Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(341)

Issue 544853002: cc: Remove static from leaf invalidation perf test (Closed)

Created:
6 years, 3 months ago by enne (OOO)
Modified:
6 years, 3 months ago
Reviewers:
danakj
CC:
cc-bugs_chromium.org, chromium-reviews
Base URL:
https://chromium.googlesource.com/chromium/src.git@master
Project:
chromium
Visibility:
Public.

Description

cc: Remove static from leaf invalidation perf test Make DidCommitAndDraw always change the value of the layer's opacity to force commits and draws rather than depending on a static bool. LayerTreeHostPerfTestLeafInvalidates had a static bool for whether to flip a layer to 1.0 or 0.5 opacity. The leaf layer happens to start at opacity 1.0. So, on the first run, the tree would draw once, DidCommitAndDraw would get called, and opacity would be set to the same value. This would cause no damage, so the tree would never draw again. Because it would never draw again, DidCommitAndDraw wouldn't get called again, and the opacity would never change again. The tree kept committing (because of the test harness) but would never draw again. The second test would do the same thing, draw once, and then encounter this static boolean (now false). It would flip the opacity to zero (an actual change), and would then keep alternating back and forth because each draw would succeed. Therefore, any test that was fast would always be followed by a slow test, and any slow test would have a 50% chance of being followed by a fast or a slow test. This was exposed by r291451 because it made perf tests use the scheduler (and would then early out of drawing for single-threaded). The threaded version always had this bug. *** Note: unsurprisingly, this will "regress" the performance of LayerTreeHostPerfTestLeafInvalidates.TenTenSingleThread because it's now doing something real. R=danakj@chromium.org BUG=407117 Committed: https://crrev.com/81145970e2b1c0ceff32af1de3239c48cb4b1d7e Cr-Commit-Position: refs/heads/master@{#293408}

Patch Set 1 #

Unified diffs Side-by-side diffs Delta from patch set Stats (+2 lines, -3 lines) Patch
M cc/trees/layer_tree_host_perftest.cc View 1 chunk +2 lines, -3 lines 0 comments Download

Messages

Total messages: 6 (1 generated)
enne (OOO)
6 years, 3 months ago (2014-09-04 22:09:16 UTC) #1
danakj
> Therefore, any test that was fast would always be followed by a slow > ...
6 years, 3 months ago (2014-09-04 22:28:09 UTC) #2
commit-bot: I haz the power
CQ is trying da patch. Follow status at https://chromium-status.appspot.com/cq/enne@chromium.org/544853002/1
6 years, 3 months ago (2014-09-04 22:36:44 UTC) #4
commit-bot: I haz the power
Committed patchset #1 (id:1) as c314017f49920b346b713e457c2c856226cfdb45
6 years, 3 months ago (2014-09-05 02:36:26 UTC) #5
commit-bot: I haz the power
6 years, 3 months ago (2014-09-10 03:35:49 UTC) #6
Message was sent while issue was closed.
Patchset 1 (id:??) landed as
https://crrev.com/81145970e2b1c0ceff32af1de3239c48cb4b1d7e
Cr-Commit-Position: refs/heads/master@{#293408}

Powered by Google App Engine
This is Rietveld 408576698