| Summary: | [radeonsi] Gimp OpenCL does not work | ||||||
|---|---|---|---|---|---|---|---|
| Product: | DRI | Reporter: | darkbasic <darkbasic> | ||||
| Component: | DRM/Radeon | Assignee: | Default DRI bug account <dri-devel> | ||||
| Status: | RESOLVED MOVED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | medium | CC: | vedran | ||||
| Version: | XOrg git | ||||||
| Hardware: | Other | ||||||
| OS: | All | ||||||
| Whiteboard: | |||||||
| i915 platform: | i915 features: | ||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 99553 | ||||||
| Attachments: |
|
||||||
|
Description
darkbasic
2014-02-14 10:57:33 UTC
Please provide more information about the crash; at least the stderr output from gimp, preferably with R600_DEBUG=cs. Created attachment 94348 [details] dump I noticed it suddently started working flawlessly so I asked myself what fixed it: patches from https://bugs.freedesktop.org/show_bug.cgi?id=72785 did fix it. I currently apply: 0001-SelectionDAG-Factor-ISD-MUL-lowering-code-out-of-DAG.patch 0002-SelectionDAG-Use-helper-function-to-improve-legaliza.patch 0001-R600-SI-Custom-select-64-bit-ADD.patch Beware that I modified "UP.Threshold = 500" to "UP.Threshold = 100" because "bfgminer --scrypt" didn't work otherwise. I attached dump from R600_DEBUG=cs GEGL_USE_OPENCL=yes gimp without the patches. Sorry I forgot one patch: https://bugs.freedesktop.org/attachment.cgi?id=94078 This is the one with "UP.Threshold = 100" instead of "UP.Threshold = 500" does GIMP use OpenGL other than GEGL library? Not sure, but I don't think so. AFAIK it' only GEGL-related stuff. (In reply to darkbasic from comment #5) > Not sure, but I don't think so. AFAIK it' only GEGL-related stuff. a lot of gegl tests pass OK for me on Turks. notably clone test that includes box-blur and gaussian-blur operations passes OK. Can you retest with the latest llvm/mesa? I will be able to test within a couple of months, at the moment I don't have access to a pc with such hardware. (In reply to darkbasic from comment #7) > I will be able to test within a couple of months, at the moment I don't have > access to a pc with such hardware. Is this still an issue? gegl testsuite mostly passes with only 3 remaining tests: contrast-curve -- 7 wrong pixels clones -- 1 wrong pixel noise-cell -- fail (many wrong pixels, I think their algorithm is wrong) comment #2 also mentions that this works OK -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/amd/issues/435. |
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.