Telemetry: Normalize fast-path results with frames.

Telemetry: Normalize fast-path results with frames.

This changes our unit from percentage-busy to ms-per-frame.
This should be less noisy to stalls where we are idle not
producing frames.

NOTRY=true
BUG=334063
No-try since this obviously doesn't break net-unittests.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=245301

[epenner, 2014-01-14 23:10:34]

Ptal. The frame-counting was uber simple as long as this is okay.

I'm beginning to like this form of 'parsing' the trace more and more, since it's
all done after the fact with python (easy) vs instrumenting run-time code (not
so easy). So maybe it's worth discussing formalizing some trace names as
constants, to make this more robust?

[epenner, 2014-01-14 23:11:03]

Also, Nat I'm happy to wait for your patch to land first.

[epenner, 2014-01-15 03:00:38]

Rebased on Nat's patch. Ptal.

This fixes two small issues I found with Nat's patch (only sum fast-path
threads, and respect 'MatchBySubstring').

Nat, currently the extra details are not normalized by the number of swap
buffers. Let me know if you want me to do that.

[nduca, 2014-01-15 20:57:44]

lgtm

[nduca, 2014-01-15 20:58:13]

(add unit test coverage now that i have tests in timeline_unittest.py)

[epennerAtGoogle, 2014-01-15 21:35:38]

On 2014/01/15 20:58:13, nduca wrote:
> (add unit test coverage now that i have tests in timeline_unittest.py)

Coming up.

[epenner, 2014-01-15 23:16:47]

Ptal for tests. Nat, I needed to decide what to do with the main thread details.
I decided to divide everything by the frame-count, since I think that makes the
most sense. So all results are now reported in ms/frame.

[epenner, 2014-01-16 01:38:58]

Ptal for reporting the correct silk/fastpath results.

https://codereview.chromium.org/138923004/diff/440001/tools/perf/benchmarks/s...
File tools/perf/benchmarks/silk.py (right):

https://codereview.chromium.org/138923004/diff/440001/tools/perf/benchmarks/s...
tools/perf/benchmarks/silk.py:15: options = {"report_silk_results": True}
I didn't see this being used, so I've changed the name and used it to indicate
the reduced set of thread reports.

[epenner, 2014-01-16 01:43:52]

I'm landing since Nat and I have talked offline and I want to fix the fast-path
benchmark.

I've tested both silk and fast-path benchmarks. I'll un-cq if there is any
issues, or follow up with another CL.

[commit-bot: I haz the power, 2014-01-16 02:12:44]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/138923004/440001

[nduca, 2014-01-16 03:13:06]

lgtm

[commit-bot: I haz the power, 2014-01-16 06:01:24]

Retried try job too often on win_rel for step(s) app_list_unittests,
ash_unittests, aura_unittests, cacheinvalidation_unittests, cc_unittests,
check_deps, check_deps2git, chrome_elf_unittests, chromedriver_unittests,
components_unittests, compositor_unittests, content_browsertests,
content_unittests, crypto_unittests, events_unittests, google_apis_unittests,
gpu_unittests, installer_util_unittests, ipc_tests, jingle_unittests,
media_unittests, mini_installer_test, nacl_integration, ppapi_unittests,
printing_unittests, remoting_unittests, sql_unittests, sync_integration_tests,
sync_unit_tests, telemetry_unittests, views_unittests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2014-01-16 16:22:38]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/138923004/440001

[commit-bot: I haz the power, 2014-01-16 18:03:40]

Retried try job too often on win_rel for step(s) net_unittests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2014-01-16 20:22:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/138923004/440001

[commit-bot: I haz the power, 2014-01-16 20:23:41]

Change committed as 245301