1630629 - Disable WebRender on Intel battery if we don't have DirectComposition

Status: VERIFIED FIXED

(Core :: Graphics: WebRender, enhancement)

Description

[Jeff Muizelaar [:jrmuizel]]

DirectComposition is important for battery usage on Intel. If bug 1630371 doesn't make it into beta, and perhaps even if it does, we should consider doing this.

Andrew do have thoughts on the best way to do this?

Comment 1

[Jeff Muizelaar [:jrmuizel]]

Attachment: Bug 1630629. Block WebRender on Intel if no DComp.

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/923c7f562468
Block WebRender on Intel if no DComp. r=aosmond

Comment 3

[Andreea Pavel [:apavel]]

Backed out as requested: https://hg.mozilla.org/integration/autoland/rev/1688dd8b01d3859ce89ca8790cd8ea30efd1ae88

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/836f649a6b09
Block WebRender on Intel if no DComp. r=aosmond

Comment 5

[Jeff Muizelaar [:jrmuizel]]

Comment on attachment 9141065 [details]
Bug 1630629. Block WebRender on Intel if no DComp.

Beta/Release Uplift Approval Request

• User impact if declined: 67 enables WebRender on some Intel laptops (bug 1615049). We want an additional limit on this group that DirectComposition is required.

DirectComposition gives performance and battery life improvements so we want to limit the risk of bug 1615049 by only shipping to user with it.

• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: No
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: Bug 1630371
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): This should be a risk reduction from what is currently in beta.
• String changes made/needed:

Comment 6

[Stefan Hindli [:stefan_hindli]]

https://hg.mozilla.org/mozilla-central/rev/836f649a6b09

Comment 7

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 9141065 [details]
Bug 1630629. Block WebRender on Intel if no DComp.

Only enables WebRender on Intel machines if DComp is also enabled. Approved for 76.0b6.

Comment 8

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

https://hg.mozilla.org/releases/mozilla-beta/rev/22b249597944

Comment 9

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Backed out on jrmuizel's request: https://hg.mozilla.org/releases/mozilla-beta/rev/d64d5ccf2b14498b6de0bb18c4e9e66abae5efce

Comment 10

[Jeff Muizelaar [:jrmuizel]]

Attachment: Bug 1630629. Block WebRender on Intel if no DComp.

This moves the DirectComposition configuration code before
the WebRender configuration code so that we can depend
on the results to decide whether to user WebRender.

It keeps the check UseWebRenderDCompWin after the initialization
of WebRender so that it's properly initialized when we read it.

Comment 11

[Jeff Muizelaar [:jrmuizel]]

Comment on attachment 9141695 [details]
Bug 1630629. Block WebRender on Intel if no DComp.

Beta/Release Uplift Approval Request

• User impact if declined: 67 enables WebRender on some Intel laptops (bug 1615049). We want an additional limit on this group that DirectComposition is required.
  DirectComposition gives performance and battery life improvements so we want to limit the risk of bug 1615049 by only shipping to user with it.
• Is this code covered by automated tests?: No
• 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): As seen in the original landing, because the ordering of some code needed to change this could introduce mis-configuration. This new version combines the patch from bug 1631312 which has been confirmed to fix the problem originally introduced by this patch.

Despite the above, the goal of this patch is a risk reduction from what is currently in beta.

• String changes made/needed:

Comment 12

[Ryan VanderMeulen [:RyanVM]]

Can QA please try to verify this fix on Nightly before uplift? Given the previous issues with this patch, it would be good to verify that everything is in working order now.

Comment 13

[Ryan VanderMeulen [:RyanVM]]

Comment on attachment 9141695 [details]
Bug 1630629. Block WebRender on Intel if no DComp.

I really wish QA would have been able to give this a quick verification on Nightly before uplifting this to Beta, but I also don't want this to wait until Friday to be uplifted. Approved for 76.0b7 now so we can get some extra testing. That said, if this bounces again, I think we should revisit what WR rollout changes we're planning to ship in 76 rather than attempting any more fixes in place.

Comment 14

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-beta/rev/7e8fc3932c44

Comment 15

[Alexandru Ionescu (needinfo me) [:alexandrui]]

== Change summary for alert #25672 (as of Wed, 22 Apr 2020 02:59:56 GMT) ==

Improvements:

22% tscrollx windows10-64-shippable-qr opt e10s stylo 1.06 -> 0.83
15% glterrain windows10-64-shippable-qr opt e10s stylo 1.83 -> 1.55
10% tart windows10-64-shippable-qr opt e10s stylo 2.82 -> 2.54
10% tart windows10-64-shippable-qr opt e10s stylo 2.83 -> 2.56

For up to date results, see: https://treeherder.mozilla.org/perf.html#/alerts?id=25672

Comment 16

[Rares Doghi, Desktop QA]

This issue is verified as fixed in Beta 76.0b7 on Windows 10.

Comment 17

[Rares Doghi, Desktop QA]

This is not Disabled in Nightly, Nightly still shows Webrender even without Direct Composition, I think it is intended. @Jeff ?

Comment 18

[Jeff Muizelaar [:jrmuizel]]

yes, it is intended.

Comment 19

[Rares Doghi, Desktop QA]

I Will update the flags for 77 based on comment 17 and Comment 18, It seems the fix is there but it will only be visible when it reaches beta.