Closed Bug 1555956 Opened 3 months ago Closed 3 months ago

Don't let D3D11 double-buffering ride the trains yet

Categories

(Core :: Graphics, defect, P2)

68 Branch
defect

Tracking

()

RESOLVED FIXED
mozilla69
Tracking Status
firefox67.0.1 --- unaffected
firefox68 --- fixed
firefox69 --- fixed

People

(Reporter: kats, Assigned: kats, NeedInfo)

References

Details

Attachments

(1 file)

Bug 1547775 turned on D3D11 double-buffering, but it seems to have caused some regressions that need investigation. Until those issues are sorted out we shouldn't let it ride the trains. See https://bugzilla.mozilla.org/show_bug.cgi?id=1555098#c6

Priority: -- → P2

Comment on attachment 9068955 [details]
Bug 1555956 - Disable double buffering on non-Nightly. r?jrmuizel

Beta/Release Uplift Approval Request

  • User impact if declined: Users may be affected by bug 1555098 and/or bug 1554610, both of which are regressions (direct or indirect) from bug 1547775 which turned on D3D11 double-buffering.
  • Is this code covered by automated tests?: Unknown
  • Has the fix been verified in Nightly?: No
  • Needs manual test from QE?: No
  • If yes, steps to reproduce:
  • List of other uplifts needed: None
  • Risk to taking this patch: Low
  • Why is the change risky/not risky? (and alternatives if risky): This is a low-risk patch to prevent the double-buffering from riding the trains as it produces regressions that are not yet understood.
  • String changes made/needed: none
Attachment #9068955 - Flags: approval-mozilla-beta?
Pushed by kgupta@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/0bb05d6474a9
Disable double buffering on non-Nightly. r=jrmuizel
Status: NEW → RESOLVED
Closed: 3 months ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla69

Comment on attachment 9068955 [details]
Bug 1555956 - Disable double buffering on non-Nightly. r?jrmuizel

disable gfx.direct3d11.use-double-buffering for 68.0b7

Attachment #9068955 - Flags: approval-mozilla-beta? → approval-mozilla-beta+

Considering the very large perf win (and power win) double buffering gives us (i.e. scrolling goes from being awful to being good on my P52), can we figure out how many D3D11 users this causes us to lose and whether it's the right tradeoff? Possibly just disabling double buffering for those users where it causes issues?

Flags: needinfo?(kats)
Flags: needinfo?(jmuizelaar)

This is actually only an issue when Advanced Layers is blacklisted for some reason, Advanced Layers already dealt with this situation.

Thanks for the patch. Any thoughts on bug 1554610?

Flags: needinfo?(kats)

(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #9)

Thanks for the patch. Any thoughts on bug 1554610?

Not without a machine where I can reproduce it :-(. Although I don't think it's bad enough that it should block double buffering.

(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #9)

Thanks for the patch. Any thoughts on bug 1554610?

My best guess here is that this is a debug feature in VirtualBox that is accidentally leaking in somewhere, they might put some kind of debugging info for their graphics driver development for double buffering in that single pixel. The easiest solution in my mind would be to disable double buffering on VirtualBox, or for that matter, perhaps blacklist it by default altogether, I'm not certain they win much by it.

(In reply to Bas Schouten (:bas.schouten) from comment #11)

The easiest solution in my mind would be to disable double buffering on VirtualBox, or for that matter, perhaps blacklist it by default altogether, I'm not certain they win much by it.

Could you write that patch?

You need to log in before you can comment on or make changes to this bug.