Issue metadata
Sign in to add a comment
|
12% regression in main thread time on cc_scroll_200_layer_grid |
||||||||||||||||||||
Issue descriptionThere is an un-recovered regression in Renderer Main thread times on Android, approximately on Aug-01-2018. I'm suspecting that a regression was confounded with crbug.com/871347 (use-zoom-for-dsf on Android) which was reverted. Other thread times mostly recovered, but Main thread has not. Creating this issue to run some bisects. Associated graphs: https://chromeperf.appspot.com/report?sid=32cc1b277af84c8bc64e2c3b097b8e6d48c8e5e687d6be0f8ed2e9546145073b
,
Aug 31
I looked through some lists of adjacent patches and can't see anything obvious. Let's see what the pinpoint job says.
,
Sep 3
Pinpoint seems to be narrowing down on another smaller regression, but it looks to me the Main thread regression on Aug-01-2008 was due to the 'use-zoom-for-dsf on Android' change between r579539 and r579543. That means the remaining Main thread time regression is a later change.
,
Sep 3
📍 Found a significant difference after 1 commit. https://pinpoint-dot-chromeperf.appspot.com/job/13b5a48b640000 Enable use-zoom-for-dsf on Android on M70 by eirage@chromium.org https://chromium.googlesource.com/chromium/src/+/d0b8bbd2dbf9c06dcbbd01a4b1285cf7a1bc2ba0 1.879 → 2.115 (+0.2361) Understanding performance regressions: http://g.co/ChromePerformanceRegressions
,
Sep 3
Unassigning from eirage@ as her change was reverted, and will keep this issue open until we can understand what else regressed on the Main thread. |
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by 42576172...@developer.gserviceaccount.com
, Aug 31