All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=760527

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=cebde7ba5a2207212ea786cf79eb2cca3997aa3145ff721be961066083453ee4


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

chromium-rel-win7-dual

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8969838448954116944

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

Hi jgruber@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 : Jakob Gruber
  Commit : 49e3bfd572dcc60f2c97c732bcc4c918cbc3fd9f
  Date   : Fri Aug 25 09:34:38 2017
  Subject: [snapshot] Move builtins to dedicated snapshot area

Bisect Details
  Configuration: win_perf_bisect
  Benchmark    : v8.browsing_desktop
  Metric       : v8-gc-incremental-step_max/browse_search/browse_search_google_india
  Change       : 85.79% | 1.72633333333 -> 3.20733333333

Revision                           Result                   N
chromium@497447                    1.72633 +- 0.683309      6      good
chromium@497451                    1.62467 +- 0.822656      6      good
chromium@497453                    1.78233 +- 0.639027      6      good
chromium@497453,v8@2ee967d253      1.57933 +- 0.51018       6      good
chromium@497453,v8@7571de3acf      1.79667 +- 0.910652      6      good
chromium@497453,v8@49e3bfd572      3.27883 +- 0.553605      6      bad       <--
chromium@497453,v8@77c7ef6750      3.36967 +- 0.40184       6      bad
chromium@497453,v8@b6158eb6be      3.28167 +- 0.414231      6      bad
chromium@497454                    3.8215 +- 3.75778        6      bad
chromium@497460                    3.146 +- 0.296189        6      bad
chromium@497473                    3.42083 +- 0.441231      6      bad
chromium@497499                    3.26183 +- 0.550726      6      bad
chromium@497551                    3.20733 +- 0.313352      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 --story-filter=browse.search.google.india v8.browsing_desktop

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

Debug information about this bisect:
  https://chromeperf.appspot.com/buildbucket_job_status/8969838448954116944


For feedback, file a bug with component Speed>Bisection

Hmm this one is surprising. The CL only refactors builtin (de)serialization. I'll have to take a closer look. First guesses are:

* Builtins are no longer immovable / in the wrong space? (Unlikely)
* The partial snapshot cache is larger.
* We may generate more write barriers during deserialization(?)

_avg and _count stay roughly unchanged while _max spikes [0]. Maybe also worth noting that only browse_media_flickr_infinite_scroll is affected.

[0] https://chromeperf.appspot.com/report?sid=ebac454c7b4ff78b17c33b7efd14da39a6479f556e935c2386ab81ecbcedd32a&start_rev=490493&end_rev=498609

Offline discussion with mlippautz@: incremental marking statistics seem to be off for 2 of the initial steps, and increase the step limit by around 10x [0]. Later steps again seem to have a reasonable limit.

Deferring further investigation until GC folks & myself return from vacation.

[0] tools/perf/run_benchmark run v8.browsing_desktop --browser=exact --browser-executable=out/Release/chrome --story-filter=browse:search:google_india  --output-format=html --show-stdout --extra-browser-args="--js-flags=\"--trace-gc --trace-incremental-marking\""

 Issue 761736  has been merged into this issue.

 Issue 763914  has been merged into this issue.

The NextAction date has arrived: 2017-10-09

I checked few graphs and they have recovered. There were many GC changes in the meantime: new write barrier, concurrent marking.

jgruber@, could you please check that the graphs are recovered and close this issue?