Arch Linux User Repository

Can you add to the package description that this is Intel and NVIDIA/Cuda backend only? (No AMD backend)

Regards!

Ahoj,

can you please remove the MAKEFLAGS="-j$(nproc) --no-print-directory" line from the PKGBUILD?

This overrides user-set MAKEFLAGS (e.g. in /etc/makepkg.conf).

The user should be able to decide the make behaviour.

If you really need --no-print-directory or -j1(!), then use MAKEFLAGS+= to append but not to replace. But do not use -j$(nproc) in a PKGBUILD -- if a user wants parallel build, the user usually sets that in e.g. /etc/makepkg.conf, and in other cases a user maybe explicitly wants non-parallel builds, or only -j2, or so.

Regards and thanks for the package!

I installed it after installing cuda but the backend was not enabled. I think these cmake parameters are missing:

-DCMAKE_CUDA_COMPILER=/opt/cuda/bin/nvcc
-DCMAKE_CUDA_COMPILER_TOOLKIT_ROOT=/opt/cuda/

@Sleipnir: helper_cuda.h is part of NVIDIA's cuda-samples. We don't have that packaged anywhere.

There's an error during build:

-- Configuring done
-- Generating done
-- Build files have been written to: /home/josh/.config/trizen/sources/gpufetch-git/src/gpufetch/build
[ 15%] Building CXX object CMakeFiles/cuda_backend.dir/src/cuda/uarch.cpp.o
[ 20%] Building CXX object CMakeFiles/intel_backend.dir/src/intel/uarch.cpp.o
[ 20%] Building CXX object CMakeFiles/cuda_backend.dir/src/cuda/cuda.cpp.o
[ 25%] Building CXX object CMakeFiles/cuda_backend.dir/src/cuda/pci.cpp.o
[ 15%] Building CXX object CMakeFiles/intel_backend.dir/src/intel/intel.cpp.o
[ 30%] Building CXX object CMakeFiles/intel_backend.dir/src/intel/pci.cpp.o
[ 35%] Building CXX object CMakeFiles/intel_backend.dir/src/intel/cpuid.cpp.o
/home/josh/.config/trizen/sources/gpufetch-git/src/gpufetch/src/cuda/cuda.cpp:1:10: fatal error: helper_cuda.h: No such file or directory
    1 | #include <helper_cuda.h>
      |          ^~~~~~~~~~~~~~~
compilation terminated.
make[2]: *** [CMakeFiles/cuda_backend.dir/build.make:76: CMakeFiles/cuda_backend.dir/src/cuda/cuda.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
[ 40%] Building CXX object CMakeFiles/intel_backend.dir/src/intel/udev.cpp.o
/home/josh/.config/trizen/sources/gpufetch-git/src/gpufetch/src/cuda/uarch.cpp:2:10: fatal error: helper_cuda.h: No such file or directory
    2 | #include <helper_cuda.h>
      |          ^~~~~~~~~~~~~~~
compilation terminated.
make[2]: *** [CMakeFiles/cuda_backend.dir/build.make:90: CMakeFiles/cuda_backend.dir/src/cuda/uarch.cpp.o] Error 1
[ 45%] Linking CXX static library libintel_backend.a
make[1]: *** [CMakeFiles/Makefile2:140: CMakeFiles/cuda_backend.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
[ 45%] Built target intel_backend
make: *** [Makefile:136: all] Error 2
==> ERROR: A failure occurred in build().
    Aborting...
:: Unable to build gpufetch-git - makepkg exited with code: 4

Let me know if you need any more information. This was run on a Lenovo P51 with hybrid proprietary drivers (Intel & NVidia)

thank you.

@yochananmarqos thanks for pointing it out, I didn't know it. I have changed the PKGBUILD and removed make from makedepends.

@theduckchannel thanks for the warning, I have just fixed it (gpufetch now uses CMake instead of Make)

Failed to build:

make: *** No targets specified and no makefile found. Stop. ==> ERROR: A failure occurred in build(). Aborting... error: failed to build 'gpufetch-git-0.10.r0.gf4a96ce-1': error: packages failed to build: gpufetch-git-0.10.r0.gf4a96ce-1

base-devel group packages should not be included in makedepends(), see https://wiki.archlinux.org/title/PKGBUILD#makedepends