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

Issue 697947 link

Starred by 4 users

Issue metadata

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



Sign in to add a comment

277% regression in smoothness.top_25_smooth at 453486:453572

Project Member Reported by sullivan@chromium.org, Mar 2 2017

Issue description

See the link to graphs below.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=697947

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgpJm87woM


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

win-zenbook

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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@453485      6.37819 +- 7.37148      21      good
chromium@453572      6.338 +- 5.76373        21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8986211418608630320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4509310131896320


| 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 Speed>Bisection.  Thank you!

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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                 N
chromium@453485      8.4071 +- 23.6235      21      good
chromium@453572      6.83105 +- 7.6918      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8986211418608630320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4509310131896320


| 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 Speed>Bisection.  Thank you!

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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@453485      7.17519 +- 8.81775      21      good
chromium@453572      9.70605 +- 31.1307      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8986211418608630320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4509310131896320


| 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 Speed>Bisection.  Thank you!
Project Member

Comment 9 by 42576172...@developer.gserviceaccount.com, Mar 10 2017


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

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : mean_input_event_latency/LinkedIn

Revision             Result                  N
chromium@453563      8.68033 +- 2.36246      21      good
chromium@453630      8.58781 +- 2.10531      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8986211418608630320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4509310131896320


| 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 Speed>Bisection.  Thank you!
Labels: Performance-Responsiveness
Project Member

Comment 12 by 42576172...@developer.gserviceaccount.com, Mar 25 2017


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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@453485      6.44314 +- 8.9414       21      good
chromium@453572      6.54971 +- 8.60899      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8984197168313711744

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4509310131896320


| 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 Speed>Bisection.  Thank you!
Cc: rogerm@chromium.org
Owner: rogerm@chromium.org

=== Auto-CCing suspected CL author rogerm@chromium.org ===

Hi rogerm@chromium.org, 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 : rogerm
  Commit : 2233a4a7cb12ac4acdebd33491b528f8fb670f02
  Date   : Mon Feb 27 23:11:16 2017
  Subject: [translate] Add translate ranker model loader.

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn
  Change       : 199.98% | 2.059 -> 6.17666666667

Revision             Result                   N
chromium@453165      2.059 +- 0.148795        6      good
chromium@453369      2.3555 +- 1.0912         6      good
chromium@453376      2.11567 +- 0.156676      6      good
chromium@453377      2.0875 +- 0.247995       6      good
chromium@453378      7.4625 +- 2.8726         6      bad       <--
chromium@453379      6.67183 +- 6.15603       6      bad
chromium@453382      6.69783 +- 1.67909       6      bad
chromium@453395      6.68067 +- 3.78955       6      bad
chromium@453420      6.69417 +- 2.80717       6      bad
chromium@453471      6.5445 +- 3.81109        6      bad
chromium@453572      6.17667 +- 3.2877        6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8980045923982576064

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=4963796403093504


| 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 Speed>Bisection.  Thank you!
The CL in question has already been reverted.

See Commit: 9ea997e1d7bbd4350d8793e1629f33c98d00ce2b

The course of this was tracked in crbug/646711
Cc: vmi...@chromium.org
rogem@ did anything related to this patch re-land?

I can see the LinkedIn first_gesture_scroll_update_latency graph recovered on Mar 02, but then regressed again and hasn't recovered.
#17 is a bisect range after the revert commit in #15.
Is the graph viewable somewhere? on what day did it regress again?
The graph links are in the first post.  See the graph in https://chromeperf.appspot.com/group_report?bug_id=697947

There is an increase in 'ChromiumPerf/win-zenbook/smoothness.top_25_smooth / first_gesture_scroll_update_latency' from around ~2ms to ~8ms on your change, then a drop on the revert, then back up to ~8ms soon after.
Project Member

Comment 21 by 42576172...@developer.gserviceaccount.com, May 10 2017

Cc: hamelphi@chromium.org
Owner: hamelphi@chromium.org

=== Auto-CCing suspected CL author hamelphi@chromium.org ===

Hi hamelphi@chromium.org, 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 : hamelphi
  Commit : d005809c83c0ad0b9cbbda6b10b7ed2c3dcd5120
  Date   : Thu Mar 02 23:38:24 2017
  Subject: [Translate] Add translate ranker model loader.

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn
  Change       : 216.70% | 2.238 -> 7.08766666667

Revision             Result                    N
chromium@454302      2.238 +- 0.956849         6      good
chromium@454375      2.08067 +- 0.155741       6      good
chromium@454411      2.07817 +- 0.373          6      good
chromium@454429      2.03317 +- 0.144454       6      good
chromium@454432      2.01033 +- 0.0834466      6      good
chromium@454433      7.20367 +- 5.56678        6      bad       <--
chromium@454434      6.27633 +- 3.93272        6      bad
chromium@454438      7.46417 +- 4.05933        6      bad
chromium@454447      7.01533 +- 1.6772         6      bad
chromium@454592      7.08767 +- 4.51732        6      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8980030082935079104

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5882852228464640


| 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 Speed>Bisection.  Thank you!
Cc: benhenry@chromium.org
It looks like this was reverted and landed multiple times, per Issue 646711.

benhenry@ this still looks like a regression in telemetry.
+yyushkina
+hamelphi

Looking at the history of that change...

Landed:     2017-02-27 (Bisect suggest culprit)
Reverted:   2017-03-02 (Seems to go back down)
Re-Landed:  2017-03-02 (Seems to go back up)
Disable:    2017-03-03 (Removed from field trial testing, no apparent change)
Reverted:   2017-03-07 (no apparent change)
Re-Landed:  2017-04-05 (refactored, added to field trials, no apparent change)

See also https://bugs.chromium.org/p/chromium/issues/detail?id=697665

 
Cc: yyushkina@chromium.org
There's a temporary drop that coincides with the revert of 2017-03-07. Running a bisect on that.

Hmmm... I'm assuming that bisect will also point out changes where metrics go down instead of up. Is that a valid assumption?
I've got a CL where I've tried:
  - Patchset 1 - disabling the fieldtrial config
  - Patchset 2 - delaying the start of load to first use

https://codereview.chromium.org/2871843003

PTAL?

If I read the results correctly (navigate to the HTML results in the perf_bisect bot output)... they seem to suggest that there's a perf gain to be had in delaying the start of load to be on-demand at first use. The gain is of similar to completely disabling the feature.

Note: We did these tests prior to landing, but we weren't aware of the regression covered by this bug, rather the ones raised in the previously filed bugs, for which we didn't observe recurrent regressions.

Project Member

Comment 27 by 42576172...@developer.gserviceaccount.com, May 10 2017


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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@455180      8.20614 +- 41.614       21      good
chromium@455379      6.02581 +- 7.34709      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8980021392878552304

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5863891357138944


| 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 Speed>Bisection.  Thank you!
Project Member

Comment 29 by 42576172...@developer.gserviceaccount.com, May 10 2017


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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@455180      7.24748 +- 15.9982      21      good
chromium@455379      6.24729 +- 9.154        21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8980012916950421568

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5863891357138944


| 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 Speed>Bisection.  Thank you!
Owner: sullivan@chromium.org
Status: Assigned (was: Untriaged)
It's difficult to say if this is telemetry or the lab as the ref builds aren't being displayed. Annie - any ideas what's going on here?
Owner: rogerm@chromium.org
Assigning to rogerm as he's actively working on this in 
https://codereview.chromium.org/2871843003/

vmiura: See #26: can you review from the perspective of answering the questions there?
Is there a way to run all of the telemetry tests ahead of submitting?
Re #32, see the "perf try job" section here: https://www.chromium.org/developers/telemetry/performance-try-bots.

This isn't LinkedIn only.

We see regressions in LinkedIn, Wordpress, booking.com, ebay.com, etc.
We see an improvement in Wikipedia, it looks like.

Because this only effects some sites, I doubt this is due to the lab.
Project Member

Comment 34 by bugdroid1@chromium.org, May 11 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d2876f4ee1c1a01ee262ca55752503ba3a30a6d5

commit d2876f4ee1c1a01ee262ca55752503ba3a30a6d5
Author: rogerm <rogerm@chromium.org>
Date: Thu May 11 22:16:40 2017

[translate] Defer start of translate ranker model load to first use.

This CL is in response to perf regression (see referenced bugs).

BUG= 697947 ,  697665 

Review-Url: https://codereview.chromium.org/2871843003
Cr-Commit-Position: refs/heads/master@{#471086}

[modify] https://crrev.com/d2876f4ee1c1a01ee262ca55752503ba3a30a6d5/components/translate/core/browser/translate_ranker_impl.cc

Where and how soon can we see graphs for the impact (or not) of this?
Owner: sullivan@chromium.org
I don't think any lingering regression is a result of the translate ranker CLs.

Looking at the perf graphs at (https://chromeperf.appspot.com/group_report?bug_id=697947) we can see:

- a regression when the initial CL was landed at r453378 on 2017-02-27
  https://codereview.chromium.org/2565873002

- recovery on revert at r454263 on 2017-03-01
  https://codereview.chromium.org/2726043003

- some small background uptick/regression

- a regression on initial reland at r454433 on 2017-03-01
  https://codereview.chromium.org/2726203002
  
- recovery on revert at r455280 on 2017-03-07
  https://codereview.chromium.org/2736853004

- additional gradual regression in the background

- no change on the fixed reland at r462038  on 2017-04-05
  https://codereview.chromium.org/2785493004

- no change on defer post first load at r471086 on 2017-05-11
  https://codereview.chromium.org/2871843003

Was this re-landed after the revert at r455280 on 2017-03-07?

I kicked off some bisects after that revert over the next few revs to see if anything shows up as causing a regression.
Project Member

Comment 40 by 42576172...@developer.gserviceaccount.com, May 15 2017


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

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : mean_input_event_latency/LinkedIn

Revision             Result                  N
chromium@455281      8.80505 +- 2.28195      21      good
chromium@455962      8.53624 +- 3.16647      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8979491007514627392

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5822233294929920


| 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 Speed>Bisection.  Thank you!
Project Member

Comment 42 by 42576172...@developer.gserviceaccount.com, May 16 2017


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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                  N
chromium@455281      11.4151 +- 68.7947      21      good
chromium@455917      6.68605 +- 9.53383      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8979490964413062560

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6466804437745664


| 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 Speed>Bisection.  Thank you!
Project Member

Comment 44 by 42576172...@developer.gserviceaccount.com, May 16 2017


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

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : mean_input_event_latency/LinkedIn

Revision             Result                  N
chromium@455281      8.37833 +- 3.74543      21      good
chromium@455962      8.68395 +- 2.12567      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8979484434451026320

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5822233294929920


| 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 Speed>Bisection.  Thank you!
Project Member

Comment 45 by 42576172...@developer.gserviceaccount.com, May 16 2017


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

Bisect Details
  Configuration: winx64_zen_perf_bisect
  Benchmark    : smoothness.top_25_smooth
  Metric       : first_gesture_scroll_update_latency/LinkedIn

Revision             Result                 N
chromium@455281      8.3219 +- 24.7979      21      good
chromium@455917      8.01538 +- 16.441      21      bad

To Run This Test
  src/tools/perf/run_benchmark -v --browser=release_x64 --output-format=chartjson --upload-results --pageset-repeat=1 --also-run-disabled-tests --story-filter=LinkedIn smoothness.top_25_smooth

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8979480657873260432

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=6466804437745664


| 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 Speed>Bisection.  Thank you!
re: c39

sullivan@ asked:

> Was this re-landed after the revert at r455280 on 2017-03-07?

It was relanded about a month later at r462038 on 2017-04-05 with no particular observed change in telemetry.

https://codereview.chromium.org/2785493004
Project Member

Comment 47 by bugdroid1@chromium.org, Jun 9 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cd4a0934f93239d0791e80d7a5da5b81eea368d9

commit cd4a0934f93239d0791e80d7a5da5b81eea368d9
Author: hamelphi <hamelphi@chromium.org>
Date: Fri Jun 09 19:45:22 2017

Revert of [translate] Defer start of translate ranker model load to first use. (patchset #2 id:40001 of https://codereview.chromium.org/2871843003/ )

Reason for revert:
The perf regression was not caused by this piece of code. Reverting to the original behavior.

See https://docs.google.com/document/d/1eeTSJg18rJ4v_oxxSJU8MK20UywnEhocWk_WrWv8kQE/edit for details of the regression investigation.

Original issue's description:
> [translate] Defer start of translate ranker model load to first use.
>
> This CL is in response to perf regression (see referenced bugs).
>
> BUG= 697947 ,  697665 
>
> Review-Url: https://codereview.chromium.org/2871843003
> Cr-Commit-Position: refs/heads/master@{#471086}
> Committed: https://chromium.googlesource.com/chromium/src/+/d2876f4ee1c1a01ee262ca55752503ba3a30a6d5

TBR=benhenry@chromium.org,vmiura@chromium.org,sullivan@chromium.org,tdresser@chromium.org,groby@chromium.org,rogerm@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 697947 ,  697665 

Review-Url: https://codereview.chromium.org/2918173002
Cr-Commit-Position: refs/heads/master@{#478379}

[modify] https://crrev.com/cd4a0934f93239d0791e80d7a5da5b81eea368d9/components/translate/core/browser/translate_ranker_impl.cc

Labels: Merge-Request-60
We ask for approval to merge   https://chromium.googlesource.com/chromium/src.git/+/cd4a0934f93239d0791e80d7a5da5b81eea368d9
 in M60
 TranslateRanker is currently enabled in Query mode in Beta, we'd like to have the model load at startup rather than first use in order to provide a more consistent experience to the users.

After a more thorough investigation, we have found out that Loading the TranslateRanker model at startup was not the cause of the regression mentioned in this bug. The cause is still under investigation, but we have not found a significant difference in first_gesture_scroll_update_latency when running perfbots several times and comparing with a control.
See https://docs.google.com/document/d/1eeTSJg18rJ4v_oxxSJU8MK20UywnEhocWk_WrWv8kQE/edit#heading=h.b12mdf4rjsun for more details on the ongoing investigation, speciffically the "Load model at startup" links for this specific metric.
Project Member

Comment 50 by sheriffbot@chromium.org, Jun 15 2017

Labels: -Merge-Request-60 Hotlist-Merge-Review Merge-Review-60
This bug requires manual review: Reverts referenced in bugdroid comments after merge request.
Please contact the milestone owner if you have questions.
Owners: amineer@(Android), cmasso@(iOS), josafat@(ChromeOS), bustamante@(Desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Please add appropriate OSs.  Thanks.
Labels: OS-All
Thanks for fixing this - can you please confirm exactly what will be merged to M60? Has this been well tested in canary/dev?
We want to merge https://chromium.googlesource.com/chromium/src.git/+/cd4a0934f93239d0791e80d7a5da5b81eea368d9 which allows TranslateRanker to initiate loading of its model at startup rather than on first use. This has been tested in Canary and Dev.
Labels: -Merge-Review-60 Merge-Approved-60
Approving merge to M60. 
Project Member

Comment 56 by sheriffbot@chromium.org, Jun 26 2017

Cc: abdulsyed@chromium.org
This issue has been approved for a merge. Please merge the fix to any appropriate branches as soon as possible!

If all merges have been completed, please remove any remaining Merge-Approved labels from this issue.

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
Project Member

Comment 57 by bugdroid1@chromium.org, Jun 27 2017

Labels: -merge-approved-60 merge-merged-3112
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ac979d32d6ce857f746f44bb009b180676b3114a

commit ac979d32d6ce857f746f44bb009b180676b3114a
Author: Roger McFarlane <rogerm@chromium.org>
Date: Tue Jun 27 17:40:21 2017

Revert of [translate] Defer start of translate ranker model load to first use. (patchset #2 id:40001 of https://codereview.chromium.org/2871843003/ )

Reason for revert:
The perf regression was not caused by this piece of code. Reverting to the original behavior.

See https://docs.google.com/document/d/1eeTSJg18rJ4v_oxxSJU8MK20UywnEhocWk_WrWv8kQE/edit for details of the regression investigation.

Original issue's description:
> [translate] Defer start of translate ranker model load to first use.
>
> This CL is in response to perf regression (see referenced bugs).
>
> BUG= 697947 ,  697665 
>
> Review-Url: https://codereview.chromium.org/2871843003
> Cr-Commit-Position: refs/heads/master@{#471086}
> Committed: https://chromium.googlesource.com/chromium/src/+/d2876f4ee1c1a01ee262ca55752503ba3a30a6d5

TBR=benhenry@chromium.org,vmiura@chromium.org,sullivan@chromium.org,tdresser@chromium.org,groby@chromium.org,rogerm@chromium.org
BUG= 697947 ,  697665 

Review-Url: https://codereview.chromium.org/2918173002
Cr-Original-Commit-Position: refs/heads/master@{#478379}
Review-Url: https://codereview.chromium.org/2956053003 .
Cr-Commit-Position: refs/branch-heads/3112@{#481}
Cr-Branched-From: b6460e24cf59f429d69de255538d0fc7a425ccf9-refs/heads/master@{#474897}

[modify] https://crrev.com/ac979d32d6ce857f746f44bb009b180676b3114a/components/translate/core/browser/translate_ranker_impl.cc

Status: Fixed (was: Assigned)
I think this is fixed now? Please reopen if I'm mistaken.

Sign in to add a comment