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

Issue 797390 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Mar 2018
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

13.8%-91.8% regression in system_health.memory_mobile at 525845:525907

Project Member Reported by briander...@chromium.org, Dec 22 2017

Issue description

See the link to graphs below.
 
Project Member

Comment 1 by 42576172...@developer.gserviceaccount.com, Dec 22 2017

All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=797390

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=8f53964ae66c0b3fda67865d34cbf6e8b4a5c944d6020e040ab2b549f7f10464


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

android-nexus5
Project Member

Comment 3 by 42576172...@developer.gserviceaccount.com, Dec 22 2017

Cc: xing...@intel.com
Owner: xing...@intel.com
Status: Assigned (was: Untriaged)

=== Auto-CCing suspected CL author xing.xu@intel.com ===

Hi xing.xu@intel.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 : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_news/load_news_irctc
  Change       : 87.21% | 5538872.0 -> 10369520.0

Revision             Result                  N
chromium@525844      5538872 +- 538433       6      good
chromium@525876      5440568 +- 0.0          6      good
chromium@525892      5440568 +- 0.0          6      good
chromium@525900      5440568 +- 0.0          6      good
chromium@525902      5440568 +- 0.0          6      good
chromium@525903      10369508 +- 159348      6      bad       <--
chromium@525904      10391364 +- 150657      6      bad
chromium@525907      10369520 +- 160083      6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.news.irctc system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959471583357829552


For feedback, file a bug with component Speed>Bisection

Comment 4 by xing...@intel.com, Dec 22 2017

Cc: danakj@chromium.org
Does it because GLRenderer keeps too much RenderPass Textures?

Comment 5 by danakj@chromium.org, Dec 22 2017

It's not supposed to be keeping more than before. Can you reproduce it and debug?

Comment 7 by xing...@intel.com, Dec 23 2017

Cc: yang...@intel.com
Sure, I will look into it.
Project Member

Comment 8 by 42576172...@developer.gserviceaccount.com, Dec 23 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_media/load_media_google_images
  Change       : 39.68% | 1248312.0 -> 1743672.0

Revision             Result              N
chromium@525844      1248312 +- 0.0      6      good
chromium@525876      1248312 +- 0.0      6      good
chromium@525892      1248312 +- 0.0      6      good
chromium@525900      1248312 +- 0.0      6      good
chromium@525902      1248312 +- 0.0      6      good
chromium@525903      1743672 +- 0.0      6      bad       <--
chromium@525904      1743672 +- 0.0      6      bad
chromium@525907      1743672 +- 0.0      6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.media.google.images system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959463970168203792


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Dec 24 2017

Issue 797402 has been merged into this issue.

Comment 10 Deleted

Comment 11 Deleted

Project Member

Comment 16 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=https.www.google.co.uk.hl.en.q.science system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959090031346943808


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 19 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browser.social.instagram system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959089616086063136


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 20 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 1.36% | 32164136.0 -> 31727336.0

Revision             Result                   N
chromium@525874      32164136 +- 0.0          6      good
chromium@525900      32299759 +- 742836       6      good
chromium@525902      32164136 +- 0.0          6      good
chromium@525903      31795229 +- 371863       6      bad       <--
chromium@525904      31727336 +- 0.0          6      bad
chromium@525907      31470279 +- 1889652      6      bad
chromium@525913      31795066 +- 370972       6      bad
chromium@525926      31727336 +- 0.0          6      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.media.google.images system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959089482869333792


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 21 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 1.67% | 29634700.0 -> 29000076.0

Revision             Result                   N
chromium@525845      29634700 +- 0.0          6       good
chromium@525876      29495436 +- 1878774      14      good
chromium@525892      29170487 +- 3805417      21      good
chromium@525900      29634700 +- 0.0          9       good
chromium@525902      29216908 +- 2993366      14      good
chromium@525903      29000076 +- 1878774      14      bad       <--
chromium@525904      28706074 +- 2431698      9       bad
chromium@525907      29000076 +- 1878774      14      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.media.google.images system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959091669592733200


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 27 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : memory.top_10_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/foreground/https_www_google_co_uk_hl_en_q_science
  Change       : 34.99% | 1248312.0 -> 1685112.0

Revision             Result              N
chromium@525902      1248312 +- 0.0      6      good
chromium@525903      1685112 +- 0.0      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959077655126475456


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 33 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : memory.top_10_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/foreground
  Change       : 2.02% | 540523.2 -> 551467.2

Revision             Result                     N
chromium@525902      540523 +- 2.85158e-10      6      good
chromium@525903      551467 +- 2.85158e-10      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959076993248064016


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 37 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : memory.top_10_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/foreground


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068167053474880


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 38 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/browse_social


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068594393094528


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 39 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_media


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068572600866608


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 40 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_media


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068517978386816


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 41 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_news


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068548878537600


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 42 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068023559303712


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 43 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : memory.top_10_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/foreground


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959067286801459536


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 44 by 42576172...@developer.gserviceaccount.com, Dec 27 2017


=== BISECT JOB RESULTS ===
NO Perf regression found, tests failed to produce values

Bisect Details
  Configuration: linux_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_news


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

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

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959068005281813024


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 46 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : memory.top_10_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/foreground
  Change       : 3.40% | 1458027.2 -> 1507563.2

Revision             Result              N
chromium@525902      1458027 +- 0.0      6      good
chromium@525903      1507563 +- 0.0      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests memory.top_10_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959077641677053056


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 51 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_one_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_news/load_news_irctc
  Change       : 6.32% | 12240163.1111 -> 11466465.3333

Revision             Result                  N
chromium@525902      12240163 +- 958650      9      good
chromium@525903      11466465 +- 608926      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=load.news.irctc system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959002635999364160


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 52 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_news


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959087950503113296


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 53 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959087869847242288


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 54 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_webview_nexus6_aosp_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:webview:all_processes:reported_by_chrome:cc:effective_size_avg/browse_tools


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-webview --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959076769024081136


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 55 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 0.62% | 31992764.8889 -> 31795066.0

Revision             Result                   N
chromium@525902      31992765 +- 1924303      9      good
chromium@525903      31795066 +- 370972       6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959000330227971904


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 56 by 42576172...@developer.gserviceaccount.com, Dec 28 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 1.67% | 29634700.0 -> 29139340.0

Revision             Result               N
chromium@525902      29634700 +- 0.0      6      good
chromium@525903      29139340 +- 0.0      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959000852187351040


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 57 by 42576172...@developer.gserviceaccount.com, Dec 29 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8958999487427863072


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 58 by 42576172...@developer.gserviceaccount.com, Dec 29 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8959000611843114496


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 63 by 42576172...@developer.gserviceaccount.com, Dec 29 2017


=== BISECT JOB RESULTS ===
NO Perf regression found

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram

Revision             Result                     N
chromium@525902      173470438 +- 54464857      21      good
chromium@525903      170200954 +- 11207591      21      bad

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=browse.social.instagram system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8958895796775156848


For feedback, file a bug with component Speed>Bisection

Comment 64 Deleted

Comment 65 Deleted

Comment 66 by xing...@intel.com, Dec 29 2017

Possible reason: memory doesn't increase or decrease. It just moves from cc to gpu. (But I don't understand why no regression was observed on desktop like linux. Does it because 5.0 MiB is a very small memory for desktop? )

I draft a table to summarize the memory used by cc and gpu: 
https://docs.google.com/spreadsheets/d/1eBd7K7I6kbhZZ-OIU4iexrGyQDT-mPyTkctgEj1bpWQ/edit#

The data analysis: 
For most case (marked red and green), memory usage in gpu increases, while in cc, it decreases.  If we sum the red and green, no obvious changes observed. 

For the yellow cases, after bisecting them twice, two "NO Perf regression found", and another two "Perf regression found with culprit".  But I can not find the details result from dashboard (Also raised 
this question here: 
https://groups.google.com/a/chromium.org/forum/#!topic/chromium-dev/F6az9yN_d2o). Do you have any ideas about this?

The code analysis:
After review https://chromeperf.appspot.com/group_report?rev=525907 and the implementation of ResourceProvider&GLES2Implementation
 (They both implement base::trace_event::MemoryDumpProvider), it seems the reason is: After this cl
(https://crrev.com/9cc56b283e51de94cc373a77214da2bb646efcbd) landed, memory metrics for local resource is calculated as part of gpu. Before this cl, it is calculated as part of cc. That’s why cc:effective_size_avg decreased 5.0MiB while gpu:effective_size_avg increased 4.8MiB.
Example: 
memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_news/load_news_irctc  9.6% 5.0 MiB (-)
memory:chrome:all_processes:reported_by_chrome:gpu:effective_size_avg/load_news/load_news_irctc 91.8% 4.8 MiB (+)

@danakj, does above comments make sense?
Project Member

Comment 67 by 42576172...@developer.gserviceaccount.com, Dec 29 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 1.36% | 32164136.0 -> 31727336.0

Revision             Result               N
chromium@525902      32164136 +- 0.0      6      good
chromium@525903      31727336 +- 0.0      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8958895974009324288


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 68 by 42576172...@developer.gserviceaccount.com, Dec 29 2017


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

Suspected Commit
  Author : Xu Xing
  Commit : 9cc56b283e51de94cc373a77214da2bb646efcbd
  Date   : Fri Dec 22 02:20:01 2017
  Subject: [viz] Implement local resource in GLRenderer

Bisect Details
  Configuration: android_nexus5_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/load_media/load_media_google_images
  Change       : 4.96% | 29634700.0 -> 28164492.0

Revision             Result                   N
chromium@525902      29634700 +- 0.0          6      good
chromium@525903      28164492 +- 2387880      6      bad       <--

Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8958895694326944496


For feedback, file a bug with component Speed>Bisection
Project Member

Comment 69 by 42576172...@developer.gserviceaccount.com, Dec 30 2017


=== BISECT JOB RESULTS ===
Bisect failed for unknown reasons

Please contact the team (see below) and report the error.


Bisect Details
  Configuration: android_nexus5X_perf_bisect
  Benchmark    : system_health.memory_mobile
  Metric       : memory:chrome:all_processes:reported_by_chrome:cc:effective_size_avg/browse_social/browse_social_instagram


Please refer to the following doc on diagnosing memory regressions:
  https://chromium.googlesource.com/chromium/src/+/master/docs/memory-infra/memory_benchmarks.md

To Run This Test
  src/tools/perf/run_benchmark -v --browser=android-chromium --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests system_health.memory_mobile

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8958895640775088912


For feedback, file a bug with component Speed>Bisection
Thanks for your analysis! I agree that is what it appears is happening. The memory used to be reported by ResourceProvider but no longer does, so it is attributed to Gpu.

However, that does mean that we have this memory that is being left unattributed, and dumped in the gpu bucket now. Maybe DirectRenderer (or the subclasses if required) can implement MemoryDumpProvider so that this memory can be attributed to viz?
Issue 798390 has been merged into this issue.
Issue 798388 has been merged into this issue.

Comment 73 by xing...@intel.com, Jan 3 2018

OK, I will investigate the memory dump on DirectrRenderer.

Cc: perezju@chromium.org
xing.xu: great analysis! any progress on investigating the memory dump on DirectRenderer?

Also +perezju, can you or someone from memory infra review the analysis in #66?
Yes, explanation makes sense. And is consistent with alerts at:
https://chromeperf.appspot.com/group_report?rev=525903

Comment 76 by xing...@intel.com, Jan 25 2018

CL for memory dump on DirectRenderer is ongoing: 
https://chromium-review.googlesource.com/c/chromium/src/+/858479

Comment 77 by dtu@chromium.org, Jan 25 2018

 Issue 797395  has been merged into this issue.

Comment 78 by xing...@intel.com, Mar 27 2018

Hi,as we have reached a consensus, this isn’t memory regression, but memory was calculated from cc tag to gpu tag. And the new memory tag viz will be used instead(crbug.com/800238). So how about close this one and track it in the viz memory dump issue?
Yes can you do that?

Comment 80 by xing...@intel.com, Mar 28 2018

Status: WontFix (was: Assigned)
Mark this as WontFix, for further action, please refer to: crbug.com/800238.

Sign in to add a comment