cc: optimize texture allocation.

cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the
primitive bounds, and the filter outsets. This means that
we may end up allocating many textures of different sizes,
defeating Skia's texture cache. Texture allocation can be
slow, so we can ameliorate the allocation pressure by
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Committed: https://crrev.com/d47a9ef38a0c81f0250e575f08fb47fb56fe1d66
Cr-Commit-Position: refs/heads/master@{#391753}

[Stephen White, 2016-04-12 22:14:51]

Description was changed from

==========
cc: optimize texture allocation.

BUG=602785
==========

to

==========
cc: optimize texture allocation.

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[Stephen White, 2016-04-12 22:14:58]

The CQ bit was checked by senorblanco@chromium.org

[Stephen White, 2016-04-12 22:15:02]

The CQ bit was unchecked by senorblanco@chromium.org

[Stephen White, 2016-04-12 22:15:08]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-04-12 22:15:32]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/1

[commit-bot: I haz the power, 2016-04-13 01:22:51]

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

[commit-bot: I haz the power, 2016-04-13 01:22:52]

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

[Stephen White, 2016-05-03 19:25:14]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-05-03 19:25:28]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/40001

[commit-bot: I haz the power, 2016-05-03 20:41:27]

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

[commit-bot: I haz the power, 2016-05-03 20:41:27]

Dry run: This issue passed the CQ dry run.

[Stephen White, 2016-05-04 21:14:10]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-05-04 21:14:30]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/80001

[Stephen White, 2016-05-04 21:19:16]

Description was changed from

==========
cc: optimize texture allocation.

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: optimize texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that we may end up asking
Skia to allocate many textures of different sizes, defeating Skia's texture
cache. Texture 
allocation can be slow, so we can ameliorate the allocation
pressure by rounding width and height up to the nearest power of 2. (This is
what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[Stephen White, 2016-05-04 21:19:16]

senorblanco@chromium.org changed reviewers:
+ enne@chromium.org

[Stephen White, 2016-05-04 21:19:53]

Description was changed from

==========
cc: optimize texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that we may end up asking
Skia to allocate many textures of different sizes, defeating Skia's texture
cache. Texture 
allocation can be slow, so we can ameliorate the allocation
pressure by rounding width and height up to the nearest power of 2. (This is
what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: optimize texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[Stephen White, 2016-05-04 21:20:04]

enne: PTAL. Thanks!

[danakj, 2016-05-04 21:24:57]

danakj@chromium.org changed reviewers:
+ danakj@chromium.org

[danakj, 2016-05-04 21:24:58]

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.cc
File cc/output/gl_renderer.cc (right):

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:145: int w, h;
you're using ints here but it's a RectF.

[Stephen White, 2016-05-04 21:27:32]

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.cc
File cc/output/gl_renderer.cc (right):

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:145: int w, h;
On 2016/05/04 21:24:58, danakj wrote:
> you're using ints here but it's a RectF.

Yep. There are implicit casts in the comparisons and the call to set_width() and
set_height(), but the multiplies are faster. :)

[enne (OOO), 2016-05-04 21:32:22]

enne@chromium.org changed reviewers:
- danakj@chromium.org

[enne (OOO), 2016-05-04 21:32:22]

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.cc
File cc/output/gl_renderer.cc (right):

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:146: for (w = 1; w < rect->width(); w *= 2) {
If you're casting to ints here for performance, might as well go all the way and
https://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 instead of
a loop.

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:1045: // Expand dst_rect size to the nearest power of
2, in order to get
Are there limits on the size that is supported here? Like, very big textures now
becoming ~2-4 times as big?

[Stephen White, 2016-05-04 21:43:57]

Description was changed from

==========
cc: optimize texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[Stephen White, 2016-05-04 21:47:17]

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.cc
File cc/output/gl_renderer.cc (right):

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:146: for (w = 1; w < rect->width(); w *= 2) {
On 2016/05/04 21:32:22, enne wrote:
> If you're casting to ints here for performance, might as well go all the way
and
> https://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 instead
of
> a loop.

Cute! I don't really care about performance here; I was just being annoying.
I've changed them to floats for clarity.

https://codereview.chromium.org/1884553003/diff/80001/cc/output/gl_renderer.c...
cc/output/gl_renderer.cc:1045: // Expand dst_rect size to the nearest power of
2, in order to get
On 2016/05/04 21:32:22, enne wrote:
> Are there limits on the size that is supported here? Like, very big textures
now
> becoming ~2-4 times as big?

If you mean, will the allocation fail due to an invalid texture dimension, I'm
pretty sure all GL texture limits are a power of 2, so no.

For overall size, these textures count against Skia's texture budget, so they
may cause other scratch textures to be evicted or reused, but they shouldn't
increase the VRAM footprint on balance.

(Note that the internal textures allocated for filter processing in Skia are
already rounded up the same way.)

[enne (OOO), 2016-05-04 21:49:21]

lgtm

[Stephen White, 2016-05-04 21:54:41]

The CQ bit was checked by senorblanco@chromium.org

[Stephen White, 2016-05-04 21:54:41]

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/1884553003/#ps120001
(title: "New baselines for rotated filters tests")

[commit-bot: I haz the power, 2016-05-04 21:55:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/120001

[Stephen White, 2016-05-04 23:35:26]

The CQ bit was checked by senorblanco@chromium.org

[Stephen White, 2016-05-04 23:35:26]

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/1884553003/#ps140001
(title: "win fix: futz with the fuzz")

[commit-bot: I haz the power, 2016-05-04 23:36:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/140001

[commit-bot: I haz the power, 2016-05-05 02:14:04]

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

[commit-bot: I haz the power, 2016-05-05 02:14:05]

Try jobs failed on following builders:
  linux_android_rel_ng on tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[Stephen White, 2016-05-05 03:37:47]

The CQ bit was checked by senorblanco@chromium.org

[commit-bot: I haz the power, 2016-05-05 03:38:03]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1884553003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1884553003/140001

[commit-bot: I haz the power, 2016-05-05 04:28:13]

Description was changed from

==========
cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[commit-bot: I haz the power, 2016-05-05 04:28:14]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2016-05-05 04:29:18]

Description was changed from

==========
cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the 
primitive bounds, and the filter outsets. This means that 
we may end up allocating many textures of different sizes, 
defeating Skia's texture cache. Texture allocation can be 
slow, so we can ameliorate the allocation pressure by 
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: optimize filter backing store texture allocation.

The backing store used for filters in cc is determined by the
primitive bounds, and the filter outsets. This means that
we may end up allocating many textures of different sizes,
defeating Skia's texture cache. Texture allocation can be
slow, so we can ameliorate the allocation pressure by
rounding width and height up to the nearest power of 2.
(This is what Skia does internally for scratch textures).

BUG=602785
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Committed: https://crrev.com/d47a9ef38a0c81f0250e575f08fb47fb56fe1d66
Cr-Commit-Position: refs/heads/master@{#391753}
==========

[commit-bot: I haz the power, 2016-05-05 04:29:19]

Patchset 8 (id:??) landed as
https://crrev.com/d47a9ef38a0c81f0250e575f08fb47fb56fe1d66
Cr-Commit-Position: refs/heads/master@{#391753}