cc: Improve performance of PaintOpBuffer::Playback with no ranges.

cc: Improve performance of PaintOpBuffer::Playback with no ranges.

In this case we were constructing 2 vectors on each call to playback()
which causes us to do malloc/free twice as well. Instead, define a
static vector and pass that which recovers basically all the lost time.

The other changes in NextOp try to improve the codegen, and seem to
maybe have up to 1ms improvement of the remaining delta of 2ms from
before cfc477b43f1c91e84e17180ee937728d6f496aa6, but it's in the
noise. However I think it's nicer with less nesting and only writing
the operator++ once anyways.

BUG=726031
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2905383002
Cr-Commit-Position: refs/heads/master@{#475936}
Committed: https://chromium.googlesource.com/chromium/src/+/4deb94b53a551930b3db578c4c339ec0ef95bc6b

[danakj, 2017-05-26 22:01:07]

Description was changed from

==========
cc: Improve performance of PaintOpBuffer::Playback with no ranges.

In this case we were constructing 2 vectors on each call to playback()
which causes us to do malloc/free twice as well. Instead, define a
static vector and pass that which recovers basically all the lost time.

The other changes in NextOp try to improve the codegen, and seem to
maybe have up to 1ms improvement of the remaining delta of 2ms from
before cfc477b43f1c91e84e17180ee937728d6f496aa6, but it's in the
noise. However I think it's nicer with less nesting and only writing
the operator++ once anyways.

BUG=726031
==========

to

==========
cc: Improve performance of PaintOpBuffer::Playback with no ranges.

In this case we were constructing 2 vectors on each call to playback()
which causes us to do malloc/free twice as well. Instead, define a
static vector and pass that which recovers basically all the lost time.

The other changes in NextOp try to improve the codegen, and seem to
maybe have up to 1ms improvement of the remaining delta of 2ms from
before cfc477b43f1c91e84e17180ee937728d6f496aa6, but it's in the
noise. However I think it's nicer with less nesting and only writing
the operator++ once anyways.

BUG=726031
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

[danakj, 2017-05-26 22:01:22]

danakj@chromium.org changed reviewers:
+ enne@chromium.org, vmpstr@chromium.org

[danakj, 2017-05-26 22:01:30]

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

[commit-bot: I haz the power, 2017-05-26 22:02:38]

Dry run: 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, 2017-05-26 23:25:02]

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

[commit-bot: I haz the power, 2017-05-26 23:25:03]

Dry run: This issue passed the CQ dry run.

[enne (OOO), 2017-05-29 23:15:21]

lgtm

[danakj, 2017-05-31 15:05:38]

The CQ bit was checked by danakj@chromium.org

[commit-bot: I haz the power, 2017-05-31 15:05:54]

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, 2017-05-31 16:25:40]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1496243138782490, "parent_rev":
"5d5052f3e33d06bf1f6710a8e1a94fd972e9c83e", "commit_rev":
"4deb94b53a551930b3db578c4c339ec0ef95bc6b"}

[commit-bot: I haz the power, 2017-05-31 16:25:58]

Description was changed from

==========
cc: Improve performance of PaintOpBuffer::Playback with no ranges.

In this case we were constructing 2 vectors on each call to playback()
which causes us to do malloc/free twice as well. Instead, define a
static vector and pass that which recovers basically all the lost time.

The other changes in NextOp try to improve the codegen, and seem to
maybe have up to 1ms improvement of the remaining delta of 2ms from
before cfc477b43f1c91e84e17180ee937728d6f496aa6, but it's in the
noise. However I think it's nicer with less nesting and only writing
the operator++ once anyways.

BUG=726031
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
==========

to

==========
cc: Improve performance of PaintOpBuffer::Playback with no ranges.

In this case we were constructing 2 vectors on each call to playback()
which causes us to do malloc/free twice as well. Instead, define a
static vector and pass that which recovers basically all the lost time.

The other changes in NextOp try to improve the codegen, and seem to
maybe have up to 1ms improvement of the remaining delta of 2ms from
before cfc477b43f1c91e84e17180ee937728d6f496aa6, but it's in the
noise. However I think it's nicer with less nesting and only writing
the operator++ once anyways.

BUG=726031
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2905383002
Cr-Commit-Position: refs/heads/master@{#475936}
Committed:
https://chromium.googlesource.com/chromium/src/+/4deb94b53a551930b3db578c4c33...
==========

[commit-bot: I haz the power, 2017-05-31 16:25:59]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/chromium/src/+/4deb94b53a551930b3db578c4c33...