Set minimum Windows wait time to 1 ms in MessagePumpDefault::Run

Set minimum Windows wait time to 1 ms in MessagePumpDefault::Run

On Windows if we call event_.TimedWait(delay); with a sub-ms delay
then the wait function will return immediately. This means we end up
busy waiting down to zero. Rounding these times up to 1 ms lets Chrome
go idle and saves a few percent of a core on some tests.

The symptom to watch for is lots of time being spent in QPCNow, called
from MessagePumpDefault::Run or from MessageLoop::DoDelayedWork.

This is the second attempt at fixing this bug. The first attempt tried
rounding down the short delays but this was stymied by other code that
didn't respect the rounding down.

One known problem is that some animations that try to run at 60 fps by
using setInterval(C,16) will drop from ~60 fps to ~58 fps. Those that
use requestAnimationFrame run at a steady rate of 60 fps.

BUG=487724
Committed: https://crrev.com/932c2457349e44eeb7b3152f8b49acd186e7057e
Cr-Commit-Position: refs/heads/master@{#331226}

[brucedawson, 2015-05-22 17:58:08]

brucedawson@chromium.org changed reviewers:
+ cpu@chromium.org, jar@chromium.org

[brucedawson, 2015-05-22 18:10:06]

Possible alternative fix. A ~2-6% reduction in CPU time consumed across all
Chrome processes was confirmed in one 60 fps animation scenario.

There may be places where this causes bigger elapsed-time regressions - when the
timer is at 15.625 ms perhaps - but the CPU consumption reductions should be
solid across the board.

[cpu_(ooo_6.6-7.5), 2015-05-22 21:36:42]

https://codereview.chromium.org/1154953002/diff/1/base/message_loop/message_p...
File base/message_loop/message_pump_default.cc (right):

https://codereview.chromium.org/1154953002/diff/1/base/message_loop/message_p...
base/message_loop/message_pump_default.cc:55: #if defined(OS_WIN)
delay = std::max(delay, TimeDelta::FromMilliseconds(1))

now if we ever get microseconds in there would we rather round up at > 500us ?

[brucedawson, 2015-05-22 22:08:47]

If we can get rounding down to work then rounding down at 500us would be worth
exploring, but currently rounding down would hit the problem that the previous
fix hit -- it requires putting the rounding down logic in multiple places. So,
currently adding rounding down at 500us would cause a busy wait in that case.

I made the change to use std::max. That is cleaner. It also made me realize that
I need to check for > zero before checking for < 1 ms, which is now fixed.

PTAL

[cpu_(ooo_6.6-7.5), 2015-05-22 22:57:54]

https://codereview.chromium.org/1154953002/diff/40001/base/message_loop/messa...
File base/message_loop/message_pump_default.cc (right):

https://codereview.chromium.org/1154953002/diff/40001/base/message_loop/messa...
base/message_loop/message_pump_default.cc:59: if (delay > TimeDelta()) {
don't use the braces in single if() statement

lgtm

[brucedawson, 2015-05-22 23:00:32]

Fixed.

[brucedawson, 2015-05-22 23:00:35]

The CQ bit was checked by brucedawson@chromium.org

[brucedawson, 2015-05-22 23:00:35]

The patchset sent to the CQ was uploaded after l-g-t-m from cpu@chromium.org
Link to the patchset: https://codereview.chromium.org/1154953002/#ps60001
(title: "Remove unnecessary braces.")

[commit-bot: I haz the power, 2015-05-22 23:02:26]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1154953002/60001

[commit-bot: I haz the power, 2015-05-23 00:14:16]

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

[commit-bot: I haz the power, 2015-05-23 00:14:17]

Try jobs failed on following builders:
  win8_chromium_rel on tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_rel...)

[brucedawson, 2015-05-23 00:34:13]

It's not clear why std::max was available in some configurations but not others.
Including <algorithm> when using std::max is a best practice anyway so making
that change.

[brucedawson, 2015-05-23 00:39:42]

The CQ bit was checked by brucedawson@chromium.org

[brucedawson, 2015-05-23 00:39:42]

The patchset sent to the CQ was uploaded after l-g-t-m from cpu@chromium.org
Link to the patchset: https://codereview.chromium.org/1154953002/#ps80001
(title: "Add missing header include.")

[commit-bot: I haz the power, 2015-05-23 00:40:02]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1154953002/80001

[commit-bot: I haz the power, 2015-05-23 03:12:11]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-05-23 03:13:01]

Patchset 5 (id:??) landed as
https://crrev.com/932c2457349e44eeb7b3152f8b49acd186e7057e
Cr-Commit-Position: refs/heads/master@{#331226}