[Perf Impact] Make frame time discrepancy absolute.

[Perf Impact] Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.

Committed: https://chromium.googlesource.com/external/github.com/catapult-project/catapult/+/b5763ff333c804d0d1884c174599e27c695b4275

[ulan, 2016-07-01 07:51:30]

ulan@chromium.org changed reviewers:
+ benjhayden@chromium.org

[ulan, 2016-07-01 07:51:30]

ptal

[benjhayden, 2016-08-10 19:27:22]

benjhayden@chromium.org changed reviewers:
+ tdresser@chromium.org

[benjhayden, 2016-08-10 19:27:23]

Tim, do y'all still use this metric? Does anybody? Whoever uses it, what do they
need out of it?

[tdresser, 2016-08-11 13:27:47]

Description was changed from

==========
Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.
==========

to

==========
Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.
==========

[tdresser, 2016-08-11 13:27:47]

tdresser@chromium.org changed reviewers:
+ vmiura@chromium.org
- tdresser@chromium.org

[tdresser, 2016-08-11 13:28:45]

tdresser@chromium.org changed reviewers:
+ tdresser@chromium.org

[tdresser, 2016-08-11 13:28:45]

I almost never look at this metric.
+vmiura@, who has more context here than I do.

[vmiura, 2016-08-11 20:22:12]

We use this in telemetry to evaluate changes in CC/GPU from time to time; it
shows spikes from decoding images or compiling GL shaders for example.

It seems fine to me to make it absolute ms.

[benjhayden, 2016-08-11 20:48:41]

lgtm

[tdresser, 2016-08-12 12:23:51]

Is this currently alerting?

If so, the subject of this commit should indicate that this is likely to trigger
spurious alerts.

Something like 
[Perf Impact] Make frame time...

Otherwise, LGTM.

[sullivan, 2016-08-12 13:19:53]

On 2016/08/12 12:23:51, tdresser wrote:
> Is this currently alerting?

If I'm understanding the metric right, I think this is the only alerting rule on
it:

ChromiumPerf/*/smoothness.tough_webgl_cases/frame_times_discrepancy/*

[benjhayden, 2016-08-18 18:18:40]

On 2016/08/12 at 13:19:53, sullivan wrote:
> On 2016/08/12 12:23:51, tdresser wrote:
> > Is this currently alerting?
> 
> If I'm understanding the metric right, I think this is the only alerting rule
on it:
> 
> ChromiumPerf/*/smoothness.tough_webgl_cases/frame_times_discrepancy/*

sullivan, where/how did you find that alerting rule, so we can do that ourselves
in the future?

[sullivan, 2016-08-18 18:23:42]

On 2016/08/18 18:18:40, benjhayden wrote:
> On 2016/08/12 at 13:19:53, sullivan wrote:
> > On 2016/08/12 12:23:51, tdresser wrote:
> > > Is this currently alerting?
> > 
> > If I'm understanding the metric right, I think this is the only alerting
rule
> on it:
> > 
> > ChromiumPerf/*/smoothness.tough_webgl_cases/frame_times_discrepancy/*
> 
> sullivan, where/how did you find that alerting rule, so we can do that
ourselves
> in the future?

https://chromeperf.appspot.com/edit_sheriffs

Long-term, I would really love to see monitoring, test owners, and things like
that configured benchmark side and passed up to the dashboard in tbm2.

[ulan, 2016-08-25 10:42:04]

Description was changed from

==========
Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.
==========

to

==========
[Perf Impact] Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.
==========

[ulan, 2016-08-25 10:42:09]

The CQ bit was checked by ulan@chromium.org

[commit-bot: I haz the power, 2016-08-25 10:42:16]

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, 2016-08-25 11:03:24]

Description was changed from

==========
[Perf Impact] Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.
==========

to

==========
[Perf Impact] Make frame time discrepancy absolute.

Currently frame time discrepancy is computed as relative, i.e.
the discrepancy value is divided by the total frame range.

Absolute frame time discrepancy has two advantages:
1. It is measured in ms, so it is easier to understand.
2. It is less noisy than relative frame time discrepancy.

Committed:
https://chromium.googlesource.com/external/github.com/catapult-project/catapu...
==========

[commit-bot: I haz the power, 2016-08-25 11:03:25]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/external/github.com/catapult-project/catapu...