Force GC in different ways for different measurements in blink_perf tests

Force GC in different ways for different measurements in blink_perf tests

Previously we force GC before each iteration of test, but for frame time
measurement, the result is much affected by GC performance. Now force
GC in different ways:

- measureTime: force GC before recording the start time of each
  iteration;
- measureRunsPerSecond: same as above;
- measureFrameTime: force GC before starting the whole test;

Other measurements either call one of the above measurement methods, or
don't call scheduleNextRun() which previously forced GC so not affected.

BUG=667811
Committed: https://crrev.com/a61df787643d5238a3f902a2acfafc85c0e6690f
Cr-Commit-Position: refs/heads/master@{#437109}

[Xianzhu, 2016-12-07 23:26:55]

The CQ bit was checked by wangxianzhu@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-12-07 23:27:43]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Xianzhu, 2016-12-07 23:39:31]

wangxianzhu@chromium.org changed reviewers:
+ pdr@chromium.org

[pdr., 2016-12-08 00:04:38]

On 2016/12/07 at 23:39:32, wangxianzhu wrote:
> 

LGTM

Before landing, can you do a sanity check with and without this patch for one of
the tests, to see if the numbers change at all?

[pdr., 2016-12-08 00:06:04]

On 2016/12/08 at 00:04:38, pdr. wrote:
> On 2016/12/07 at 23:39:32, wangxianzhu wrote:
> > 
> 
> LGTM
> 
> Before landing, can you do a sanity check with and without this patch for one
of the tests, to see if the numbers change at all?

Actually... that's what the perf bots are for. They will let us know quickly if
this affects the numbers. LGTM to land as-is.

[Xianzhu, 2016-12-08 00:12:01]

On 2016/12/08 00:06:04, pdr. wrote:
> On 2016/12/08 at 00:04:38, pdr. wrote:
> > On 2016/12/07 at 23:39:32, wangxianzhu wrote:
> > > 
> > 
> > LGTM
> > 
> > Before landing, can you do a sanity check with and without this patch for
one
> of the tests, to see if the numbers change at all?
> 
> Actually... that's what the perf bots are for. They will let us know quickly
if
> this affects the numbers. LGTM to land as-is.

I did schedule perf jobs. Will land after they finish.

[commit-bot: I haz the power, 2016-12-08 00:58:21]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-12-08 00:58:22]

Dry run: This issue passed the CQ dry run.

[Xianzhu, 2016-12-08 01:10:40]

The CQ bit was checked by wangxianzhu@chromium.org

[commit-bot: I haz the power, 2016-12-08 01:11:33]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-12-08 01:21:40]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1481159440544770, "parent_rev":
"e355a412f2593159207cc758f337b87faec4d19a", "commit_rev":
"e5e2358b7bd0aa946736cd429527907fc7ac164b"}

[commit-bot: I haz the power, 2016-12-08 01:22:09]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-12-08 01:25:49]

Description was changed from

==========
Force GC in different ways for different measurements in blink_perf tests

Previously we force GC before each iteration of test, but for frame time
measurement, the result is much affected by GC performance. Now force
GC in different ways:

- measureTime: force GC before recording the start time of each
  iteration;
- measureRunsPerSecond: same as above;
- measureFrameTime: force GC before starting the whole test;

Other measurements either call one of the above measurement methods, or
don't call scheduleNextRun() which previously forced GC so not affected.

BUG=667811
==========

to

==========
Force GC in different ways for different measurements in blink_perf tests

Previously we force GC before each iteration of test, but for frame time
measurement, the result is much affected by GC performance. Now force
GC in different ways:

- measureTime: force GC before recording the start time of each
  iteration;
- measureRunsPerSecond: same as above;
- measureFrameTime: force GC before starting the whole test;

Other measurements either call one of the above measurement methods, or
don't call scheduleNextRun() which previously forced GC so not affected.

BUG=667811

Committed: https://crrev.com/a61df787643d5238a3f902a2acfafc85c0e6690f
Cr-Commit-Position: refs/heads/master@{#437109}
==========

[commit-bot: I haz the power, 2016-12-08 01:25:50]

Patchset 1 (id:??) landed as
https://crrev.com/a61df787643d5238a3f902a2acfafc85c0e6690f
Cr-Commit-Position: refs/heads/master@{#437109}