New issue
Advanced search Search tips

Issue 668179 link

Starred by 1 user
Status: Fixed
Owner:
robertph...@google.com
Closed: Dec 2016
Cc:
nzolghadr@chromium.org
robertph...@google.com
bsalo...@google.com
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

14.6% regression in smoothness.gpu_rasterization.tough_filters_cases at 433630:433666

Project Member Reported by nzolghadr@chromium.org, Nov 23 2016 
See the link to graphs below.

Comment 1 by nzolghadr@chromium.org, Nov 23 2016 

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=668179

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg69_utAkM


Bot(s) for this bug's original alert(s):

android-one
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Nov 24 2016

Cc: robertph...@google.com
Owner: robertph...@google.com

=== Auto-CCing suspected CL author robertphillips@google.com ===

Hi robertphillips@google.com, the bisect results pointed to your CL below as possibly
causing a regression. Please have a look at this info and see whether
your CL be related.


===== BISECT JOB RESULTS =====
Status: completed


===== SUSPECTED CL(s) =====
Subject : Defer more renderTargetContexts in the GPU image filter paths - take 2
Author  : robertphillips
Commit description:
  
This is a reland of https://skia-review.googlesource.com/c/4767/ (Defer more renderTargetContexts in the GPU image filter paths).
The addition of guards on instantiation && accessRenderTarget failure should prevent a reoccurence of this Nexus7 failures.

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2514543002

TBR=bsalomon@google.com

Review-Url: https://codereview.chromium.org/2514543002
Commit  : d728f0c1a94fe926b59d8ebc9ae174019ccd3606
Date    : Mon Nov 21 19:05:03 2016


===== TESTED REVISIONS =====
Revision                         Mean     Std Dev  N   Good?
chromium@433629                  579.423  2158.99  85  good
chromium@433648                  573.933  1692.97  83  good
chromium@433653                  577.948  2189.02  85  good
chromium@433653,skia@14d2899145  579.137  2137.05  86  good
chromium@433653,skia@43c847b7f2  573.528  1960.58  84  good
chromium@433653,skia@d728f0c1a9  644.015  2298.03  74  bad    <--
chromium@433653,skia@99938a8ef2  642.145  1815.72  76  bad
chromium@433654                  641.06   2139.13  73  bad
chromium@433655                  647.3    2554.05  77  bad
chromium@433657                  643.327  2304.4   76  bad
chromium@433666                  638.616  2624.88  76  bad

Bisect job ran on: android_one_perf_bisect
Bug ID: 668179

Test Command: src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=http...rawgit.com.WebKit.webkit.master.PerformanceTests.Animometer.developer.html.test.interval.20.d smoothness.gpu_rasterization.tough_filters_cases
Test Metric: frame_times/http___rawgit.com_WebKit_webkit_master_PerformanceTests_Animometer_developer.html?test-interval_20_d
Relative Change: 10.22%

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/android_one_perf_bisect/builds/1825
Job details: https://chromeperf.appspot.com/buildbucket_job_status/8995187844066430272


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5855221329166336

| O O | Visit http://www.chromium.org/developers/speed-infra/perf-bug-faq
|  X  | for more information addressing perf regression bugs. For feedback,
| / \ | file a bug with component Tests>AutoBisect.  Thank you!
Project Member

Comment 4 by bugdroid1@chromium.org, Dec 1 2016 

The following revision refers to this bug:
  https://skia.googlesource.com/skia.git/+/155b29224d64b5c12bf7fc789edb5c257b6281c8

commit 155b29224d64b5c12bf7fc789edb5c257b6281c8
Author: Robert Phillips <robertphillips@google.com>
Date: Thu Dec 01 17:57:20 2016

Remove use of makeDeferredRenderTargetContextWithFallback

This is a partial revert of https://codereview.chromium.org/2514543002 (Defer more renderTargetContexts in the GPU image filter paths - take 2)

I have been unable to reproduce the performance regression in  crbug.com/668179  locally so intend to revert the above CL piecemeal.

BUG= 668179 

Change-Id: Idf3d66d71c452f5718d7b1d8d945ca7ff46b75d8
Reviewed-on: https://skia-review.googlesource.com/5441
Commit-Queue: Robert Phillips <robertphillips@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>

[modify] https://crrev.com/155b29224d64b5c12bf7fc789edb5c257b6281c8/src/effects/SkAlphaThresholdFilter.cpp
[modify] https://crrev.com/155b29224d64b5c12bf7fc789edb5c257b6281c8/src/effects/SkBlurMaskFilter.cpp
[modify] https://crrev.com/155b29224d64b5c12bf7fc789edb5c257b6281c8/src/gpu/GrBlurUtils.cpp
Project Member

Comment 5 by bugdroid1@chromium.org, Dec 1 2016 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7b8f77623038030d4997e21048ae009a1e3811dd

commit 7b8f77623038030d4997e21048ae009a1e3811dd
Author: skia-deps-roller <skia-deps-roller@chromium.org>
Date: Thu Dec 01 22:48:31 2016

Roll src/third_party/skia/ 71b762f2a..97aadfce6 (23 commits).

https://skia.googlesource.com/skia.git/+log/71b762f2aca1..97aadfce654c

$ git log 71b762f2a..97aadfce6 --date=short --no-merges --format='%ad %ae %s'
2016-12-01 msarett Update skimage version, test CMYK images
2016-12-01 msarett Revert "SkColorSpaceXform bug fixes"
2016-12-01 mtklein Avoid creating std::function in run_pipeline().
2016-12-01 raftias Added CMYK support for ICC profiles.
2016-12-01 brianosman Revert "Add RasterPipeline implementation for SkColorSpaceXform"
2016-12-01 robertphillips Remove use of makeDeferredRenderTargetContextWithFallback
2016-12-01 brianosman Revert of Enable sRGB on iOS, make sRGB decode support optional (patchset #11 id:200001 of https://codereview.chromium.org/2539993002/ )
2016-12-01 msarett Add RasterPipeline implementation for SkColorSpaceXform
2016-12-01 robertphillips Add animating blur image filter GM/slide/bench
2016-12-01 fmalita SkNoDrawCanvas - a public non-drawing canvas base class
2016-11-29 msarett Add srgb and f16 modes to fiddle
2016-12-01 halcanary third_party/ktx: put WriteBitmapToKTX back
2016-12-01 reed Revert "Revert "Revert "remove (empty) SkXfermode.h"""
2016-12-01 brianosman Two (related) changes here:
2016-12-01 bsalomon Rename GrDrawBatch->GrDrawOp
2016-12-01 mtklein Revert "Added CMYK support for ICC profiles."
2016-11-30 raftias Added CMYK support for ICC profiles.
2016-11-30 mtklein move all memset() logic into blitter
2016-12-01 bsalomon Rename GrBatch to GrOp
2016-12-01 bsalomon Remove pipeline info dump from GrDrawBatch
2016-12-01 reed Revert "Revert "remove (empty) SkXfermode.h""
2016-12-01 halcanary No SkEncodeImageAsKTX for ANDROID_FRAMEWORK
2016-11-30 msarett SkColorSpaceXform bug fixes

BUG= 668179 , 668925 

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
TBR=brianosman@google.com

Review-Url: https://codereview.chromium.org/2545913002
Cr-Commit-Position: refs/heads/master@{#435746}

[modify] https://crrev.com/7b8f77623038030d4997e21048ae009a1e3811dd/DEPS
Project Member

Comment 6 by bugdroid1@chromium.org, Dec 1 2016 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7b8f77623038030d4997e21048ae009a1e3811dd

commit 7b8f77623038030d4997e21048ae009a1e3811dd
Author: skia-deps-roller <skia-deps-roller@chromium.org>
Date: Thu Dec 01 22:48:31 2016

Roll src/third_party/skia/ 71b762f2a..97aadfce6 (23 commits).

https://skia.googlesource.com/skia.git/+log/71b762f2aca1..97aadfce654c

$ git log 71b762f2a..97aadfce6 --date=short --no-merges --format='%ad %ae %s'
2016-12-01 msarett Update skimage version, test CMYK images
2016-12-01 msarett Revert "SkColorSpaceXform bug fixes"
2016-12-01 mtklein Avoid creating std::function in run_pipeline().
2016-12-01 raftias Added CMYK support for ICC profiles.
2016-12-01 brianosman Revert "Add RasterPipeline implementation for SkColorSpaceXform"
2016-12-01 robertphillips Remove use of makeDeferredRenderTargetContextWithFallback
2016-12-01 brianosman Revert of Enable sRGB on iOS, make sRGB decode support optional (patchset #11 id:200001 of https://codereview.chromium.org/2539993002/ )
2016-12-01 msarett Add RasterPipeline implementation for SkColorSpaceXform
2016-12-01 robertphillips Add animating blur image filter GM/slide/bench
2016-12-01 fmalita SkNoDrawCanvas - a public non-drawing canvas base class
2016-11-29 msarett Add srgb and f16 modes to fiddle
2016-12-01 halcanary third_party/ktx: put WriteBitmapToKTX back
2016-12-01 reed Revert "Revert "Revert "remove (empty) SkXfermode.h"""
2016-12-01 brianosman Two (related) changes here:
2016-12-01 bsalomon Rename GrDrawBatch->GrDrawOp
2016-12-01 mtklein Revert "Added CMYK support for ICC profiles."
2016-11-30 raftias Added CMYK support for ICC profiles.
2016-11-30 mtklein move all memset() logic into blitter
2016-12-01 bsalomon Rename GrBatch to GrOp
2016-12-01 bsalomon Remove pipeline info dump from GrDrawBatch
2016-12-01 reed Revert "Revert "remove (empty) SkXfermode.h""
2016-12-01 halcanary No SkEncodeImageAsKTX for ANDROID_FRAMEWORK
2016-11-30 msarett SkColorSpaceXform bug fixes

BUG= 668179 , 668925 

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
TBR=brianosman@google.com

Review-Url: https://codereview.chromium.org/2545913002
Cr-Commit-Position: refs/heads/master@{#435746}

[modify] https://crrev.com/7b8f77623038030d4997e21048ae009a1e3811dd/DEPS
Project Member

Comment 7 by bugdroid1@chromium.org, Dec 2 2016 

The following revision refers to this bug:
  https://skia.googlesource.com/skia.git/+/51d77ffdc635b3208350262a3edea8d685453583

commit 51d77ffdc635b3208350262a3edea8d685453583
Author: Robert Phillips <robertphillips@google.com>
Date: Fri Dec 02 19:56:45 2016

Second part of piecemeal revert

This is a partial revert of https://codereview.chromium.org/2514543002 (Defer more renderTargetContexts in the GPU image filter paths - take 2)

I have been unable to reproduce the performance regression in  crbug.com/668179  locally so intend to revert the above CL piecemeal.

BUG= 668179 

Change-Id: Iee9d0164f85ae33ff8dfa9b3eb01c07fd825f017
Reviewed-on: https://skia-review.googlesource.com/5541
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>

[modify] https://crrev.com/51d77ffdc635b3208350262a3edea8d685453583/src/core/SkBlurImageFilter.cpp
Project Member

Comment 8 by bugdroid1@chromium.org, Dec 2 2016 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7697725c4bb455dbe58d3256604ee988e1f05b18

commit 7697725c4bb455dbe58d3256604ee988e1f05b18
Author: skia-deps-roller <skia-deps-roller@chromium.org>
Date: Fri Dec 02 23:13:33 2016

Roll src/third_party/skia/ 439ace939..51d77ffdc (3 commits).

https://skia.googlesource.com/skia.git/+log/439ace939b6a..51d77ffdc635

$ git log 439ace939..51d77ffdc --date=short --no-merges --format='%ad %ae %s'
2016-12-02 robertphillips Second part of piecemeal revert
2016-12-02 mtklein Tricky float -> byte conversion in store_8888.
2016-12-01 msarett Refine constant sRGB and Adobe RGB to D50 matrices

BUG= 668179 

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
TBR=brianosman@google.com

Review-Url: https://codereview.chromium.org/2548583005
Cr-Commit-Position: refs/heads/master@{#436073}

[modify] https://crrev.com/7697725c4bb455dbe58d3256604ee988e1f05b18/DEPS

Comment 9 by robertphillips@chromium.org, Dec 5 2016 

https://skia-review.googlesource.com/c/5541/ (Second part of piecemeal revert) rolled into Chrome at 436073. 

It seems to completely "fix" the regression on the Mac Retina & Win7-gpu bots.
It looks like it "fixes" most of the regression on the Android1 bot.

Comment 10 by robertphillips@chromium.org, Dec 6 2016

Cc: bsalo...@google.com
Status: Fixed (was: Untriaged)
https://skia-review.googlesource.com/c/5573/ (Revert "Remove use of makeDeferredRenderTargetContextWithFallback") rolled into Chrome at 436325 with no perf regression on the charts so, makeDeferredRenderTargetContextWithFallback is exonerated.

Marking as fixed at this point. We can re-open, or start a new bug, if/when we defer Skia's image filters at the root.
►

Sign in to add a comment