Open Bug 1560972 Opened 5 years ago Updated 2 years ago

profile with both webrender on and off

Categories

(Firefox Build System :: General, enhancement)

enhancement

Tracking

(Not tracked)

People

(Reporter: froydnj, Unassigned)

References

Details

(Keywords: in-triage)

Attachments

(2 files)

One of these codepaths is not getting optimized with our normal profiling runs. We should probably try to fix that.

What's needed to do that? Do we just have another step in profileserver.py to re-run the test suite with a different Firefox profile?

(In reply to Michael Shal [:mshal] from comment #1)

What's needed to do that? Do we just have another step in profileserver.py to re-run the test suite with a different Firefox profile?

I think that's pretty much it. The only open question is whether there is some webrender-specific benchmark that would be useful here, but I think jrmuizel said starting the browser should be good enough.

I tried this out on both Linux & Windows, but I didn't see any changes in talos benchmarks. Is there a way to verify that webrender is actually being used? And are there any specific tests that we would expect to see change?

What kind of method are you looking for? about:support will say if WebRender is being used in the Compositor section.

tsvgx gives different results depending on whether WebRender is on or off:
https://treeherder.mozilla.org/perf.html#/graphs?series=mozilla-central,1662821,1,1&series=mozilla-central,1538058,1,1

Flags: needinfo?(mshal)

Sorry I never got back to this. I'll post the patches I was testing with in case someone else wants to pick it up. Maybe there's something obvious I'm missing...

Flags: needinfo?(mshal)

Depends on D66663

Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: