New issue
Advanced search Search tips

Issue 764245 link

Starred by 3 users

Issue metadata

Status: WontFix
Owner:
Closed: Oct 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug-Regression



Sign in to add a comment

Performance regression on CSS transforms test cases in Spielzeugz

Reported by as...@opera.com, Sep 12 2017

Issue description

Performance regression on CSS transforms test case from   https://chromium.googlesource.com/chromium/src/+/4d3622e28d891b3a445275d132a6817bb83973ed (removing the bilerp-ignorning hack). Related to https://bugs.chromium.org/p/skia/issues/detail?id=6835

Chrome Version: 61.0.3156.0
OS: Linux x86

What steps will reproduce the problem?
(1) build content-shell (preferably non-debug build for best performance, I used args: 'dcheck_always_on=false is_debug=false remove_webcore_debug_symbols=true')
(2) run: ./content_shell --slow-down-raster-scale-factor=25 http://spielzeugz.de/html5/compare/css3-imgs-transform.html
(3) observe the reported fps count

What is the expected result?
~12-14 fps reported, same as in previous commit on master@8ccae6f5

What happens instead?
~4-6 fps reported


 

Comment 1 by as...@opera.com, Sep 12 2017

Description: Show this description

Comment 2 by f...@opera.com, Oct 12 2017

Components: Blink>Paint
Cc: -reed@chromium.org
Components: -Blink>Paint Internals>Skia
Labels: -Pri-3 ReleaseBlock-Stable M-63 Pri-1
Owner: fmalita@chromium.org
Status: Assigned (was: Untriaged)
There was another bug related to stuttering of transforms. This bug may also explain that issue.
Please add appropriate OSs.
Cc: reed@google.com
- prior to https://chromium.googlesource.com/chromium/src/+/4d3622e28d891b3a445275d132a6817bb83973ed, Skia was ignoring blierp filtering requests for subpixel translations, in software mode;  that is arguably incorrect, and not aligned with the GPU backend behavior.

- post change, Skia respects the filtering requests for subpixel positioned images and the regression reflects the cost of bilerp.

AFAICT, this is the correct behavior.

If the intent of the benchmark is to exercise filtering, then the new results are correct and expected.  If not, it should be updated to avoid subpixel positioning.

Unless you disagree, I'm inclined to close this as WontFix (WAI).
Project Member

Comment 6 by sheriffbot@chromium.org, Oct 13 2017

Cc: schenney@chromium.org
This issue is marked as a release blocker with no OS labels associated. Please add an appropriate OS label.

All release blocking issues should have OS labels associated to it, so that the issue can tracked and promptly verified, once it gets fixed.

Thanks for your time! To disable nags, add the Disable-Nags label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Labels: -ReleaseBlock-Stable
Status: WontFix (was: Assigned)

Comment 8 by as...@vewd.com, Oct 16 2017

Hi,

Thanks for this answer! With that a won't fix is fine with me.

Sign in to add a comment