Bug 86302
| Summary: | [Chromium] REGRESSION(r116825): 15% regressions on Dromaeo/dromaeo-3d-cube due to Chromium DEPS roll | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Ryosuke Niwa <rniwa> |
| Component: | WebKit Misc. | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | Normal | CC: | enne, jamesr, morrita, ojan, tony |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Ryosuke Niwa
Our score on Dromaeo-3d-cube regressed 15% on Chromium Linux after my DEPS roll:
http://webkit-perf.appspot.com/graph.html#tests=[[42011,2001,173262]]&sel=1336768884920.8948,1336828569131.4211,997.3568281938326,1494.273127753304&displayrange=7&datatype=running
The regression range is r136044 to r136884 (840 revision :( ). However, with this kind of regression, you might be able to bi-sect by running the test locally.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Ojan Vafai
Is anyone looking into this? Whoever was gardener at this time is basically on the hook for it.
Ryosuke Niwa
It appears that this has been fixed :)
http://webkit-perf.appspot.com/graph.html#tests=[[42011,2001,173262]]&sel=1337047933042.644,1337077479602.9631,911.3924050632911,1500&displayrange=7&datatype=running
Hajime Morrita
(In reply to comment #2)
> It appears that this has been fixed :)
>
> http://webkit-perf.appspot.com/graph.html#tests=[[42011,2001,173262]]&sel=1337047933042.644,1337077479602.9631,911.3924050632911,1500&displayrange=7&datatype=running
The unit of this test is runs/sec. So It looks what you are pointing
is a slow down. not speed up.
Actually the original regression at 12th looks
having been fixed once, and taken back again around 15th.