Bug 33824

Summary:	[r600g, tiling] mipmap rendering errors / block artifacts
Product:	Mesa	Reporter:	Tobias Jakobi <liquid.acid>
Component:	Drivers/Gallium/r600	Assignee:	Default DRI bug account <dri-devel>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	medium	CC:	tdroste
Version:	git
Hardware:	x86-64 (AMD64)
OS:	Linux (All)
Whiteboard:
i915 platform:		i915 features:

Description Tobias Jakobi 2011-02-01 13:22:45 UTC

Hi there,

this is a update of bug #31046, where the main issue was GPU resets caused by enabling tiling (R600_FORCE_TILING=1) with the r600g driver.

The situation however was changed, and now R600_FORCE_TILING only causes mipmap rendering errors in a few application. Last time I tested ioquake3, doom3 (demo) and both ut2003 and ut2004.

ut03 and ut04 are still affected by the blocky rendering errors (see the other bug for screenshots illustrating the problem), but both ioquake3 and doom3 now look perfectly fine.

My hardware:
ATI Technologies Inc Radeon HD 4770 [RV740]

libdrm, mesa and xf86-video-ati are all git master tip.
ColorTiling is enabled in the xorg.conf, verified this by checking the Xorg
logfile.
kernel is d-r-t

Greets,
Tobias

Comment 1 Tobias Jakobi 2011-05-06 17:26:22 UTC

Can no longer reproduce this issue. So it looks fixed.

Comment 2 Tobias Droste 2011-05-07 05:07:33 UTC

It's still there. Tiling is just disabled by default and the environment variable changed to R600_TILING. Just try the piglit test fbo-generatemipmap-formats with: R600_TILING=on piglit/bin/fbo-generatemipmap-formats and switch to NPOT textures with 'd' and go through various test sets ('n'). Some of them are still rendered wrong.

Comment 3 Tobias Jakobi 2011-05-07 06:07:31 UTC

Yeah, you're right -- I completly forgot that the envvar changed :(

Comment 4 Tobias Jakobi 2011-05-07 09:14:58 UTC

Retested with the correct envvar and I still can't reproduce the artifacts. They don't appear anymore in ut2003 and ut2004, which were the only remaining games on my list.

Leaving this open, since Droste still seems to have issues (I didn't check these tests).

Comment 5 Tobias Droste 2011-05-08 20:05:38 UTC

Could you test nexuiz (just to check that it's not my fault in some way)? It should be visible there too. I can't test ut2k3/2k4.

Comment 6 Tobias Jakobi 2011-05-11 17:04:17 UTC

No, also no errors visible there. Which kernel do you use? Using the drm-next branch here.

Comment 7 Tobias Droste 2011-05-11 17:09:45 UTC

Currently I'm running drm-fixes (built ~1h ago). Most probably it's only visible on some cards (evergreen/juniper here).

(libdrm, mesa and ddx were also updated to current git master ~1h ago)

Comment 8 Tobias Jakobi 2011-05-11 17:22:47 UTC

I'm closing this one then, since it was primarly about the R700 series and the artifacts I experienced in games using the ut engine.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.