New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 671200 link

Starred by 1 user

Issue metadata

Status: Archived
Owner:
Closed: Dec 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug



Sign in to add a comment

Figure out some way of reliably detecting CrOS WebRTC regressions amid the noise

Project Member Reported by phoglund@chromium.org, Dec 5 2016

Issue description

We've had our first reports of webrtc.stress CPU, and we realized we basically have no idea what to do with the data. How do we bisect into 5k+ CLs? https://bugs.chromium.org/p/chromium/issues/detail?id=670707

We need to either find a more reliable metric than CPU or write some tools to find WebRTC rolls, Chromium <video> renderer changes, etc.
 
Summary: Figure out some way of reliably detecting CrOS WebRTC regressions amid the noise (was: Figure out some way of reliably detecting CrOS WebRTC regressions)
Owner: ehmaldonado@chromium.org
Cc: sullivan@chromium.org simonhatch@chromium.org
Components: Speed
I have a feeling we've improved here, but CC'ing some folks just in case. Can we close this?
Cc: kjellander@chromium.org
ehmaldonado, kjellander, how is this going?
AFAIK we have auto-bisection into WebRTC rolls for the bots in chromium.perf now, but I'm not sure it has actually been used in a real-world regression yet. Edward do you know?

Other than that, Edward has been working on improving our performance tests a lot the past year, so I'll leave it to him to comment here and possibly closing this bug as Fixed/WontFix.
Status: Fixed (was: Assigned)
I'm going to proactively close as fixed.
Status: Archived (was: Fixed)

Sign in to add a comment