cc: Changed to prediction histograms in tile manager

cc: Changed to prediction histograms in tile manager

I don't think we want all of these histograms uploaded,
we just care about them on local runs. I switched them
to be non-UMA

This change also updates the RasterTimes histogram
to be in microseconds.


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=192625

[vmpstr, 2013-04-04 19:26:19]

Please take a look.

[enne (OOO), 2013-04-04 20:42:09]

https://codereview.chromium.org/13575007/diff/1/cc/resources/tile_manager.cc
File cc/resources/tile_manager.cc (left):

https://codereview.chromium.org/13575007/diff/1/cc/resources/tile_manager.cc#...
cc/resources/tile_manager.cc:991:
UMA_HISTOGRAM_BOOLEAN("Renderer4.ColorPredictor.WrongActualNotSolid", true);
I'm not sure I agree that making these not uma histograms is a good choice. 
It'd be nice to be able to see in the field if the color predictor starts
failing.

[vmpstr, 2013-04-04 20:53:19]

https://codereview.chromium.org/13575007/diff/1/cc/resources/tile_manager.cc
File cc/resources/tile_manager.cc (left):

https://codereview.chromium.org/13575007/diff/1/cc/resources/tile_manager.cc#...
cc/resources/tile_manager.cc:991:
UMA_HISTOGRAM_BOOLEAN("Renderer4.ColorPredictor.WrongActualNotSolid", true);
On 2013/04/04 20:42:09, enne wrote:
> I'm not sure I agree that making these not uma histograms is a good choice. 
> It'd be nice to be able to see in the field if the color predictor starts
> failing.

These would only be recorded if both gpu-benchmarking and
prediction-benchmarking is turned on. If someone is running with those flags,
then I imagine they would be manually inspecting the results :)

I don't mind changing those back though, since we'll still record at least as
much data... Do you think all of my UMA -> non-UMA changes should be undone or
just the color predictor ones?

[enne (OOO), 2013-04-04 20:55:16]

Ohhhhh, I missed the bit where these aren't on in normal builds.  In that case
this is fine.

lgtm

[commit-bot: I haz the power, 2013-04-04 20:57:53]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/vmpstr@chromium.org/13575007/1

[tomhudson, 2013-04-04 21:06:08]

On 2013/04/04 20:55:16, enne wrote:
> Ohhhhh, I missed the bit where these aren't on in normal builds.  In that case
> this is fine.
> 
> lgtm

[tomhudson, 2013-04-04 21:07:38]

Actually, I had on my TODO list to try to get the ones that aren't
compute-intensive to figure out moved out from behind the if () so that we could
get field data. We've got the cheapness estimator live in M27 but we don't have
a sense of how conservative it is outside ~75 test sites we've run in telemetry.

[vmpstr, 2013-04-04 21:10:14]

On 2013/04/04 21:07:38, tomhudson wrote:
> Actually, I had on my TODO list to try to get the ones that aren't
> compute-intensive to figure out moved out from behind the if () so that we
could
> get field data. We've got the cheapness estimator live in M27 but we don't
have
> a sense of how conservative it is outside ~75 test sites we've run in
telemetry.

Oh.. Sorry about committing quickly.. I mean we can still pick and choose the
ones that we want to switch back, and make them on my default. WDYT?

[commit-bot: I haz the power, 2013-04-04 21:24:24]

Retried try job too often on win_x64_rel for step(s) compile
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_x64_re...

[commit-bot: I haz the power, 2013-04-05 17:19:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/vmpstr@chromium.org/13575007/1

[commit-bot: I haz the power, 2013-04-05 20:21:36]

Change committed as 192625