[benchmark] Do not call measureValue in warm-up iterations

[benchmark] Do not call measureValue in warm-up iterations

R=ricea@chromium.org

Committed: https://code.google.com/p/pywebsocket/source/detail?r=854

[hiroshige, 2014-10-22 21:32:51]

PTAL when you have time.

[Adam Rice, 2014-10-23 02:46:00]

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js
File src/example/benchmark.js (right):

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js#newc...
src/example/benchmark.js:53: calculateAndLogResult(config, size,
benchmark.startTimeInMs, totalSize,
I think since we have to change all the calls to calculateAndLogResult() anyway,
it would be better to do

if (!isWarmUp) {
  calculateAndLogResult(config, size, benchmark.startTimeInMs, totalSize);
}

and then calculateAndLogResult() itself doesn't have to change.

[hiroshige, 2014-10-23 05:06:04]

Thank you for reviewing!

Patch Set 2: I noticed addResultClearingTask is no longer needed and removed it.

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js
File src/example/benchmark.js (right):

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js#newc...
src/example/benchmark.js:53: calculateAndLogResult(config, size,
benchmark.startTimeInMs, totalSize,
I'd like to call config.addToLog (at the bottom of calculateAndLogResult) to
show the elapsed times for warmup iterations (but exclude the times from the
average reported), to check whether the performance changes during the warmup or
not.

[Adam Rice, 2014-10-23 08:13:24]

lgtm

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js
File src/example/benchmark.js (right):

https://codereview.appspot.com/154680043/diff/1/src/example/benchmark.js#newc...
src/example/benchmark.js:53: calculateAndLogResult(config, size,
benchmark.startTimeInMs, totalSize,
On 2014/10/23 05:06:04, hiroshige wrote:
> I'd like to call config.addToLog (at the bottom of calculateAndLogResult) to
> show the elapsed times for warmup iterations (but exclude the times from the
> average reported), to check whether the performance changes during the warmup
or
> not.

Okay.

[hiroshige, 2014-10-23 08:16:06]

Committed patchset #2 (id:20001) manually as r854 (presubmit successful).