Upgrade to cuda 11
drahnr opened this issue · comments
Currently the repository contains cuda version 10.0, but it would be great to have 11.0.
Are there any plans to update the cuda* packages to the latest version for fedora 32?
Ref negativo17/compat-nvidia-repo#1
Ref fff-rs/juice#114
Ref fff-rs/juice#115
Yes I will, hopefully soon. I've been asked to help with the Blender build in Fedora, will do just after as I'm building Blender with CUDA support as part of the CUDA packages.
Do you think you will find time for this soon~ish? Much appreciated!
Sorry, moved house and got a few weeks without internet. Catching up now.
As you can see it's not a small change: 239c733
I will rewrite this commit as I make progress.
I am happy to test this early on once you have test packages to grab :)
Pushing the updated 11.1.1 now. The Java applications Nsight Compute and Nsight Systems are no longer part of the base CUDA package, they will be packaged separately (not sure though).
There is conflicting reports in the documentation, here it says GCC 10 is supported:
And here is not:
Can you confirm that GCC 10 works with nvcc from CUDA 11.1.1? I haven't built the cuda-gcc package yet for Fedora 33.
I will be able to test this towards the weekend. Thanks for all the work you put into this! 🙇
I will be able to test this towards the weekend. Thanks for all the work you put into this!
Any feedback? Does it work properly with GCC 10?
I tested with the quicksort sample on a pascal gpu, it compiles fine and runs fine.
$ g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/x86_64-redhat-linux/10/lto-wrapper
OFFLOAD_TARGET_NAMES=nvptx-none
OFFLOAD_TARGET_DEFAULT=1
Target: x86_64-redhat-linux
Configured with: ../configure --enable-bootstrap --enable-languages=c,c++,fortran,objc,obj-c++,ada,go,d,lto --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --with-bugurl=http://bugzilla.redhat.com/bugzilla --enable-shared --enable-threads=posix --enable-checking=release --enable-multilib --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-gnu-unique-object --enable-linker-build-id --with-gcc-major-version-only --with-linker-hash-style=gnu --enable-plugin --enable-initfini-array --with-isl --enable-offload-targets=nvptx-none --without-cuda-driver --enable-gnu-indirect-function --enable-cet --with-tune=generic --with-arch_32=i686 --build=x86_64-redhat-linux
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 10.2.1 20201016 (Red Hat 10.2.1-6) (GCC)
../../bin/x86_64/linux/release/cdpSimpleQuicksort
GPU Device 0: "Pascal" with compute capability 6.1
Initializing data:
Running quicksort on 128 elements
Launching kernel on the GPU
Validating results: OK
The samples currently come with makefiles that expect nvcc in a different path, so you might want to adjust that, to just use the one from $PATH and strip all the dependencies on the compat-gcc.
Sorry for the long delay.
@scaronni ^^^^
Dependency on cuda-gcc has been removed.