Summary: | Artifacts in CSGO map de_cache when shader level is "high" or "very high (default)" | ||
---|---|---|---|
Product: | Mesa | Reporter: | Kristian Nyborg Dahl <krdahl> |
Component: | Drivers/Gallium/radeonsi | Assignee: | Default DRI bug account <dri-devel> |
Status: | RESOLVED FIXED | QA Contact: | Default DRI bug account <dri-devel> |
Severity: | normal | ||
Priority: | medium | CC: | network723 |
Version: | 17.2 | ||
Hardware: | x86-64 (AMD64) | ||
OS: | Linux (All) | ||
See Also: | https://github.com/ValveSoftware/csgo-osx-linux/issues/1567 | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: | screenshot of artifacts |
Description
Kristian Nyborg Dahl
2017-12-10 11:28:43 UTC
This was kisak-valve's comment on their github: "Hello @Heis, on my AMD test box, mesa 17.2.6 built against llvm 5.0.0 is affected, while llvm 4.0.1 is not affected. This should also be brought to the attention of your driver vendor if it has not been already." I had same issue on my RX480 (bug 103056), but it appears the video card had a dead VRAM chip. Now I'm back to HD7950 with amdgpu kernel driver, and it looks like TAHITI card is not affected. I can confirm seeing the same artifacts in CSGO on a R9 285 (Tonga) using Ubuntu 17.10 with Mesa 17.3.2, LLVM 5.0 (Padoka Stable PPA). *** Bug 103056 has been marked as a duplicate of this bug. *** (In reply to Kristian Nyborg Dahl from comment #1) > This was kisak-valve's comment on their github: > > "Hello @Heis, on my AMD test box, mesa 17.2.6 built against llvm 5.0.0 is > affected, while llvm 4.0.1 is not affected. > > This should also be brought to the attention of your driver vendor if it has > not been already." I just tested with latest available mesa (17.3.7) and llvm (6.0.0.4). Seems to be solved. (In reply to Kristian Nyborg Dahl from comment #5) > (In reply to Kristian Nyborg Dahl from comment #1) > > This was kisak-valve's comment on their github: > > > > "Hello @Heis, on my AMD test box, mesa 17.2.6 built against llvm 5.0.0 is > > affected, while llvm 4.0.1 is not affected. > > > > This should also be brought to the attention of your driver vendor if it has > > not been already." > > I just tested with latest available mesa (17.3.7) and llvm (6.0.0.4). Seems > to be solved. Great! Thanks for retesting. Is it a good idea to close this bug? There seem to be a number of bugs affected by a problem with llvm 5.0. Unfortunatly, LLVM 5.0 series releases no stable builds, so a fix cant be backported there. I guess it would be also too hard to code somekind of workaround in mesa. Maybe restrict the use of LLVM 5.0 in stable releases? |
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.