Package Details: opencl-amd 1:6.4.0-1

Git Clone URL: https://aur.archlinux.org/opencl-amd.git (read-only, click to copy)
Package Base: opencl-amd
Description: ROCm components repackaged from AMD's Ubuntu releases (ROCr runtime, ROCm runtime, HIP runtime) - This package is intended to work along with the free amdgpu stack.
Upstream URL: http://www.amd.com
Keywords: amd amdgpu computing gpgpu opencl radeon
Licenses: custom:AMD
Conflicts: amd-smi-lib, comgr, hip, hip-dev, hip-doc, hip-runtime-amd, hip-samples, hipcc, hsa-amd-aqlprofile, hsa-rocr, hsa-rocr-dev, libdrm-amdgpu-amdgpu1, openmp-extras-runtime, rocdecode, rocdecode-dev, rocjpeg, rocjpeg-dev, rocm-cmake, rocm-core, rocm-dbgapi, rocm-debug-agent, rocm-device-libs, rocm-gdb, rocm-hip-runtime, rocm-language-runtime, rocm-ocl-icd, rocm-opencl, rocm-opencl-dev, rocm-opencl-icd-loader, rocm-opencl-runtime, rocm-smi-lib, rocm-utils, rocminfo, rocprofiler, rocprofiler-dev, rocprofiler-plugins, rocprofiler-register, roctracer, roctracer-dev
Provides: amd-smi-lib, comgr, hip, hip-dev, hip-doc, hip-runtime-amd, hip-samples, hipcc, hsa-amd-aqlprofile, hsa-rocr, hsa-rocr-dev, libdrm-amdgpu-amdgpu1, opencl-driver, openmp-extras-runtime, rocdecode, rocdecode-dev, rocjpeg, rocjpeg-dev, rocm-cmake, rocm-core, rocm-dbgapi, rocm-debug-agent, rocm-device-libs, rocm-gdb, rocm-hip-runtime, rocm-language-runtime, rocm-ocl-icd, rocm-opencl, rocm-opencl-dev, rocm-opencl-icd-loader, rocm-opencl-runtime, rocm-smi-lib, rocm-utils, rocminfo, rocprofiler, rocprofiler-dev, rocprofiler-plugins, rocprofiler-register, roctracer, roctracer-dev
Submitter: grmat
Maintainer: sperg512 (luciddream)
Last Packager: luciddream
Votes: 133
Popularity: 0.46
First Submitted: 2016-12-01 03:45 (UTC)
Last Updated: 2025-04-11 22:54 (UTC)

Required by (132)

Sources (38)

Pinned Comments

nho1ix commented on 2023-12-29 08:43 (UTC) (edited on 2024-02-10 07:13 (UTC) by nho1ix)

Note for anyone who has a Polaris GPU (Radeon RX 5xx) debugging issues with this package; Packages that use OpenCL like clinfo or davinci-resolve-studio will need you to downgrade opencl-amd to 1:5.7.1-1 as well as amdgpu-pro-oglp to 23.10_1620044-1 to avoid coredumps & segfaults.

DVR would not open unless these 2 packages were downgraded (along with their dependencies). Had to figure it out the hard way after hours using valgrind and rebooting over and over. Hopefully someone else will not have to pull their hair out trying to resolve their issue.

Latest Comments

« First ‹ Previous 1 .. 38 39 40 41 42 43 44 45 46 47 48 .. 81 Next › Last »

sperg512 commented on 2021-04-22 03:41 (UTC)

@iKevin shouldn't be necessary. Since they're so out-of-date, I don't even know if it's any different than the ones in extra. Later I'll probably see if I can diff the one in the tarball and an equally old version of the "official" headers. No promises for when or if I'll do it, like I said I'm good at lying.

Arup commented on 2021-04-22 00:03 (UTC)

The latest one works fine here with RX570 card, clinfo returns all the right parameters as well as Darktable.

iKevin commented on 2021-04-21 19:52 (UTC)

Hi @sperg512. I should be more clear. By "off to the races" I meant I can get going. I haven't tried to compile yet. Also, as you pointed out, I noticed the opencl-headers in the regular repo and they are out of sync with this package.

Perhaps there could be a separate AUR package with the c/c++ headers called opencl-amd-dev that depends on this package.

luciddream commented on 2021-04-21 19:03 (UTC)

I guess it compiles because it doesn't really need it. The headers are already included in the project.

I just compared opencl-headers package with the one from amdgpu-pro package, it seems that amdgpu-pro CL files are from 2019, and has no support for OpenCL 3.0.

So I'm not sure if there is any benefit in including them with opencl-amd

sperg512 commented on 2021-04-21 18:35 (UTC) (edited on 2021-04-21 18:36 (UTC) by sperg512)

Apparently I'm really good at lying because I forgot to test last night, anyways here's my results:

  • geekbench acts really weird, with >=2 windows open, it freezes most stuff and my entire display just completely spergs out. With 1 window open though, it just exits (status 255)

  • Blender and LibreOffice with OCL on just segfault

  • Detected by clinfo just fine

@hpohl I couldn't really figure anything of significance out from that. I think I'll send that to the devs to see if they can.

@iKevin @luciddream OCL headers are in opencl-headers package, with C++ stuff available in opencl-clhpp. I have no clue how you got that to compile (maybe you have one of those installed?), but either way maybe we could add those as optdeps or something.

@apaz That's normal. Seems to be doing that for pretty much everyone's.

iKevin commented on 2021-04-21 18:03 (UTC)

Thanks @luciddream. You solved the problem in a different way. I was trying to build it and cmake was complaining that it couldn't find opencl due to the includes files not being found. I am off to the races!

luciddream commented on 2021-04-21 17:43 (UTC) (edited on 2021-04-21 17:52 (UTC) by luciddream)

@iKevin I made some tests, although keep in mind I've never worked with C or C++. I'm not sure if the header files are necessary.

All I did was add ${CMAKE_DL_LIBS} to target_link_libraries, ran ccmake ., set OpenCL_LIBRARY=/usr/lib/libamdocl64.so, set OpenCL_INCLUDE_DIR to something random, generate makefile, and ran make.

Then it looks that it is identifying the GPU correctly. I have no idea how to run this though.

luciddream@arch ~/p/VerthashMiner (main)> ./VerthashMiner -l
[2021-04-21 20:43:09] INFO  Found 1 OpenCL devices.

Device list:
==================
OpenCL devices:
    Index: 0. Name: gfx1010:xnack+
              Platform index: 0
              Platform name: Advanced Micro Devices, Inc.
              pcieId: 0d:00:0

iKevin commented on 2021-04-21 16:23 (UTC)

Hi @luciddream. My use case would be to compile a local copy of verthash-miner.

apaz commented on 2021-04-21 16:01 (UTC)

Everything is fine for me with the new drivers (RX 5700XT). The only change is that clinfo reports "Device Name gfx1010:xnack+" while previously it reported only "gfx1010". I don't know what that means.

luciddream commented on 2021-04-21 15:20 (UTC) (edited on 2021-04-21 15:20 (UTC) by luciddream)

@iKevin we actually don't install the header files at the moment. But there are some header files inside the archive, and we can make tests to see if they are useful for OpenCL / HIP compilation. Do you have a specific case in mind? I was thinking using something like Pytorch for ROCM as a test project, but that's for HIP.

« First ‹ Previous 1 .. 38 39 40 41 42 43 44 45 46 47 48 .. 81 Next › Last »