Bug 77401
| Summary: | runner.js for performance test should use Bessel's correction | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Ryosuke Niwa <rniwa> |
| Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | Normal | CC: | abarth, eric, morrita, zoltan |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Bug Depends on: | 77074 | ||
| Bug Blocks: | 77037 | ||
Ryosuke Niwa
It's kind of bogus to use standard deviation for sample size smaller than 20 (for most tests). We should use some unbiased estimates for standard deviation. As a starter, we can use Bessel's correction.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Hajime Morrita
Also, I'd like to see stdev or some distribution metrics in the chart.
This is another story though.
Ryosuke Niwa
*** This bug has been marked as a duplicate of bug 98115 ***