Add ANGLE workaround to prefer flushes over VRAM usage

Add ANGLE workaround to prefer flushes over VRAM usage

On the whole, https://codereview.chromium.org/1286203002/ (Defer flushes if kPreferNoIO is specified) improved performance but it did cause a performance regression on ANGLE. This CL disables the deferral of flushes on ANGLE until we can add a separate incremental flushing mechanism.

TBR=bsalomon@google.com

BUG=skia:4201
BUG=521529
Committed: https://skia.googlesource.com/skia/+/63926683c583e8497d9d907977e773663cb4bd9e

[robertphillips, 2015-08-20 14:19:03]

robertphillips@google.com changed reviewers:
+ joshualitt@google.com, jvanverth@google.com

[joshualitt, 2015-08-20 15:35:20]

On 2015/08/20 14:20:06, robertphillips wrote:

lgtm

[jvanverth1, 2015-08-20 15:37:50]

LGTM + suggestion

https://codereview.chromium.org/1287193008/diff/1/src/gpu/GrResourceCache.cpp
File src/gpu/GrResourceCache.cpp (right):

https://codereview.chromium.org/1287193008/diff/1/src/gpu/GrResourceCache.cpp...
src/gpu/GrResourceCache.cpp:264: // We would always like to defer a flush in
favor of consuming more available
We would prefer to consume more available VRAM rather than flushing immediately,
but on ANGLE [...]

[robertphillips, 2015-08-20 16:30:58]

robertphillips@google.com changed reviewers:
+ bsalomon@google.com

[robertphillips, 2015-08-20 16:31:04]

https://codereview.chromium.org/1287193008/diff/1/src/gpu/GrResourceCache.cpp
File src/gpu/GrResourceCache.cpp (right):

https://codereview.chromium.org/1287193008/diff/1/src/gpu/GrResourceCache.cpp...
src/gpu/GrResourceCache.cpp:264: // We would always like to defer a flush in
favor of consuming more available
On 2015/08/20 15:37:49, jvanverth1 wrote:
> We would prefer to consume more available VRAM rather than flushing
immediately,
> but on ANGLE [...]

Done.

[robertphillips, 2015-08-20 16:31:11]

The CQ bit was checked by robertphillips@google.com to run a CQ dry run

[commit-bot: I haz the power, 2015-08-20 16:31:26]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1287193008/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1287193008/20001

[commit-bot: I haz the power, 2015-08-20 16:37:56]

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

[commit-bot: I haz the power, 2015-08-20 16:37:57]

Dry run: This issue passed the CQ dry run.

[robertphillips, 2015-08-20 16:38:12]

The CQ bit was checked by robertphillips@google.com

[robertphillips, 2015-08-20 16:38:13]

The patchset sent to the CQ was uploaded after l-g-t-m from
jvanverth@google.com, joshualitt@google.com
Link to the patchset: https://codereview.chromium.org/1287193008/#ps20001
(title: "Update comment")

[commit-bot: I haz the power, 2015-08-20 16:38:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1287193008/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1287193008/20001

[commit-bot: I haz the power, 2015-08-20 16:39:05]

Committed patchset #2 (id:20001) as
https://skia.googlesource.com/skia/+/63926683c583e8497d9d907977e773663cb4bd9e

[bsalomon, 2015-08-21 15:41:31]

https://codereview.chromium.org/1287193008/diff/20001/src/gpu/GrCaps.cpp
File src/gpu/GrCaps.cpp (right):

https://codereview.chromium.org/1287193008/diff/20001/src/gpu/GrCaps.cpp#newc...
src/gpu/GrCaps.cpp:118: fPreferVRAMUseOverFlushes = true;
Hmm... our other workarounds are worded the other way where false means don't
apply the workaround. Otherwise lgtm.

[robertphillips, 2015-08-21 15:47:15]

https://codereview.chromium.org/1287193008/diff/20001/src/gpu/GrCaps.cpp
File src/gpu/GrCaps.cpp (right):

https://codereview.chromium.org/1287193008/diff/20001/src/gpu/GrCaps.cpp#newc...
src/gpu/GrCaps.cpp:118: fPreferVRAMUseOverFlushes = true;
On 2015/08/21 15:41:31, bsalomon wrote:
> Hmm... our other workarounds are worded the other way where false means don't
> apply the workaround. Otherwise lgtm.

I did it that way at first and found it a bit confusing.