test(ivy): fined tune micro benchmark parameters (#33341)

Decrease `MIN_SAMPLE_DURATION` to make it more likely that we cane fit into single time slice.
Increase `MIN_SAMPLE_COUNT_NO_IMPROVEMENT` to make it more likely to find the best

PR Close #33341
This commit is contained in:
Misko Hevery 2019-10-22 20:23:58 -07:00 committed by Andrew Kushnir
parent d40ee6a259
commit a1d7b6bb86
1 changed files with 2 additions and 2 deletions

View File

@ -8,10 +8,10 @@
const performance = require('perf_hooks').performance; const performance = require('perf_hooks').performance;
// Higher number here makes it more likely that we are more sure of the result. // Higher number here makes it more likely that we are more sure of the result.
const MIN_SAMPLE_COUNT_NO_IMPROVEMENT = 100; const MIN_SAMPLE_COUNT_NO_IMPROVEMENT = 500;
// A smaller number here means that we are coming too close on timer resultion, but it also means // A smaller number here means that we are coming too close on timer resultion, but it also means
// that it is less likely that we will be bothered by GC or preemptive multi tasking. // that it is less likely that we will be bothered by GC or preemptive multi tasking.
const MIN_SAMPLE_DURATION = 10; const MIN_SAMPLE_DURATION = 3;
const UNITS = ['ms', 'us', 'ns', 'ps']; const UNITS = ['ms', 'us', 'ns', 'ps'];
export interface Benchmark { export interface Benchmark {