Bug 166675 - Provide a setting for clients to always prefer low-power WebGL
Summary: Provide a setting for clients to always prefer low-power WebGL
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Tim Horton
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2017-01-04 00:09 PST by Tim Horton
Modified: 2017-01-04 11:02 PST (History)
12 users (show)

See Also:

Attachments
Patch (8.92 KB, patch)
2017-01-04 00:10 PST, Tim Horton 		mitz: review+
Details | Formatted Diff | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Tim Horton 2017-01-04 00:09:57 PST 
Provide a setting for clients to always prefer low-power WebGL
Comment 1 Tim Horton 2017-01-04 00:10:22 PST 
Created attachment 297996 [details]
Patch
Comment 2 mitz 2017-01-04 00:39:56 PST 
Comment on attachment 297996 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=297996&action=review

> Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:403
> +    if (frame->settings().preferLowPowerWebGLRendering())
> +        attributes.preferLowPowerToHighPerformance = true;

Should this check the current value of preferLowPowerToHighPerformance before consulting Settings, like we do with the antialias attribute above?
Comment 3 Tim Horton 2017-01-04 00:43:46 PST 
Comment on attachment 297996 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=297996&action=review

>> Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:403
>> +        attributes.preferLowPowerToHighPerformance = true;
> 
> Should this check the current value of preferLowPowerToHighPerformance before consulting Settings, like we do with the antialias attribute above?

I guess it could? Just to save the cost of asking Settings?
Comment 4 Tim Horton 2017-01-04 00:56:46 PST 
https://trac.webkit.org/changeset/210273
Comment 5 Tim Horton 2017-01-04 00:57:06 PST 
<rdar://problem/29834093>
Comment 6 Myles C. Maxfield 2017-01-04 10:46:39 PST 
Comment on attachment 297996 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=297996&action=review

>>> Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:403
>>> +        attributes.preferLowPowerToHighPerformance = true;
>> 
>> Should this check the current value of preferLowPowerToHighPerformance before consulting Settings, like we do with the antialias attribute above?
> 
> I guess it could? Just to save the cost of asking Settings?

Why do you call the internal setting "prefer?" The implementation seems a lot like "force" or "supports" to me.
Comment 7 Tim Horton 2017-01-04 11:02:38 PST 
(In reply to comment #6)
> Comment on attachment 297996 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=297996&action=review
> 
> >>> Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:403
> >>> +        attributes.preferLowPowerToHighPerformance = true;
> >> 
> >> Should this check the current value of preferLowPowerToHighPerformance before consulting Settings, like we do with the antialias attribute above?
> > 
> > I guess it could? Just to save the cost of asking Settings?
> 
> Why do you call the internal setting "prefer?" The implementation seems a
> lot like "force" or "supports" to me.

Because the whole thing is a "prefer" (if the discrete card is up, we'll still use that). It might be "force prefer", but it's definitely not just "force-use low power".