Issue metadata
Sign in to add a comment
|
3.6%-23.1% regression in webrtc_perf_tests at 18449:18449 |
||||||||||||||||||||
Issue descriptionFairly clear change in the graphs, but the candidate Chromium roll doesn't contain any libvpx or compiler changes, or anything else suspicious. Could this be a device/infra change? Mostly improvements, so I don't think it is worth spending too much time on it.
,
Jun 16 2017
It looks similar to https://bugs.chromium.org/p/chromium/issues/detail?id=733554. Chromium recently switched from gcc to clang for android (https://chromium.googlesource.com/chromium/src/build/+/5a7ba0fdfa1c5c98291d5a66f5d2a085051de215) and this can cause a perf regression (see crbug.com/481675 ). I think this is the root cause. Let me know if you want more investigation.
,
Jun 16 2017
I don't think that's the case here. The roll doesn't appear to contain the gcc -> clang change. However, I agree that it isn't worth spending too much time on. |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by terelius@chromium.org
, Jun 16 2017