cc: Mark tiles as processed for solid color analysis.

cc: Mark tiles as processed for solid color analysis.

This patch ensures we don't process tiles for solid color analysis
many times. This wasn't as much of an issue before we separated raster
and decode images, since we'd only process a tile once. Now, we can
visit tiles more than once.

This shows pretty significant gains on perftests.
Before:
*RESULT prepare_tiles: 2_100= 2236.82763671875 runs/s
*RESULT prepare_tiles: 2_500= 8168.21533203125 runs/s
*RESULT prepare_tiles: 2_1000= 10103.5654296875 runs/s
*RESULT prepare_tiles: 10_100= 8856.08984375 runs/s
*RESULT prepare_tiles: 10_500= 8581.669921875 runs/s
*RESULT prepare_tiles: 10_1000= 8932.677734375 runs/s
*RESULT prepare_tiles: 50_100= 3708.318359375 runs/s
*RESULT prepare_tiles: 50_500= 3758.5810546875 runs/s
*RESULT prepare_tiles: 50_1000= 3686.30078125 runs/s

After:
*RESULT prepare_tiles: 2_100= 9835.1611328125 runs/s
*RESULT prepare_tiles: 2_500= 20325.25390625 runs/s
*RESULT prepare_tiles: 2_1000= 21301.900390625 runs/s
*RESULT prepare_tiles: 10_100= 16529.826171875 runs/s
*RESULT prepare_tiles: 10_500= 16585.15625 runs/s
*RESULT prepare_tiles: 10_1000= 16813.125 runs/s
*RESULT prepare_tiles: 50_100= 4764.2998046875 runs/s
*RESULT prepare_tiles: 50_500= 4804.51220703125 runs/s
*RESULT prepare_tiles: 50_1000= 4789.98486328125 runs/s

However, I'll also be changing the tests to actually reset this state
when I change the perftests to test specific scenarios.

R=enne
BUG=645569
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/3e9481d0d0a80567a1f16c10f59058377048bd03
Cr-Commit-Position: refs/heads/master@{#418662}

[vmpstr, 2016-09-14 18:22:22]

Description was changed from

==========
cc: Mark tiles as processed for solid color analysis.

This patch ensures we don't process tiles for solid color analysis
many times. This wasn't as much of an issue before we separated raster
and decode images, since we'd only process a tile once. Now, we can
visit tiles more than once.

This shows pretty significant gains on perftests.
Before:
*RESULT prepare_tiles: 2_100= 2236.82763671875 runs/s
*RESULT prepare_tiles: 2_500= 8168.21533203125 runs/s
*RESULT prepare_tiles: 2_1000= 10103.5654296875 runs/s
*RESULT prepare_tiles: 10_100= 8856.08984375 runs/s
*RESULT prepare_tiles: 10_500= 8581.669921875 runs/s
*RESULT prepare_tiles: 10_1000= 8932.677734375 runs/s
*RESULT prepare_tiles: 50_100= 3708.318359375 runs/s
*RESULT prepare_tiles: 50_500= 3758.5810546875 runs/s
*RESULT prepare_tiles: 50_1000= 3686.30078125 runs/s

After:
*RESULT prepare_tiles: 2_100= 9835.1611328125 runs/s
*RESULT prepare_tiles: 2_500= 20325.25390625 runs/s
*RESULT prepare_tiles: 2_1000= 21301.900390625 runs/s
*RESULT prepare_tiles: 10_100= 16529.826171875 runs/s
*RESULT prepare_tiles: 10_500= 16585.15625 runs/s
*RESULT prepare_tiles: 10_1000= 16813.125 runs/s
*RESULT prepare_tiles: 50_100= 4764.2998046875 runs/s
*RESULT prepare_tiles: 50_500= 4804.51220703125 runs/s
*RESULT prepare_tiles: 50_1000= 4789.98486328125 runs/s

However, I'll also be changing the tests to actually reset this state
when I change the perftests to test specific scenarios.

R=enne
BUG=645569
==========

to

==========
cc: Mark tiles as processed for solid color analysis.

This patch ensures we don't process tiles for solid color analysis
many times. This wasn't as much of an issue before we separated raster
and decode images, since we'd only process a tile once. Now, we can
visit tiles more than once.

This shows pretty significant gains on perftests.
Before:
*RESULT prepare_tiles: 2_100= 2236.82763671875 runs/s
*RESULT prepare_tiles: 2_500= 8168.21533203125 runs/s
*RESULT prepare_tiles: 2_1000= 10103.5654296875 runs/s
*RESULT prepare_tiles: 10_100= 8856.08984375 runs/s
*RESULT prepare_tiles: 10_500= 8581.669921875 runs/s
*RESULT prepare_tiles: 10_1000= 8932.677734375 runs/s
*RESULT prepare_tiles: 50_100= 3708.318359375 runs/s
*RESULT prepare_tiles: 50_500= 3758.5810546875 runs/s
*RESULT prepare_tiles: 50_1000= 3686.30078125 runs/s

After:
*RESULT prepare_tiles: 2_100= 9835.1611328125 runs/s
*RESULT prepare_tiles: 2_500= 20325.25390625 runs/s
*RESULT prepare_tiles: 2_1000= 21301.900390625 runs/s
*RESULT prepare_tiles: 10_100= 16529.826171875 runs/s
*RESULT prepare_tiles: 10_500= 16585.15625 runs/s
*RESULT prepare_tiles: 10_1000= 16813.125 runs/s
*RESULT prepare_tiles: 50_100= 4764.2998046875 runs/s
*RESULT prepare_tiles: 50_500= 4804.51220703125 runs/s
*RESULT prepare_tiles: 50_1000= 4789.98486328125 runs/s

However, I'll also be changing the tests to actually reset this state
when I change the perftests to test specific scenarios.

R=enne
BUG=645569
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[vmpstr, 2016-09-14 18:23:36]

Please take a look.

[enne (OOO), 2016-09-14 18:29:34]

lgtm

https://codereview.chromium.org/2334003007/diff/1/cc/tiles/tile.h
File cc/tiles/tile.h (right):

https://codereview.chromium.org/2334003007/diff/1/cc/tiles/tile.h#newcode155
cc/tiles/tile.h:155: bool is_solid_color_analysis_performed_ = false;
How about sticking this bool with the other bools where it can be packed better?

[vmpstr, 2016-09-14 18:48:13]

https://codereview.chromium.org/2334003007/diff/1/cc/tiles/tile.h
File cc/tiles/tile.h (right):

https://codereview.chromium.org/2334003007/diff/1/cc/tiles/tile.h#newcode155
cc/tiles/tile.h:155: bool is_solid_color_analysis_performed_ = false;
On 2016/09/14 18:29:33, enne wrote:
> How about sticking this bool with the other bools where it can be packed
better?

Done.

[vmpstr, 2016-09-14 18:48:18]

The CQ bit was checked by vmpstr@chromium.org

[vmpstr, 2016-09-14 18:48:18]

The patchset sent to the CQ was uploaded after l-g-t-m from enne@chromium.org
Link to the patchset: https://codereview.chromium.org/2334003007/#ps20001
(title: "update")

[commit-bot: I haz the power, 2016-09-14 18:49:18]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-14 20:41:18]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-09-14 20:43:24]

Description was changed from

==========
cc: Mark tiles as processed for solid color analysis.

This patch ensures we don't process tiles for solid color analysis
many times. This wasn't as much of an issue before we separated raster
and decode images, since we'd only process a tile once. Now, we can
visit tiles more than once.

This shows pretty significant gains on perftests.
Before:
*RESULT prepare_tiles: 2_100= 2236.82763671875 runs/s
*RESULT prepare_tiles: 2_500= 8168.21533203125 runs/s
*RESULT prepare_tiles: 2_1000= 10103.5654296875 runs/s
*RESULT prepare_tiles: 10_100= 8856.08984375 runs/s
*RESULT prepare_tiles: 10_500= 8581.669921875 runs/s
*RESULT prepare_tiles: 10_1000= 8932.677734375 runs/s
*RESULT prepare_tiles: 50_100= 3708.318359375 runs/s
*RESULT prepare_tiles: 50_500= 3758.5810546875 runs/s
*RESULT prepare_tiles: 50_1000= 3686.30078125 runs/s

After:
*RESULT prepare_tiles: 2_100= 9835.1611328125 runs/s
*RESULT prepare_tiles: 2_500= 20325.25390625 runs/s
*RESULT prepare_tiles: 2_1000= 21301.900390625 runs/s
*RESULT prepare_tiles: 10_100= 16529.826171875 runs/s
*RESULT prepare_tiles: 10_500= 16585.15625 runs/s
*RESULT prepare_tiles: 10_1000= 16813.125 runs/s
*RESULT prepare_tiles: 50_100= 4764.2998046875 runs/s
*RESULT prepare_tiles: 50_500= 4804.51220703125 runs/s
*RESULT prepare_tiles: 50_1000= 4789.98486328125 runs/s

However, I'll also be changing the tests to actually reset this state
when I change the perftests to test specific scenarios.

R=enne
BUG=645569
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
cc: Mark tiles as processed for solid color analysis.

This patch ensures we don't process tiles for solid color analysis
many times. This wasn't as much of an issue before we separated raster
and decode images, since we'd only process a tile once. Now, we can
visit tiles more than once.

This shows pretty significant gains on perftests.
Before:
*RESULT prepare_tiles: 2_100= 2236.82763671875 runs/s
*RESULT prepare_tiles: 2_500= 8168.21533203125 runs/s
*RESULT prepare_tiles: 2_1000= 10103.5654296875 runs/s
*RESULT prepare_tiles: 10_100= 8856.08984375 runs/s
*RESULT prepare_tiles: 10_500= 8581.669921875 runs/s
*RESULT prepare_tiles: 10_1000= 8932.677734375 runs/s
*RESULT prepare_tiles: 50_100= 3708.318359375 runs/s
*RESULT prepare_tiles: 50_500= 3758.5810546875 runs/s
*RESULT prepare_tiles: 50_1000= 3686.30078125 runs/s

After:
*RESULT prepare_tiles: 2_100= 9835.1611328125 runs/s
*RESULT prepare_tiles: 2_500= 20325.25390625 runs/s
*RESULT prepare_tiles: 2_1000= 21301.900390625 runs/s
*RESULT prepare_tiles: 10_100= 16529.826171875 runs/s
*RESULT prepare_tiles: 10_500= 16585.15625 runs/s
*RESULT prepare_tiles: 10_1000= 16813.125 runs/s
*RESULT prepare_tiles: 50_100= 4764.2998046875 runs/s
*RESULT prepare_tiles: 50_500= 4804.51220703125 runs/s
*RESULT prepare_tiles: 50_1000= 4789.98486328125 runs/s

However, I'll also be changing the tests to actually reset this state
when I change the perftests to test specific scenarios.

R=enne
BUG=645569
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/3e9481d0d0a80567a1f16c10f59058377048bd03
Cr-Commit-Position: refs/heads/master@{#418662}
==========

[commit-bot: I haz the power, 2016-09-14 20:43:25]

Patchset 2 (id:??) landed as
https://crrev.com/3e9481d0d0a80567a1f16c10f59058377048bd03
Cr-Commit-Position: refs/heads/master@{#418662}