All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=826819

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=e59bbe9ed78b4da4ac521bbc5269cfde700936169d85cb87d5fcc39567b5011f


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

android-nexus5
android-nexus5X
android-nexus6
android-webview-nexus5X
chromium-rel-win7-dual
chromium-rel-win7-x64-dual

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14d4f4bb440000

📍 Found significant differences after each of 2 commits.
https://pinpoint-dot-chromeperf.appspot.com/job/14d4f4bb440000

Revert "Use new Skia GPU read/write pixel code paths" by robertphillips@google.com
https://chromium.googlesource.com/chromium/src/+/56044665258df106f9c5dd353ed9662cab28aefd

Revert "Remove "Skipped" file and Port.skips_perf_test" by eyaich@chromium.org
https://chromium.googlesource.com/chromium/src/+/0d5e363089648f237f01341b28143fb27f5261f3

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

+simonhatch@chromium.org to triage pinpoint issue, looks like some jobs are failing with:

Traceback (most recent call last):
  File "/base/data/home/apps/s~chromeperf/pinpoint:dev-dtu-f2bde269.408767502487382406/dashboard/pinpoint/models/quest/execution.py", line 99, in Poll
    self._Poll()
  File "/base/data/home/apps/s~chromeperf/pinpoint:dev-dtu-f2bde269.408767502487382406/dashboard/pinpoint/models/quest/find_isolate.py", line 73, in _Poll
    self._CheckBuildStatus()
  File "/base/data/home/apps/s~chromeperf/pinpoint:dev-dtu-f2bde269.408767502487382406/dashboard/pinpoint/models/quest/find_isolate.py", line 103, in _CheckBuildStatus
    raise BuildError('Build failed: ' + status['build']['failure_reason'])
BuildError: Build failed: INVALID_BUILD_DEFINITION

+dtu

Is this related to the builder migration that's going on?

Yes,  issue 828778 . It should be mostly working again today. I'll re-run.

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/14eceb28c40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/14eceb28c40000

Revert "Use new Skia GPU read/write pixel code paths" by robertphillips@google.com
https://chromium.googlesource.com/chromium/src/+/56044665258df106f9c5dd353ed9662cab28aefd

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

robertphillips: bisect is showing a pretty clear memory jump at your CL. Can you take a look?

The CL identified by pinpoint was a revert due to a perf regression of the original CL.

I believe Brian has relanded the original CL w/ enhancements to reduce the perf impact.

It was reverted again and I haven't had a chance to get back to it.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6d0559c006da5625314490e3b5b522a7615113ba

commit 6d0559c006da5625314490e3b5b522a7615113ba
Author: Brian Salomon <bsalomon@google.com>
Date: Fri May 11 18:20:34 2018

Use Skia's new simplified read/write pixels code paths

Bug:  826819 
Change-Id: Ia6eaca75fe43f9d9ec94f8fe644a44e8e7436675
Reviewed-on: https://chromium-review.googlesource.com/1054520
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Brian Salomon <bsalomon@chromium.org>
Cr-Commit-Position: refs/heads/master@{#557943}
[modify] https://crrev.com/6d0559c006da5625314490e3b5b522a7615113ba/skia/config/SkUserConfig.h

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/835ae72c15a9702d56fd7936a83f020d9206118e

commit 835ae72c15a9702d56fd7936a83f020d9206118e
Author: Timothy Loh <timloh@chromium.org>
Date: Mon May 14 06:34:57 2018

Revert "Use Skia's new simplified read/write pixels code paths"

This reverts commit 6d0559c006da5625314490e3b5b522a7615113ba.

Reason for revert: Breaks ImageBitmapTest.AvoidGPUReadback on MSAN, e.g. https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.memory%2FLinux_MSan_Tests%2F9783%2F%2B%2Frecipes%2Fsteps%2Fwebkit_unit_tests%2F0%2Flogs%2FImageBitmapTest.AvoidGPUReadback%2F0

Original change's description:
> Use Skia's new simplified read/write pixels code paths
> 
> Bug:  826819 
> Change-Id: Ia6eaca75fe43f9d9ec94f8fe644a44e8e7436675
> Reviewed-on: https://chromium-review.googlesource.com/1054520
> Reviewed-by: Robert Phillips <robertphillips@google.com>
> Commit-Queue: Brian Salomon <bsalomon@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#557943}

TBR=bsalomon@chromium.org,robertphillips@google.com

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug:  826819 
Change-Id: Ia4dca1186fbb79569672fbf5cdabd9657c702762
Reviewed-on: https://chromium-review.googlesource.com/1056847
Reviewed-by: Timothy Loh <timloh@chromium.org>
Commit-Queue: Timothy Loh <timloh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#558215}
[modify] https://crrev.com/835ae72c15a9702d56fd7936a83f020d9206118e/skia/config/SkUserConfig.h

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d74706673fcd86bc4c6fd6ff880f104379602ca9

commit d74706673fcd86bc4c6fd6ff880f104379602ca9
Author: Brian Salomon <bsalomon@chromium.org>
Date: Tue May 15 16:03:40 2018

Reland "Use Skia's new simplified read/write pixels code paths"

This reverts commit 835ae72c15a9702d56fd7936a83f020d9206118e.

Reason for revert: Fix for MSAN issue landed here: https://skia.googlesource.com/skia/+/3d945e840c1955b5756fdeb696599c2818ed95fc

Original change's description:
> Revert "Use Skia's new simplified read/write pixels code paths"
> 
> This reverts commit 6d0559c006da5625314490e3b5b522a7615113ba.
> 
> Reason for revert: Breaks ImageBitmapTest.AvoidGPUReadback on MSAN, e.g. https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.memory%2FLinux_MSan_Tests%2F9783%2F%2B%2Frecipes%2Fsteps%2Fwebkit_unit_tests%2F0%2Flogs%2FImageBitmapTest.AvoidGPUReadback%2F0
> 
> Original change's description:
> > Use Skia's new simplified read/write pixels code paths
> > 
> > Bug:  826819 
> > Change-Id: Ia6eaca75fe43f9d9ec94f8fe644a44e8e7436675
> > Reviewed-on: https://chromium-review.googlesource.com/1054520
> > Reviewed-by: Robert Phillips <robertphillips@google.com>
> > Commit-Queue: Brian Salomon <bsalomon@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#557943}
> 
> TBR=bsalomon@chromium.org,robertphillips@google.com
> 
> # Not skipping CQ checks because original CL landed > 1 day ago.
> 
> Bug:  826819 
> Change-Id: Ia4dca1186fbb79569672fbf5cdabd9657c702762
> Reviewed-on: https://chromium-review.googlesource.com/1056847
> Reviewed-by: Timothy Loh <timloh@chromium.org>
> Commit-Queue: Timothy Loh <timloh@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#558215}

TBR=bsalomon@chromium.org,timloh@chromium.org,robertphillips@google.com

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug:  826819 
Change-Id: Idd8043535288d6f224531f90d381f85ca548c01d
Reviewed-on: https://chromium-review.googlesource.com/1059793
Reviewed-by: Brian Salomon <bsalomon@chromium.org>
Commit-Queue: Brian Salomon <bsalomon@chromium.org>
Cr-Commit-Position: refs/heads/master@{#558727}
[modify] https://crrev.com/d74706673fcd86bc4c6fd6ff880f104379602ca9/skia/config/SkUserConfig.h

The situation here is that I landed a change that decreased memory usage but regressed performance. When that was reverted this was filed because memory usage went back up. The change has since relanded but now we continue to favor performance over memory usage (a GL texture) in these scenarios. I believe this is the correct decision.