Closed Bug 846842 (gl-thread) Opened 12 years ago Closed 2 years ago

Offload OpenGL commands to a dedicated GL thread

Categories

(Core :: Graphics: CanvasWebGL, enhancement)

enhancement

Tracking

()

RESOLVED DUPLICATE of bug 1607940

People

(Reporter: jgilbert, Unassigned)

References

(Blocks 1 open bug)

Details

(Whiteboard: webgl-perf)

We're seeing non-trivial overhead for GL calls in WebGL-heavy demos. One way to help with this is to do the actual GL calls on a different thread, synchronizing as necessary. Most GL commands can be async, while others (bufferData, texImage2D) should often be sync.
(In reply to Jeff Gilbert [:jgilbert] from comment #0) > We're seeing non-trivial overhead for GL calls in WebGL-heavy demos. One way > to help with this is to do the actual GL calls on a different thread, > synchronizing as necessary. > > Most GL commands can be async, while others (bufferData, texImage2D) should > often be sync. FWIW, some OpenGL stacks already support this. i.e. on OS X you can opt in and on mesa there are patches to add support.
Great, sounds like we should try those out as well. Doing it ourselves provides a couple other potential advantages, but more immediate solutions (for even a subset of platforms) could be quite useful.
Blocks: 1268629
Whiteboard: webgl-perf
Severity: normal → enhancement
Severity: normal → S3
Component: Graphics → Graphics: CanvasWebGL
Status: NEW → RESOLVED
Closed: 2 years ago
Duplicate of bug: 1607940
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.