cc: Reduce duplicated tiling data difference iterator code

cc: Reduce duplicated tiling data difference iterator code

This adds a new BaseDifferenceIterator as a common base class for all
tiling data difference iterators and moves duplicated code from
different tiling data difference iterators to that common base class.
This reduces code duplication.

Committed: https://crrev.com/b3fa4672766fecb53cde887295ede63fb783e8b5
Cr-Commit-Position: refs/heads/master@{#321540}

[USE eero AT chromium.org, 2015-03-17 16:50:16]

e.hakkinen@samsung.com changed reviewers:
+ danakj@chromium.org, vmpstr@chromium.org

[USE eero AT chromium.org, 2015-03-17 16:50:41]

PTAL.

[danakj, 2015-03-17 16:56:09]

https://codereview.chromium.org/1017713003/diff/1/cc/base/tiling_data.h
File cc/base/tiling_data.h (right):

https://codereview.chromium.org/1017713003/diff/1/cc/base/tiling_data.h#newco...
cc/base/tiling_data.h:111: bool has_consider_rect() const;
You must use CamelCase for out-of-line methods.

[vmpstr, 2015-03-17 17:31:53]

This looks like a nice refactor. Are there any noticeable perf impact from this?

[USE eero AT chromium.org, 2015-03-18 09:07:27]

https://codereview.chromium.org/1017713003/diff/1/cc/base/tiling_data.h
File cc/base/tiling_data.h (right):

https://codereview.chromium.org/1017713003/diff/1/cc/base/tiling_data.h#newco...
cc/base/tiling_data.h:111: bool has_consider_rect() const;
On 2015/03/17 16:56:08, danakj wrote:
> You must use CamelCase for out-of-line methods.

Done.

[USE eero AT chromium.org, 2015-03-18 09:27:06]

On 2015/03/17 17:31:53, vmpstr wrote:
> Are there any noticeable perf impact from this?

I would assume that there will be none, but I will measure.

[vmpstr, 2015-03-18 15:13:59]

On 2015/03/18 09:27:06, e_hakkinen wrote:
> On 2015/03/17 17:31:53, vmpstr wrote:
> > Are there any noticeable perf impact from this?
> 
> I would assume that there will be none, but I will measure.

The reason I ask is that it's a pretty low level functionality that is used
heavily in iterators, so small differences can have impact :) I would also
assume however that this is basically a no-impact change

[USE eero AT chromium.org, 2015-03-19 14:22:03]

On 2015/03/18 15:13:59, vmpstr wrote:
> The reason I ask is that it's a pretty low level functionality that is used
> heavily in iterators, so small differences can have impact :) I would also
> assume however that this is basically a no-impact change

I would say that this really is a no-impact change.

There are some positive and some negative differences between results with and
without the patch. All in all, they are quite small and to both directions.
There are also some quite big difference between different runs of the code, so
most differences are probably just noise. Out-lier filtering seems to mandatory
to get any meaningful results and the filtering parameters also have an effect
on results.

Here are some results:

 *RESULT tile_manager_raster_tile_queue_construct: 2=	-0,12 %
 *RESULT tile_manager_raster_tile_queue_construct: 10=	+0,83 %
 *RESULT tile_manager_raster_tile_queue_construct: 50=	+1,02 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 2_16=	+0,27 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 2_32=	+0,83 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 2_64=	+0,98 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 2_128=	+1,16 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 10_16=	+0,41 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 10_32=	+0,18 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 10_64=	+0,21 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 10_128=	+0,33 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 50_16=	+0,78 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 50_32=	+0,10 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 50_64=	+0,06 %
 *RESULT tile_manager_raster_tile_queue_construct_and_iterate: 50_128=	-0,09 %
 *RESULT tile_manager_eviction_tile_queue_construct: 2=	+0,19 %
 *RESULT tile_manager_eviction_tile_queue_construct: 10=	-0,19 %
 *RESULT tile_manager_eviction_tile_queue_construct: 50=	-1,11 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 2_16=	+0,16 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 2_32=	-0,63 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 2_64=	-0,47 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 2_128=	+0,00 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 10_16=	-0,56 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 10_32=	-0,59 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 10_64=	-0,30 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 10_128=	+0,16 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 50_16=	-0,78 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 50_32=	-0,64 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 50_64=	-0,24 %
 *RESULT tile_manager_eviction_tile_queue_construct_and_iterate: 50_128=	+0,08 %

[vmpstr, 2015-03-19 23:22:59]

lgtm

[USE eero AT chromium.org, 2015-03-20 08:04:21]

The CQ bit was checked by e.hakkinen@samsung.com

[commit-bot: I haz the power, 2015-03-20 08:04:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1017713003/20001

[commit-bot: I haz the power, 2015-03-20 10:06:54]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-03-20 10:08:07]

Patchset 2 (id:??) landed as
https://crrev.com/b3fa4672766fecb53cde887295ede63fb783e8b5
Cr-Commit-Position: refs/heads/master@{#321540}