset++] & 0xFF;
+        final long byte118 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte116 & 31) << 14) | (byte117 << 6) | (byte118 >>> 2);
+        final long byte119 = blocks[blocksOffset++] & 0xFF;
+        final long byte120 = blocks[blocksOffset++] & 0xFF;
+        final long byte121 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte118 & 3) << 17) | (byte119 << 9) | (byte120 << 1) | (byte121 >>> 7);
+        final long byte122 = blocks[blocksOffset++] & 0xFF;
+        final long byte123 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte121 & 127) << 12) | (byte122 << 4) | (byte123 >>> 4);
+        final long byte124 = blocks[blocksOffset++] & 0xFF;
+        final long byte125 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte123 & 15) << 15) | (byte124 << 7) | (byte125 >>> 1);
+        final long byte126 = blocks[blocksOffset++] & 0xFF;
+        final long byte127 = blocks[blocksOffset++] & 0xFF;
+        final long byte128 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte125 & 1) << 18) | (byte126 << 10) | (byte127 << 2) | (byte128 >>> 6);
+        final long byte129 = blocks[blocksOffset++] & 0xFF;
+        final long byte130 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte128 & 63) << 13) | (byte129 << 5) | (byte130 >>> 3);
+        final long byte131 = blocks[blocksOffset++] & 0xFF;
+        final long byte132 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte130 & 7) << 16) | (byte131 << 8) | byte132;
+        final long byte133 = blocks[blocksOffset++] & 0xFF;
+        final long byte134 = blocks[blocksOffset++] & 0xFF;
+        final long byte135 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = (byte133 << 11) | (byte134 << 3) | (byte135 >>> 5);
+        final long byte136 = blocks[blocksOffset++] & 0xFF;
+        final long byte137 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte135 & 31) << 14) | (byte136 << 6) | (byte137 >>> 2);
+        final long byte138 = blocks[blocksOffset++] & 0xFF;
+        final long byte139 = blocks[blocksOffset++] & 0xFF;
+        final long byte140 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte137 & 3) << 17) | (byte138 << 9) | (byte139 << 1) | (byte140 >>> 7);
+        final long byte141 = blocks[blocksOffset++] & 0xFF;
+        final long byte142 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte140 & 127) << 12) | (byte141 << 4) | (byte142 >>> 4);
+        final long byte143 = blocks[blocksOffset++] & 0xFF;
+        final long byte144 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte142 & 15) << 15) | (byte143 << 7) | (byte144 >>> 1);
+        final long byte145 = blocks[blocksOffset++] & 0xFF;
+        final long byte146 = blocks[blocksOffset++] & 0xFF;
+        final long byte147 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte144 & 1) << 18) | (byte145 << 10) | (byte146 << 2) | (byte147 >>> 6);
+        final long byte148 = blocks[blocksOffset++] & 0xFF;
+        final long byte149 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte147 & 63) << 13) | (byte148 << 5) | (byte149 >>> 3);
+        final long byte150 = blocks[blocksOffset++] & 0xFF;
+        final long byte151 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = ((byte149 & 7) << 16) | (byte150 << 8) | byte151;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 45) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 7) | ((values[valuesOffset] & 0xffffffffL) >>> 12);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 33) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset] & 0xffffffffL) >>> 5);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 59) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 21) | ((values[valuesOffset++] & 0xffffffffL) << 2) | ((values[valuesOffset] & 0xffffffffL) >>> 17);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 47) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 9) | ((values[valuesOffset] & 0xffffffffL) >>> 10);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 35) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset] & 0xffffffffL) >>> 3);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 61) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 23) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset] & 0xffffffffL) >>> 15);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 49) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 11) | ((values[valuesOffset] & 0xffffffffL) >>> 8);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 37) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset] & 0xffffffffL) >>> 1);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 63) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 25) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset] & 0xffffffffL) >>> 13);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 51) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 13) | ((values[valuesOffset] & 0xffffffffL) >>> 6);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 39) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 1) | ((values[valuesOffset] & 0xffffffffL) >>> 18);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 27) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset] & 0xffffffffL) >>> 11);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 53) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 15) | ((values[valuesOffset] & 0xffffffffL) >>> 4);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 41) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 3) | ((values[valuesOffset] & 0xffffffffL) >>> 16);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 29) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset] & 0xffffffffL) >>> 9);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 55) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 17) | ((values[valuesOffset] & 0xffffffffL) >>> 2);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 43) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 5) | ((values[valuesOffset] & 0xffffffffL) >>> 14);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 31) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset] & 0xffffffffL) >>> 7);
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 57) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 19) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 45) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 7) | (values[valuesOffset] >>> 12);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 52) | (values[valuesOffset++] << 33) | (values[valuesOffset++] << 14) | (values[valuesOffset] >>> 5);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 59) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 21) | (values[valuesOffset++] << 2) | (values[valuesOffset] >>> 17);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 47) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 9) | (values[valuesOffset] >>> 10);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 54) | (values[valuesOffset++] << 35) | (values[valuesOffset++] << 16) | (values[valuesOffset] >>> 3);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 61) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 23) | (values[valuesOffset++] << 4) | (values[valuesOffset] >>> 15);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 49) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 11) | (values[valuesOffset] >>> 8);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 56) | (values[valuesOffset++] << 37) | (values[valuesOffset++] << 18) | (values[valuesOffset] >>> 1);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 63) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 25) | (values[valuesOffset++] << 6) | (values[valuesOffset] >>> 13);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 51) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 13) | (values[valuesOffset] >>> 6);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 58) | (values[valuesOffset++] << 39) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 1) | (values[valuesOffset] >>> 18);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 46) | (values[valuesOffset++] << 27) | (values[valuesOffset++] << 8) | (values[valuesOffset] >>> 11);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 53) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 15) | (values[valuesOffset] >>> 4);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 60) | (values[valuesOffset++] << 41) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 3) | (values[valuesOffset] >>> 16);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 48) | (values[valuesOffset++] << 29) | (values[valuesOffset++] << 10) | (values[valuesOffset] >>> 9);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 55) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 17) | (values[valuesOffset] >>> 2);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 43) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 5) | (values[valuesOffset] >>> 14);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 50) | (values[valuesOffset++] << 31) | (values[valuesOffset++] << 12) | (values[valuesOffset] >>> 7);
+        blocks[blocksOffset++] = (values[valuesOffset++] << 57) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 19) | values[valuesOffset++];
+      }
+    }
+
+}

lucene/core/src/java/org/apache/lucene/util/packed/BulkOperationPacked2.java

@@ -0,0 +1,232 @@
+// This file has been automatically generated, DO NOT EDIT
+
+package org.apache.lucene.util.packed;
+
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * Efficient sequential read/write of packed integers.
+ */
+final class BulkOperationPacked2 extends BulkOperation {
+    @Override
+    public int blockCount() {
+      return 1;
+    }
+
+    @Override
+    public int valueCount() {
+      return 32;
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = (int) (block0 >>> 62);
+        values[valuesOffset++] = (int) ((block0 >>> 60) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 58) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 56) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 54) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 52) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 50) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 48) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 46) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 44) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 42) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 40) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 38) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 36) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 34) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 32) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 30) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 28) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 26) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 24) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 22) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 20) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 18) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 16) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 14) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 12) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 10) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 8) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 6) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 4) & 3L);
+        values[valuesOffset++] = (int) ((block0 >>> 2) & 3L);
+        values[valuesOffset++] = (int) (block0 & 3L);
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, int[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final int byte0 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte0 >>> 6;
+        values[valuesOffset++] = (byte0 >>> 4) & 3;
+        values[valuesOffset++] = (byte0 >>> 2) & 3;
+        values[valuesOffset++] = byte0 & 3;
+        final int byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte1 >>> 6;
+        values[valuesOffset++] = (byte1 >>> 4) & 3;
+        values[valuesOffset++] = (byte1 >>> 2) & 3;
+        values[valuesOffset++] = byte1 & 3;
+        final int byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte2 >>> 6;
+        values[valuesOffset++] = (byte2 >>> 4) & 3;
+        values[valuesOffset++] = (byte2 >>> 2) & 3;
+        values[valuesOffset++] = byte2 & 3;
+        final int byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte3 >>> 6;
+        values[valuesOffset++] = (byte3 >>> 4) & 3;
+        values[valuesOffset++] = (byte3 >>> 2) & 3;
+        values[valuesOffset++] = byte3 & 3;
+        final int byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte4 >>> 6;
+        values[valuesOffset++] = (byte4 >>> 4) & 3;
+        values[valuesOffset++] = (byte4 >>> 2) & 3;
+        values[valuesOffset++] = byte4 & 3;
+        final int byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte5 >>> 6;
+        values[valuesOffset++] = (byte5 >>> 4) & 3;
+        values[valuesOffset++] = (byte5 >>> 2) & 3;
+        values[valuesOffset++] = byte5 & 3;
+        final int byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte6 >>> 6;
+        values[valuesOffset++] = (byte6 >>> 4) & 3;
+        values[valuesOffset++] = (byte6 >>> 2) & 3;
+        values[valuesOffset++] = byte6 & 3;
+        final int byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte7 >>> 6;
+        values[valuesOffset++] = (byte7 >>> 4) & 3;
+        values[valuesOffset++] = (byte7 >>> 2) & 3;
+        values[valuesOffset++] = byte7 & 3;
+      }
+    }
+
+    @Override
+    public void decode(long[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long block0 = blocks[blocksOffset++];
+        values[valuesOffset++] = block0 >>> 62;
+        values[valuesOffset++] = (block0 >>> 60) & 3L;
+        values[valuesOffset++] = (block0 >>> 58) & 3L;
+        values[valuesOffset++] = (block0 >>> 56) & 3L;
+        values[valuesOffset++] = (block0 >>> 54) & 3L;
+        values[valuesOffset++] = (block0 >>> 52) & 3L;
+        values[valuesOffset++] = (block0 >>> 50) & 3L;
+        values[valuesOffset++] = (block0 >>> 48) & 3L;
+        values[valuesOffset++] = (block0 >>> 46) & 3L;
+        values[valuesOffset++] = (block0 >>> 44) & 3L;
+        values[valuesOffset++] = (block0 >>> 42) & 3L;
+        values[valuesOffset++] = (block0 >>> 40) & 3L;
+        values[valuesOffset++] = (block0 >>> 38) & 3L;
+        values[valuesOffset++] = (block0 >>> 36) & 3L;
+        values[valuesOffset++] = (block0 >>> 34) & 3L;
+        values[valuesOffset++] = (block0 >>> 32) & 3L;
+        values[valuesOffset++] = (block0 >>> 30) & 3L;
+        values[valuesOffset++] = (block0 >>> 28) & 3L;
+        values[valuesOffset++] = (block0 >>> 26) & 3L;
+        values[valuesOffset++] = (block0 >>> 24) & 3L;
+        values[valuesOffset++] = (block0 >>> 22) & 3L;
+        values[valuesOffset++] = (block0 >>> 20) & 3L;
+        values[valuesOffset++] = (block0 >>> 18) & 3L;
+        values[valuesOffset++] = (block0 >>> 16) & 3L;
+        values[valuesOffset++] = (block0 >>> 14) & 3L;
+        values[valuesOffset++] = (block0 >>> 12) & 3L;
+        values[valuesOffset++] = (block0 >>> 10) & 3L;
+        values[valuesOffset++] = (block0 >>> 8) & 3L;
+        values[valuesOffset++] = (block0 >>> 6) & 3L;
+        values[valuesOffset++] = (block0 >>> 4) & 3L;
+        values[valuesOffset++] = (block0 >>> 2) & 3L;
+        values[valuesOffset++] = block0 & 3L;
+      }
+    }
+
+    @Override
+    public void decode(byte[] blocks, int blocksOffset, long[] values, int valuesOffset, int iterations) {
+      assert blocksOffset + 8 * iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        final long byte0 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte0 >>> 6;
+        values[valuesOffset++] = (byte0 >>> 4) & 3;
+        values[valuesOffset++] = (byte0 >>> 2) & 3;
+        values[valuesOffset++] = byte0 & 3;
+        final long byte1 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte1 >>> 6;
+        values[valuesOffset++] = (byte1 >>> 4) & 3;
+        values[valuesOffset++] = (byte1 >>> 2) & 3;
+        values[valuesOffset++] = byte1 & 3;
+        final long byte2 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte2 >>> 6;
+        values[valuesOffset++] = (byte2 >>> 4) & 3;
+        values[valuesOffset++] = (byte2 >>> 2) & 3;
+        values[valuesOffset++] = byte2 & 3;
+        final long byte3 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte3 >>> 6;
+        values[valuesOffset++] = (byte3 >>> 4) & 3;
+        values[valuesOffset++] = (byte3 >>> 2) & 3;
+        values[valuesOffset++] = byte3 & 3;
+        final long byte4 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte4 >>> 6;
+        values[valuesOffset++] = (byte4 >>> 4) & 3;
+        values[valuesOffset++] = (byte4 >>> 2) & 3;
+        values[valuesOffset++] = byte4 & 3;
+        final long byte5 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte5 >>> 6;
+        values[valuesOffset++] = (byte5 >>> 4) & 3;
+        values[valuesOffset++] = (byte5 >>> 2) & 3;
+        values[valuesOffset++] = byte5 & 3;
+        final long byte6 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte6 >>> 6;
+        values[valuesOffset++] = (byte6 >>> 4) & 3;
+        values[valuesOffset++] = (byte6 >>> 2) & 3;
+        values[valuesOffset++] = byte6 & 3;
+        final long byte7 = blocks[blocksOffset++] & 0xFF;
+        values[valuesOffset++] = byte7 >>> 6;
+        values[valuesOffset++] = (byte7 >>> 4) & 3;
+        values[valuesOffset++] = (byte7 >>> 2) & 3;
+        values[valuesOffset++] = byte7 & 3;
+      }
+    }
+
+    @Override
+    public void encode(int[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = ((values[valuesOffset++] & 0xffffffffL) << 62) | ((values[valuesOffset++] & 0xffffffffL) << 60) | ((values[valuesOffset++] & 0xffffffffL) << 58) | ((values[valuesOffset++] & 0xffffffffL) << 56) | ((values[valuesOffset++] & 0xffffffffL) << 54) | ((values[valuesOffset++] & 0xffffffffL) << 52) | ((values[valuesOffset++] & 0xffffffffL) << 50) | ((values[valuesOffset++] & 0xffffffffL) << 48) | ((values[valuesOffset++] & 0xffffffffL) << 46) | ((values[valuesOffset++] & 0xffffffffL) << 44) | ((values[valuesOffset++] & 0xffffffffL) << 42) | ((values[valuesOffset++] & 0xffffffffL) << 40) | ((values[valuesOffset++] & 0xffffffffL) << 38) | ((values[valuesOffset++] & 0xffffffffL) << 36) | ((values[valuesOffset++] & 0xffffffffL) << 34) | ((values[valuesOffset++] & 0xffffffffL) << 32) | ((values[valuesOffset++] & 0xffffffffL) << 30) | ((values[valuesOffset++] & 0xffffffffL) << 28) | ((values[valuesOffset++] & 0xffffffffL) << 26) | ((values[valuesOffset++] & 0xffffffffL) << 24) | ((values[valuesOffset++] & 0xffffffffL) << 22) | ((values[valuesOffset++] & 0xffffffffL) << 20) | ((values[valuesOffset++] & 0xffffffffL) << 18) | ((values[valuesOffset++] & 0xffffffffL) << 16) | ((values[valuesOffset++] & 0xffffffffL) << 14) | ((values[valuesOffset++] & 0xffffffffL) << 12) | ((values[valuesOffset++] & 0xffffffffL) << 10) | ((values[valuesOffset++] & 0xffffffffL) << 8) | ((values[valuesOffset++] & 0xffffffffL) << 6) | ((values[valuesOffset++] & 0xffffffffL) << 4) | ((values[valuesOffset++] & 0xffffffffL) << 2) | (values[valuesOffset++] & 0xffffffffL);
+      }
+    }
+
+    @Override
+    public void encode(long[] values, int valuesOffset, long[] blocks, int blocksOffset, int iterations) {
+      assert blocksOffset + iterations * blockCount() <= blocks.length;
+      assert valuesOffset + iterations * valueCount() <= values.length;
+      for (int i = 0; i < iterations; ++i) {
+        blocks[blocksOffset++] = (values[valuesOffset++] << 62) | (values[valuesOffset++] << 60) | (values[valuesOffset++] << 58) | (values[valuesOffset++] << 56) | (values[valuesOffset++] << 54) | (values[valuesOffset++] << 52) | (values[valuesOffset++] << 50) | (values[valuesOffset++] << 48) | (values[valuesOffset++] << 46) | (values[valuesOffset++] << 44) | (values[valuesOffset++] << 42) | (values[valuesOffset++] << 40) | (values[valuesOffset++] << 38) | (values[valuesOffset++] << 36) | (values[valuesOffset++] << 34) | (values[valuesOffset++] << 32) | (values[valuesOffset++] << 30) | (values[valuesOffset++] << 28) | (values[valuesOffset++] << 26) | (values[valuesOffset++] << 24) | (values[valuesOffset++] << 22) | (values[valuesOffset++] << 20) | (values[valuesOffset++] << 18) | (values[valuesOffset++] << 16) | (values[valuesOffset++] << 14) | (values[valuesOffset++] << 12) | (values[valuesOffset++] << 10) | (values[valuesOffset++] << 8) | (values[valuesOffset++] << 6) | (values[valuesOffset++] << 4) | (values[valuesOffset++] << 2) | values[valuesOffset++];
+      }
+    }
+
+}