Hmmh, the graph in #0 is comparing between the max of onload metric in pcv2 against pcv1's onload.

If we compare the average onload, then pcv2's avg onload metric is less than pcv1's onload:

https://chromeperf.appspot.com/report?sid=d0cdccbb2cd2de38fee5a23fd4f1202858eca8096d6d15aa010b15be4e8f89fb&start_rev=397302&end_rev=398580

+Kouhei: I thought each trace in pcv2 should produce a single onload metric only?

Confirmed on arstechnica run:
https://chromeperf.appspot.com/report?sid=a2b6c02e63ec6a9412f1e55a329c3110f10e7c886a00c302523049f3eab9b111&start_rev=397302&end_rev=398580

Investigating

Good catch Kouhei!

To be clearer, here is the combined graph of arstechnica run:

https://chromeperf.appspot.com/report?sid=c31d588144cd4a4c5fa333318db7c17dc34b5432f66d79479781f26c776d0504&start_rev=397302&end_rev=398580

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a16ec0d3ba867c0e3e46206c7f3ab36c4f146728

commit a16ec0d3ba867c0e3e46206c7f3ab36c4f146728
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Thu Jun 09 07:38:33 2016

Roll src/third_party/catapult/ 8ff23141e..20c004c9f (1 commit).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/8ff23141e1df..20c004c9fccd

$ git log 8ff23141e..20c004c9f --date=short --no-merges --format='%ad %ae %s'

BUG= 618156 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2044363003
Cr-Commit-Position: refs/heads/master@{#398807}

[modify] https://crrev.com/a16ec0d3ba867c0e3e46206c7f3ab36c4f146728/DEPS

Kouhei: your change make the avg closer to the max value of pcv2 onload. Now we still have 200ms diff between pcv2 & pcv1 for both timeToOnload_avg & timeToOnload_max:

https://chromeperf.appspot.com/report?sid=c31d588144cd4a4c5fa333318db7c17dc34b5432f66d79479781f26c776d0504&start_rev=398398&end_rev=398812

The 200-300ms diff between pcv1 onload & pcv2 onload is very persistent for every single page:
https://chromeperf.appspot.com/report?sid=2b74ed6f6cf69fdd6f40dc43c9699ec3df966007c2756f6a456ad2f52710522b&start_rev=397892&end_rev=398812

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4b11d01326b69961a8d3c75103e0b3a423fcecbd

commit 4b11d01326b69961a8d3c75103e0b3a423fcecbd
Author: kouhei <kouhei@chromium.org>
Date: Thu Jun 09 18:42:15 2016

[PCv2] remove wildcard from tracing category filter

Before this CL, the chrome traced high-overhead categories like NetLog.
This CL removes wildcard from the tracing category filter so that PCv2
records tracing with minimal overhead.

BUG= 618156 

Review-Url: https://codereview.chromium.org/2057553002
Cr-Commit-Position: refs/heads/master@{#398960}

[modify] https://crrev.com/4b11d01326b69961a8d3c75103e0b3a423fcecbd/tools/perf/benchmarks/page_cycler_v2.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a16ec0d3ba867c0e3e46206c7f3ab36c4f146728

commit a16ec0d3ba867c0e3e46206c7f3ab36c4f146728
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Thu Jun 09 07:38:33 2016

Roll src/third_party/catapult/ 8ff23141e..20c004c9f (1 commit).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/8ff23141e1df..20c004c9fccd

$ git log 8ff23141e..20c004c9f --date=short --no-merges --format='%ad %ae %s'

BUG= 618156 

TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2044363003
Cr-Commit-Position: refs/heads/master@{#398807}

[modify] https://crrev.com/a16ec0d3ba867c0e3e46206c7f3ab36c4f146728/DEPS

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/4b11d01326b69961a8d3c75103e0b3a423fcecbd

commit 4b11d01326b69961a8d3c75103e0b3a423fcecbd
Author: kouhei <kouhei@chromium.org>
Date: Thu Jun 09 18:42:15 2016

[PCv2] remove wildcard from tracing category filter

Before this CL, the chrome traced high-overhead categories like NetLog.
This CL removes wildcard from the tracing category filter so that PCv2
records tracing with minimal overhead.

BUG= 618156 

Review-Url: https://codereview.chromium.org/2057553002
Cr-Commit-Position: refs/heads/master@{#398960}

[modify] https://crrev.com/4b11d01326b69961a8d3c75103e0b3a423fcecbd/tools/perf/benchmarks/page_cycler_v2.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b07fec5fa70ba1f13eb9479f077a452871a2cf64

commit b07fec5fa70ba1f13eb9479f077a452871a2cf64
Author: catapult-deps-roller <catapult-deps-roller@chromium.org>
Date: Wed Oct 12 02:22:24 2016

Roll src/third_party/catapult/ 4fa0ede5e..daccdeb8c (2 commits).

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/4fa0ede5e814..daccdeb8c823

$ git log 4fa0ede5e..daccdeb8c --date=short --no-merges --format='%ad %ae %s'
2016-10-11 eyaich Removing stale code for page_cycler_v1
2016-10-11 sullivan Fix PathsForDeployment

BUG= 618156 

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel
TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2411003005
Cr-Commit-Position: refs/heads/master@{#424650}

[modify] https://crrev.com/b07fec5fa70ba1f13eb9479f077a452871a2cf64/DEPS