All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=633943

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICg5pL1vwkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgpqO5rwoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgICgpsW1vwkM


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

chromium-rel-mac-retina
chromium-rel-mac10
chromium-rel-mac11

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9005355276668846896

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

Hi kojii@chromium.org, 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 : More LayoutLocale refactor with additional Chinese support
Author  : kojii
Commit description:
  
Following the initial LayoutLocale refactoring CL[1], this patch:

1. Support 14 encompassed languages within the Chinese macrolanguage[2].
2. Add "mo" (Macau) as "Traditional by default", as pointed out by W3C
   I18N WG and match to Firefox.
3. Better and more spec conformance to parse BCP-47 language tags[3].
4. Change "und-Zsye" (Emoji) priority from the lowest to the highest.
5. Unify the logic for disambiguation of the Unified Han Ideographs for
   Linux/Android and Windows.
6. Merge duplicated code in AcceptLanguagesResolver to LayoutLocale.
7. Centralize locale-related methods more to LayoutLocale for better
   discoverability, caching, and code sharing.

[1] https://codereview.chromium.org/2161683002
[2] http://www-01.sil.org/iso639-3/documentation.asp?id=zho
[3] https://tools.ietf.org/html/bcp47

BUG= 586517 ,  611817 

Review-Url: https://codereview.chromium.org/2192703002
Cr-Commit-Position: refs/heads/master@{#409157}
Commit  : 3ec109dc729a647b433136b2328e9da01ba9c8b8
Date    : Tue Aug 02 08:28:23 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev      N  Good?
chromium@409146  1014360  1.64636e-10  5  good
chromium@409152  1014360  0.0          5  good
chromium@409155  1014360  1.16415e-10  5  good
chromium@409156  1014360  0.0          5  good
chromium@409157  1026648  0.0          5  bad    <--

Bisect job ran on: mac_10_10_perf_bisect
Bug ID: 633943

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --also-run-disabled-tests system_health.memory_desktop
Test Metric: load_games-memory:chrome:all_processes:reported_by_chrome:partition_alloc:effective_size_avg/load_games_lazors
Relative Change: 1.21%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_10_10_perf_bisect/builds/2262
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9005355276668846896


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5798569525641216

| 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!

I don't think the CL can increase memory on Mac by 15kb. Most of the changes apply only to Win/Linux/Android, one global change can possibly use 50 bytes or so memory, but not the order of 15kb.

Is this possible from other CL?

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/9005326749375664304

The result seems pretty conclusive (given the standard deviations), but I started one more bisect.

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


===== SUSPECTED CL(s) =====
Subject : More LayoutLocale refactor with additional Chinese support
Author  : kojii
Commit description:
  
Following the initial LayoutLocale refactoring CL[1], this patch:

1. Support 14 encompassed languages within the Chinese macrolanguage[2].
2. Add "mo" (Macau) as "Traditional by default", as pointed out by W3C
   I18N WG and match to Firefox.
3. Better and more spec conformance to parse BCP-47 language tags[3].
4. Change "und-Zsye" (Emoji) priority from the lowest to the highest.
5. Unify the logic for disambiguation of the Unified Han Ideographs for
   Linux/Android and Windows.
6. Merge duplicated code in AcceptLanguagesResolver to LayoutLocale.
7. Centralize locale-related methods more to LayoutLocale for better
   discoverability, caching, and code sharing.

[1] https://codereview.chromium.org/2161683002
[2] http://www-01.sil.org/iso639-3/documentation.asp?id=zho
[3] https://tools.ietf.org/html/bcp47

BUG= 586517 ,  611817 

Review-Url: https://codereview.chromium.org/2192703002
Cr-Commit-Position: refs/heads/master@{#409157}
Commit  : 3ec109dc729a647b433136b2328e9da01ba9c8b8
Date    : Tue Aug 02 08:28:23 2016


===== TESTED REVISIONS =====
Revision         Mean     Std Dev      N  Good?
chromium@409146  1014360  1.16415e-10  5  good
chromium@409153  1014360  0.0          5  good
chromium@409155  1014360  1.64636e-10  5  good
chromium@409156  1014360  0.0          5  good
chromium@409157  1026648  0.0          5  bad    <--
chromium@409160  1026648  0.0          5  bad

Bisect job ran on: mac_retina_perf_bisect
Bug ID: 633943

Test Command: src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --upload-results --also-run-disabled-tests system_health.memory_desktop
Test Metric: load_games-memory:chrome:all_processes:reported_by_chrome:partition_alloc:effective_size_avg/load_games_lazors
Relative Change: 1.21%
Score: 99.9

Buildbot stdio: http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_retina_perf_bisect/builds/1522
Job details: https://chromeperf.appspot.com/buildbucket_job_status/9005326749375664304


Not what you expected? We'll investigate and get back to you!
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5807497688907776

| 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!

Ok, I'll see if I can #if out some code not relevant to Mac.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f7f744c29306b76d03bd3a9146562bb1ca2f830f

commit f7f744c29306b76d03bd3a9146562bb1ca2f830f
Author: kojii <kojii@chromium.org>
Date: Mon Aug 08 20:17:31 2016

Defer computing localeForHan until needed

This patch defers computing LayoutLocale::localeForHan() until it is
needed.

On Mac, localeForHan() is not used. On other platforms, it is used only
when fallback for Han characters is needed.

BUG= 586517 ,  633943 

Review-Url: https://codereview.chromium.org/2213393002
Cr-Commit-Position: refs/heads/master@{#410424}

[modify] https://crrev.com/f7f744c29306b76d03bd3a9146562bb1ca2f830f/third_party/WebKit/Source/platform/LayoutLocale.cpp
[modify] https://crrev.com/f7f744c29306b76d03bd3a9146562bb1ca2f830f/third_party/WebKit/Source/platform/LayoutLocale.h
[modify] https://crrev.com/f7f744c29306b76d03bd3a9146562bb1ca2f830f/third_party/WebKit/Source/platform/fonts/AcceptLanguagesResolver.cpp
[modify] https://crrev.com/f7f744c29306b76d03bd3a9146562bb1ca2f830f/third_party/WebKit/Source/platform/fonts/AcceptLanguagesResolver.h
[modify] https://crrev.com/f7f744c29306b76d03bd3a9146562bb1ca2f830f/third_party/WebKit/Source/platform/fonts/AcceptLanguagesResolverTest.cpp

 Issue 634400  has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/23d537bfe5050e53dd965cc2805cbb8d0268630a

commit 23d537bfe5050e53dd965cc2805cbb8d0268630a
Author: kojii <kojii@chromium.org>
Date: Wed Aug 10 03:14:40 2016

Avoid relying on NVRO for HashMap

The previous CL[1] relies on NRVO (Named Return Value Optimization)[2],
which is defined as "may" in C++ spec. This CL allows non-NRVO
compilers.

[1] https://codereview.chromium.org/2192703002
[2] http://en.cppreference.com/w/cpp/language/copy_elision

BUG= 586517 ,  633943 

Review-Url: https://codereview.chromium.org/2221963002
Cr-Commit-Position: refs/heads/master@{#410940}

[modify] https://crrev.com/23d537bfe5050e53dd965cc2805cbb8d0268630a/third_party/WebKit/Source/platform/text/LocaleToScriptMapping.cpp

Perf sheriff ping: reminder to follow up on possible performance issues

Temporarily declaring bankruptcy on the *desktop* system health benchmark.
The number of alerts became unmanageable and the overall process needs to be improved to make it sustainable.
The alerts have been turned off and I'm archiving the outstanding regressions.
Note: this is just about desktop, the mobile system health stays up.