Closed Bug 757642 Opened 12 years ago Closed 5 years ago

WebGL antialiasing: un-blacklist "ATI Technologies Inc. -- ATI Radeon HD 6770M OpenGL Engine -- 2.1 ATI-7.18.18"

Categories

(Core :: Graphics: CanvasWebGL, defect, P1)

defect

Tracking

()

RESOLVED FIXED
mozilla69
Tracking Status
firefox69 --- fixed

People

(Reporter: ehsan.akhgari, Assigned: jgilbert)

References

(Blocks 1 open bug)

Details

(Whiteboard: webgl-driver)

Attachments

(1 file, 1 obsolete file)

I set the webgl.msaa-force to true, and ran a bunch of WebGL pages, with no apparent problems.  The renderings were correctly anti-aliased.

My about:support gfx section:

Graphics
      

      
        
        Vendor ID0x1002Device ID0x6740WebGL RendererATI Technologies Inc. -- ATI Radeon HD 6770M OpenGL Engine -- 2.1 ATI-7.18.18GPU Accelerated Windows4/4 OpenGL

        
        AzureBackendquartz
Summary: Please un-blacklist → Please un-blacklist "ATI Technologies Inc. -- ATI Radeon HD 6770M OpenGL Engine -- 2.1 ATI-7.18.18"
Blocks: gecko-games
Summary: Please un-blacklist "ATI Technologies Inc. -- ATI Radeon HD 6770M OpenGL Engine -- 2.1 ATI-7.18.18" → WebGL antialiasing: un-blacklist "ATI Technologies Inc. -- ATI Radeon HD 6770M OpenGL Engine -- 2.1 ATI-7.18.18"
I forgot to ask: what's your OSX version?
10.7.4.
Please run the 1.0.1 conformance tests and detail which failures are found:
https://cvs.khronos.org/svn/repos/registry/trunk/public/webgl/conformance-suites/1.0.1/webgl-conformance-tests.html
No longer blocks: gecko-games
I also have a Macbook Pro with the 6770M running 10.7.4. Here's the conformance test failures with today's nightly:

WebGL Conformance Test Results
Version 1.0.1

-------------------

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:16.0) Gecko/16.0 Firefox/16.0
WebGL VENDOR: Mozilla
WebGL VERSION: WebGL 1.0
WebGL RENDERER: Mozilla
Unmasked VENDOR: undefined
Unmasked RENDERER: undefined
WebGL R/G/B/A/Depth/Stencil bits (default config): 8/8/8/8/32/0

-------------------

Test Summary (8881 total tests):
Tests PASSED: 8766
Tests FAILED: 115
Tests TIMED OUT: 0

-------------------

Failures:

conformance/attribs/gl-vertexattribpointer-offsets.html: 60 tests failed
conformance/glsl/functions/glsl-function-distance.html: 2 tests failed
conformance/glsl/functions/glsl-function-dot.html: 2 tests failed
conformance/glsl/functions/glsl-function-length.html: 2 tests failed
conformance/glsl/variables/gl-pointcoord.html: 48 tests failed
conformance/programs/program-test.html: 1 tests failed
Attached patch Patch v1 (obsolete) — Splinter Review
Here's the conformance tests run with this patch:

WebGL Conformance Test Results
Version 1.0.1

-------------------

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:16.0) Gecko/16.0 Firefox/16.0
WebGL VENDOR: Mozilla
WebGL VERSION: WebGL 1.0
WebGL RENDERER: Mozilla
Unmasked VENDOR: undefined
Unmasked RENDERER: undefined
WebGL R/G/B/A/Depth/Stencil bits (default config): 8/8/8/8/32/0

-------------------

Test Summary (8881 total tests):
Tests PASSED: 8766
Tests FAILED: 115
Tests TIMED OUT: 0

-------------------

Failures:

conformance/attribs/gl-vertexattribpointer-offsets.html: 60 tests failed
conformance/glsl/functions/glsl-function-distance.html: 2 tests failed
conformance/glsl/functions/glsl-function-dot.html: 2 tests failed
conformance/glsl/functions/glsl-function-length.html: 2 tests failed
conformance/glsl/variables/gl-pointcoord.html: 48 tests failed
conformance/programs/program-test.html: 1 tests failed
Assignee: nobody → jon
Status: NEW → ASSIGNED
Attachment #637622 - Flags: review?(jgilbert)
Sounds good, but we need to dig up the reason this was blocked to begin with, and check that this driver is unaffected by the original issue.
(In reply to Jeff Gilbert [:jgilbert] from comment #6)
> Sounds good, but we need to dig up the reason this was blocked to begin
> with, and check that this driver is unaffected by the original issue.

I believe Chrome had a big block list because they did limited testing and wanted to play it safe. We followed suit. It seems they have since relaxed it.
(In reply to Jeff Muizelaar [:jrmuizel] from comment #7)
> (In reply to Jeff Gilbert [:jgilbert] from comment #6)
> > Sounds good, but we need to dig up the reason this was blocked to begin
> > with, and check that this driver is unaffected by the original issue.
> 
> I believe Chrome had a big block list because they did limited testing and
> wanted to play it safe. We followed suit. It seems they have since relaxed
> it.

That was largely for the driver dates, iirc. I believe the issues we were having with ATI+Mac+AA were texture corruption issues.
Interesting. So they switched from selectively blocking cards to just blocking ATI on OSX < 10.7.2.

We should at least allow ATI cards on versions newer than that. We should consider matching their criteria and blocklisting all ATI on older OSX.
Comment on attachment 637622 [details] [diff] [review]
Patch v1

Review of attachment 637622 [details] [diff] [review]:
-----------------------------------------------------------------

We need to unblock OSX >=10.7.2 and ATI, not add a universal exception for this card. That is, unless we can dig up the testcases for the MSAA bugs, and this card passes. bjacob might better remember where these are.
Attachment #637622 - Flags: review?(jgilbert) → review-
Assignee: jon → nobody
Status: ASSIGNED → NEW
Whiteboard: webgl-driver
Priority: -- → P3
Assignee: nobody → jgilbert
Priority: P3 → P1
Blocks: 1372168

Previously we were hardcoding to disable on AMD. Modern MacOS versions
don't have MSAA corruption, and anything we find that does should go on
the downloadable blocklist instead.

Attachment #637622 - Attachment is obsolete: true
Pushed by jgilbert@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/fb974fbba61b
Rely on downloadable blocklist for Mac MSAA disabling. r=jrmuizel
Status: NEW → RESOLVED
Closed: 5 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla69
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: