All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=694045

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguPuthwgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDghKLNoAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-POltgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-Lrd4gkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-OKn8QgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-IvYoAoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-PHS8wsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguL-wqwsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-LT5qgoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguNfM_wgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-JejuAkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguK7Z_woM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-JfusAkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-MLkgAgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguKne_AkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-I3rpwsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-JKV7AgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-ITL5AkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-JCA4QoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-IqFkAkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-ITL5AoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguLOY5gkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-LrdogsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-PbioAsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-Lyt_gsM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-LDSvgkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-ISf9QkM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDguMqhwQgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-Kqu5goM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-IrAlwgM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg-OzlhwoM


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

chromium-rel-mac-retina
chromium-rel-mac11-air
chromium-rel-win10
chromium-rel-win7-dual
chromium-rel-win7-gpu-ati
chromium-rel-win7-gpu-intel
chromium-rel-win7-gpu-nvidia
chromium-rel-win7-x64-dual
chromium-rel-win8-dual
win-high-dpi

Started bisect job https://chromeperf.appspot.com/buildbucket_job_status/8987214341606053120

likely the same as   Issue 694043 

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

Hi darin@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 : darin
  Commit : 2d145fe56a1d995fc1a016da03c1f91f217612e0
  Date   : Thu Feb 16 15:22:35 2017
  Subject: HTML MessagePort as mojo::MessagePipeHandle

Bisect Details
  Configuration: winx64_high_dpi_perf_bisect
  Benchmark    : service_worker.service_worker_micro_benchmark
  Metric       : concurrent_100_response_50_percentile/concurrent_100_response_50_percentile
  Change       : 45.41% | 74.8125 -> 108.788333333

Revision             Result                  N
chromium@450964      74.8125 +- 12.2505      6      good
chromium@450969      75.6533 +- 5.03963      6      good
chromium@450970      98.0133 +- 1.45808      6      bad       <--
chromium@450971      109.49 +- 3.89224       6      bad
chromium@450973      104.871 +- 4.77063      6      bad
chromium@450981      105.11 +- 3.72099       6      bad
chromium@450997      116.801 +- 9.78116      6      bad
chromium@451030      104.152 +- 10.0869      6      bad
chromium@451096      108.788 +- 6.03256      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 service_worker.service_worker_micro_benchmark

Debug Info
  https://chromeperf.appspot.com/buildbucket_job_status/8987214341606053120

Is this bisect wrong?
  https://chromeperf.appspot.com/bad_bisect?try_job_id=5782048729464832


| 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!

I wonder if this could be due to making Blob registration IPCs synchronous.

Yes, this benchmark is very sensitive to changes in Blob code. It measures a SW responding to many (up to 100) concurrent XHR requests using something like respondWith(new Response(‘hello’)), which ends up creating a Blob.

Is the MessagePort commit expected to slow down blob creation? Since this microbenchmark isn't super representative of real world usage, we might just accept this regression.

Yes, the MessagePort includes the change that turns Blob registration IPC into Synchronous, therefore it is pretty much expected that it slows down Blob creation.

The benchmark isn't really representative as you say so I suppose we can probably just accept (or update the benchmark code), but let's also check if we don't have any important real-world usage this change might badly affect. (I can't think of for now)

Yeah, I don't know of one and a cursory investigation of top SW sites didn't show usage.

I think we can just accept this. If it turns out to matter we could also consider transferring string-constructed responses as streams rather than blobs.