Summary: | force_glsl_extensions_warn=true breaks Unigine Sanctuary & Tropics demos when GL_ARB_gpu_shader5 exposed | ||
---|---|---|---|
Product: | Mesa | Reporter: | zhoujian <jianx.zhou> |
Component: | glsl-compiler | Assignee: | Chris Forbes <chrisf> |
Status: | RESOLVED WONTFIX | QA Contact: | Intel 3D Bugs Mailing List <intel-3d-bugs> |
Severity: | major | ||
Priority: | medium | CC: | 375gnu, christophe.prigent, eero.t.tamminen, lilix.cheng, wendy.wang |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux (All) | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: | GL errors from Sanctuary apitrace trace |
Description
zhoujian
2014-08-21 08:22:38 UTC
Why is this a bug? To clarify, are you seeing any change in the rendering as well, or is the performance just suspiciously better? The Unigine demos are using the force_glsl_extensions_warn driconf workaround, which now enables GL_ARB_gpu_shader5, turning 'sample' into a reserved word. So a bunch of the shaders fail to compile. *** Bug 82900 has been marked as a duplicate of this bug. *** Created attachment 105083 [details]
GL errors from Sanctuary apitrace trace
Attached are first GL errors output from Sanctuary apitrace trace (in total, error log file was tens of MBs).
The issue also exists on mesa-10.3-rc1 testing. ------------------------------------------------ kernel: 3.16.1 Mesa: (10.3)git-627d31d Xf86_video_intel:2.99.914 Libdrm:libdrm-2.4.56 Libva: 1.3.1 vaapi-intel-driver: 1.3.2 Cairo: 1.12.16 Xserver:xorg-server-1.16.0 This affects only Sanctuary & Tropics, Heaven and Valley seem to work fine. Yeah, the workarounds are only necessary for the older, discontinued, extremely non-compliant version of the Unigine engine. The new Unigine engine used in Heaven and Valley is actually quite good, and doesn't need so many workarounds. That's why I keep advocating for dropping Sanctuary and Tropics. They're demos for an engine I haven't seen used anywhere except those demos, and you can't even buy it anymore. Since Unigine Sanctuary and Tropics are "broken" benchmarks, no bug that only shows up on these benchmarks will be considered P1 going forward. Heaven and Valley bugs, however, will be considered for P1 designation. *** Bug 91778 has been marked as a duplicate of this bug. *** I think it's probably safe to close this as won't fix. |
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.