It looks like I am not the only user who has difficulties in getting OpenCL up and running. Surely, if a system is discovered, during the installation period, that the hardware is OpenCL capable, then should it not be an intrinsic part of the basic installation process?
Moderator action: Moved from Graphics & Display to Feedback as there is only a vague statement and rhetorical question
OpenCL, Rocm or Cuda are mainly used by developers to access the GPU directly through a low-level API. So no, the is no need to install it for every system. Devs usually know how to do it.
If we interpret the original post, for the benefit of them and others, as needing guidance with OpenCL then most everything needed to know should be available at the archwiki; https://wiki.archlinux.org/title/GPGPU#OpenCL
Megavolt should know that OpenCl is used in an increasing number of common graphic applications. This has not been a dev-only procedure for some years.
You are right. Point taken. But it is still a library that is used by very specific programs. Take Blender, for example. For all programs known to me that can use OpenCL, it is optional and it also works without it. Accordingly, as with blender, the libraries are optional:
$ pacman -Si blender
Repository : extra
Name : blender
Version : 17:4.3.2-7
Description : A fully integrated 3D graphics creation suite
Architecture : x86_64
URL : https://www.blender.org
Licenses : Apache-2.0 BSD-2-Clause BSD-3-Clause GPL-2.0-or-later
GPL-3.0-or-later LGPL-2.1-or-later MIT MPL-2.0 Zlib
Groups : None
Provides : None
Depends On : alembic bash boost-libs draco embree expat ffmpeg fftw
freetype2 gcc-libs glew glibc gmp hicolor-icon-theme imath
intel-oneapi-compiler-dpcpp-cpp-runtime-libs
intel-oneapi-compiler-shared-runtime-libs jack jemalloc
level-zero-loader libepoxy libharu libjpeg-turbo libpng
libsndfile libspnav libtiff libwebp libx11 libxfixes libxi
libxkbcommon libxml2 libxrender libxxf86vm llvm-libs
materialx onetbb openal opencolorio openexr openimagedenoise
openimageio openjpeg2 openpgl openshadinglanguage opensubdiv
openvdb openxr potrace pugixml pystring python python-numpy
python-requests sdl2 shared-mime-info usd xdg-utils yaml-cpp
zlib zstd
Optional Deps : cuda: Cycles renderer CUDA support
intel-compute-runtime: Cycles renderer Intel OneAPI support
hip-runtime-amd: Cycles renderer AMD ROCm support
hiprt: Ray tracing AMD ROCm support
libdecor: wayland support
Conflicts With : None
Replaces : None
Download Size : 147.40 MiB
Installed Size : 429.67 MiB
Packager : Caleb Maclennan <alerque@archlinux.org>
Build Date : Wed Jan 22 12:45:08 2025
Validated By : MD5 Sum SHA-256 Sum Signature
I don’t think it bothers to integrate each of these on the ISO via MHWD, but it’s more a question of disk space, as cuda would also be ~2GB (decompressed ~5GB) extra. The other libraries for AMD and INTEL are manageably small.