Closed
Bug 1424047
Opened 7 years ago
Closed 7 years ago
ResolveFlattening is the hottest function during WR DisplayList building on facebook-refresh
Categories
(Core :: Graphics: WebRender, enhancement)
Core
Graphics: WebRender
Tracking
()
RESOLVED
INVALID
People
(Reporter: jrmuizel, Unassigned)
References
(Blocks 1 open bug)
Details
Attachments
(1 file)
(deleted),
text/x-review-board-request
|
Details |
https://perfht.ml/2nFUZ4T mstange thinks we don't need to spend so much time here
Comment hidden (mozreview-request) |
Comment 3•7 years ago
|
||
The container of the stack seems to be the reason. I don't see the ResolveFlattening in the gecko profiler[1] after using std::list, though I'm not sure why the AutoTArray is slower in this case. Jeff, can you check if the patch helps? Maybe ClearInvalidationStateBits has the same problem. I'll investigate more deeply. [1] https://perfht.ml/2Bihu60
Flags: needinfo?(ethlin)
Updated•7 years ago
|
Flags: needinfo?(jmuizelaar)
Updated•7 years ago
|
Whiteboard: [wr-mvp] [triage]
Reporter | ||
Comment 4•7 years ago
|
||
So I looked at this more closely in Instruments. It turns out this is just the cache miss that we get for visiting the next nsDisplayItem.
Status: NEW → RESOLVED
Closed: 7 years ago
Flags: needinfo?(jmuizelaar)
Resolution: --- → INVALID
Updated•7 years ago
|
Whiteboard: [wr-mvp] [triage]
Updated•2 years ago
|
Blocks: wr-displaylist-perf
You need to log in
before you can comment on or make changes to this bug.
Description
•