CUDA 6.0 RPM on Fedora 19

I’m trying to install CUDA 6.0 using the RPM on Fedora 19. I ran into some problems that I’m trying to solve.

  1. First I downloaded this from the download page:
    http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/cuda-repo-fedora19-6.0-37.x86_64.rpm

I have no idea what checksum to expect since unlike the checksum for the .run file, the checksum for the RPM doesn’t seem to be listed on the website. Anyway, the md5sum and sha256sum of the file I downloaded are respectively 58c327682ff19470dcb21136804d0e34 and c254e8f4ea104c3a40236b0451776c48a66b17d34290719f8367d18c5fd39320.

Well, I “yum localinstall” the RPM anyway, and then “yum install cuda”, upon which I noticed that:

  1. The “cuda” package depends on the “cuda-5-5” package, which is odd since this is supposed to be CUDA 6.

  2. I got this error:
    Transaction check error:
    file /usr/share/man/man1/nvidia-xconfig.1.gz conflicts between attempted installs of nvidia-xconfig-319.37-27.fc18.x86_64 and xorg-x11-drv-nvidia-1:331.67-1.fc19.x86_64
    file /usr/bin/nvidia-settings conflicts between attempted installs of nvidia-settings-319.37-30.fc18.x86_64 and xorg-x11-drv-nvidia-1:331.67-1.fc19.x86_64
    file /usr/share/applications/nvidia-settings.desktop conflicts between attempted installs of nvidia-settings-319.37-30.fc18.x86_64 and xorg-x11-drv-nvidia-1:331.67-1.fc19.x86_64
    file /usr/share/man/man1/nvidia-settings.1.gz conflicts between attempted installs of nvidia-settings-319.37-30.fc18.x86_64 and xorg-x11-drv-nvidia-1:331.67-1.fc19.x86_64

  3. Fine, I try disabling the rpmfusion repo:
    yum install cuda --disablerepo=rpmfusion*

Then I got these errors:
Downloading packages:
No Presto metadata available for cuda
nvidia-modprobe-319.37-1.fc18. FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/nvidia-modprobe-319.37-1.fc18.x86_64.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
nvidia-kmod-319.37-1.fc18.x86_ FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/nvidia-kmod-319.37-1.fc18.x86_64.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
xorg-x11-drv-nvidia-319.37-2.f FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/xorg-x11-drv-nvidia-319.37-2.fc18.x86_64.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
xorg-x11-drv-nvidia-devel-319. FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/xorg-x11-drv-nvidia-devel-319.37-2.fc18.i686.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
xorg-x11-drv-nvidia-devel-319. FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/xorg-x11-drv-nvidia-devel-319.37-2.fc18.x86_64.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
xorg-x11-drv-nvidia-libs-319.3 FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/xorg-x11-drv-nvidia-libs-319.37-2.fc18.i686.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.
xorg-x11-drv-nvidia-libs-319.3 FAILED
http://developer.download.nvidia.com/compute/cuda/repos/fedora19/x86_64/xorg-x11-drv-nvidia-libs-319.37-2.fc18.x86_64.rpm: [Errno 14] HTTP Error 404 - Not Found ] 0.0 B/s | 0 B --:–:-- ETA
Trying other mirror.

Error downloading packages:
1:xorg-x11-drv-nvidia-devel-319.37-2.fc18.x86_64: [Errno 256] No more mirrors to try.
1:xorg-x11-drv-nvidia-libs-319.37-2.fc18.x86_64: [Errno 256] No more mirrors to try.
1:xorg-x11-drv-nvidia-319.37-2.fc18.x86_64: [Errno 256] No more mirrors to try.
1:xorg-x11-drv-nvidia-libs-319.37-2.fc18.i686: [Errno 256] No more mirrors to try.
1:xorg-x11-drv-nvidia-devel-319.37-2.fc18.i686: [Errno 256] No more mirrors to try.
nvidia-modprobe-319.37-1.fc18.x86_64: [Errno 256] No more mirrors to try.
1:nvidia-kmod-319.37-1.fc18.x86_64: [Errno 256] No more mirrors to try.

So, how should I proceed? Just use the RUN file? It’s not an RPM, so given the results I’ve been having with the RPM, I’m afraid that if the RUN installation fails, it will leave a big mess on my machine.

It doesn’t look like you followed the directions in the getting started guide:

http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/index.html#package-manager-installation

I managed to install CUDA 6.0 and then run some samples. All I was missing was the “sudo yum clean expire-cache” However, Problem 1 still stands. I still can’t find the checksum to make sure I downloaded the correct RPM content.