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

Issue 759029 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Aug 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

35.6% regression in blink_perf.canvas at 496423:496550

Project Member Reported by kraynov@chromium.org, Aug 25 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Aug 25 2017

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=759029

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=467c7046b311579026f803a55f0852a0beb91f2013ec4613f326ff1b0f8eda78


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

chromium-rel-mac-retina
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Aug 26 2017

Cc: bsalo...@google.com
Owner: bsalo...@google.com
Status: Assigned (was: Untriaged)

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

Hi bsalomon@google.com, the bisect results pointed to your CL, please take a look at the
results.


=== BISECT JOB RESULTS ===
Perf regression found with culprit

Suspected Commit
  Author : Brian Salomon
  Commit : a184ac7e0cdb26919369151f606c194358c94ff5
  Date   : Tue Aug 22 21:36:40 2017
  Subject: Revert "Revert "Revert "Add GrTextureOp and use to implement SkGpuDevice::drawImage[Rect]() when possible"""

Bisect Details
  Configuration: mac_retina_perf_bisect
  Benchmark    : blink_perf.canvas
  Metric       : draw-static-canvas-2d-to-hw-accelerated-canvas-2d/draw-static-canvas-2d-to-hw-accelerated-canvas-2d
  Change       : 30.92% | 256942.053542 -> 177499.926331

Revision                             Result                 N
chromium@496422                      256942 +- 46065.5      6      good
chromium@496486                      237119 +- 62782.5      6      good
chromium@496518                      254179 +- 57916.6      6      good
chromium@496522                      255253 +- 80183.5      6      good
chromium@496522,skia@d6b28f7bc8      272407 +- 58251.9      6      good
chromium@496522,skia@82e314534c      254939 +- 89182.4      6      good
chromium@496522,skia@a184ac7e0c      171434 +- 50586.8      6      bad       <--
chromium@496523                      178464 +- 44358.1      6      bad
chromium@496524                      180709 +- 21968.9      6      bad
chromium@496526                      182620 +- 36332.8      6      bad
chromium@496534                      202281 +- 24568.8      6      bad
chromium@496550                      177500 +- 35797.6      6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests blink_perf.canvas

More information on addressing performance regressions:
  http://g.co/ChromePerformanceRegressions

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8970275446451954320


For feedback, file a bug with component Speed>Bisection

Comment 4 by bsalo...@google.com, Aug 26 2017

The regression is due to a revert of a change that improved performance for simple image draws. I'll be relanding that change once I add back an optimization to disable bilerp that the new code path omitted which caused some regressions on Android.

Comment 5 by bsalo...@google.com, Aug 30 2017

Status: Fixed (was: Assigned)
The change relanded and the perf is recovered.

Sign in to add a comment