New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 670100 link

Starred by 2 users

Issue metadata

Status: ExternalDependency
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 2
Type: Bug



Sign in to add a comment

conformance/extensions/angle-instanced-arrays.html on Intel HD 6000 on macOS 10.12

Project Member Reported by kainino@chromium.org, Nov 30 2016

Issue description

Chrome Version: Canary 57.0.2937.0

The test conformance/extensions/angle-instanced-arrays.html

fails on:
* macOS 10.12.1 on Intel HD 6000
* macOS 10.12.2 beta 4 on Intel HD 6000 (different machine)

does NOT fail on:
* macOS 10.11.? on Intel HD 5000
* macOS 10.12.1 on Intel HD 5000 (same machine)
* macOS 10.12.1 on AMD FirePro D500 (zmo's)

Seems specific to Intel HD 6000.

Haven't been able to test on macOS 10.11.x on HD 6000, so I don't know if it's a macOS regression.
 
Cc: yunchao...@intel.com qiankun....@intel.com
Owner: kbr@chromium.org
Status: Assigned (was: Untriaged)
Assigning to kbr since this likely requires a radar to be filed.

+yunchao.he +qiankun.miao since this is Intel-specific.

Comment 2 by kbr@chromium.org, Dec 2 2016

Owner: qiankun....@intel.com
Qiankun, Yunchao, could one of you please triage this and prepare a Radar template if necessary? Do you have a machine on which you could reproduce this? (Does it happen with the newer GPUs in the MacBook and MacBook Pro?)

Comment 3 by yang...@intel.com, Dec 2 2016

Cc: yang...@intel.com
I tested this on our Macbook at hand. I didn't see this failure. Configs of the two Macbooks are Mac OSX Sierra (10.12) with MacBook Pro-Mid 2015-Intel Iris Pro Graphics 5200 (GEN7) and MacBook-Early 2016-Intel HD Graphics 515 (GEN9).

This 2.0.1 failure persists on macOS 10.12.4 beta 2 (16E154a), HD 6000.

Comment 6 by yang...@intel.com, Feb 24 2017

Why did you test this case against 2.0.1? I found it's only valid for WebGL 1.
In file sdk/tests/conformance/extensions/00_test_list.txt:
--min-version 1.0.3 --max-version 1.9.9 angle-instanced-arrays.html

Comment 7 by kbr@chromium.org, Mar 6 2017

Cc: kainino@chromium.org
Owner: kainino@chromium.org
Sorry for not picking this up again.

Kai, could you please revisit this particular failure? Intel rightly pointed out that this test shouldn't be tested against WebGL 2.0 contexts in the first place. Maybe that was just a mistake in updating this bug.

We need to know exactly what's failing in the test on the affected hardware. The failure log would be helpful.

Sorry for dropping this, it's been a long time.

> Why did you test this case against 2.0.1?

I think that may have been a typo. It still fails on Intel HD 6000 under WebGL 1.0, macOS 10.12.4, Chrome 60.0.3096.0.

Testing drawArraysInstancedANGLE with param 'first' > 0
FAIL should be 255,0,0,255
at (13, 25) expected: 255,0,0,255 was 0,0,0,0
FAIL should be 0,255,0,255
at (38, 25) expected: 0,255,0,255 was 0,0,0,0
FAIL should be 0,0,255,255
at (13, 0) expected: 0,0,255,255 was 0,0,0,0
PASS should be 255,255,0,255

Testing that the 'first' parameter to drawArraysInstancedANGLE is only an offset into the non-instanced vertex attributes.
PASS getError was expected value: NO_ERROR : drawArraysInstancedANGLE should succeed
PASS getError was expected value: NO_ERROR : drawArraysInstancedANGLE should succeed
FAIL should be 255,0,0,255
at (25, 25) expected: 255,0,0,255 was 0,0,0,0

I'll have it on my todo list to file a radar for this.
Owner: ----
Status: ExternalDependency (was: Assigned)
Filed as Radar 32212685

Sign in to add a comment