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

Issue 661091 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 2
Type: Bug-Regression



Sign in to add a comment

16.8% regression in smoothness.top_25_smooth/mean_frame_time on cros-veyron_minnie at 29040000894600001:29050000894700000

Project Member Reported by hctsai@chromium.org, Nov 1 2016

Issue description

Performance dashboard identified a 16.8% regression in smoothness.top_25_smooth/mean_frame_time on cros-veyron_minnie at revision range 29040000894600001:29050000894700000. Graph: https://chromeperf.appspot.com/report?masters=ChromeOSPerf&bots=cros-veyron_minnie&tests=smoothness.top_25_smooth%2Fmean_frame_time&checked=mean_frame_time%2Cmean_frame_time_ref%2Cref&rev=29050000894700000


This happens on many boards and looks like a real regression not flaky results. Trying to run bisect tool between Chrome 56.0.2904.0 and  56.0.2905.0.
 
Bisect Chrome between 56.0.2904.0 and 56.0.2905.0 on both ChromeOS 8946.0.0 and 8947.0.0. All the Chrome changes during bisection on 8946.0.0 reports good results while all the Chrome changes during bisection on ChromeOS 8947.0.0 reports bad results. I suspect the cause is from ChromeOS changes.

Comment 2 by cmt...@chromium.org, Nov 11 2016

Just double checked to see if this might be due to the compiler shift from GCC to LLVM (which happened for veyron_minnie on Oct. 31, around 5:30pm PDT).  According to the CL, the compiler change started with Chrome version 8952, so this problem was occurring before the compiler was switched.
Components: Internals>GPU>Metrics

Comment 4 by sadrul@chromium.org, Mar 29 2018

Cc: wutao@chromium.org
Owner: cywang@chromium.org
Status: Assigned (was: Untriaged)
--> cywang@ for triage (pretty old bug though, so maybe WontFix?)

Sign in to add a comment