I am unable to use CUDA-10 on Tumbleweed. Earlier I installed CUDA from http://developer.download.nvidia.com/compute/cuda/repos/opensuse15/x86_64 but it broke the xorg due to NVIDIAG04/05 driver mismatch, later I had to uninstall CUDA and reinstall NVIDIA drivers. Please let me know what I need to do in order to get it working. I’ve logs, but they are large and can’t post here or make a paste.
What’s stopping you from doing a pastebin? Do you have the susepaste rpm installed? If it doesn’t seem to work, there are many places you can upload using a web browser, including:
http://susepaste.org/
http://paste.opensuse.org/
http://pastebin.com/
https://gist.github.com/
http://paste.debian.net/
http://paste.fedoraproject.org
http://paste.ubuntu.com/
Also, it’s not enough to simply state where you downloaded CUDA from…
You also need to provide links to whatever guide you’re using and a detailed description or the error itself.
A quick search turned up the following nVidia Forums thread that might provide links to two different procedures installing CUDA 9.2, bot resulting in successes. Of course you may need to modify slightly for TW
TSU
Later in the evening I will do some regex on logs, filter unique lines and post. I repeated the procedures several times and now there is a lot of garbage in log files.
There is a possibility of compiling docker container, I am exploring that at the moment. I think there is some issue with YaST NVIDIA drivers and CUDA.
I followed these documents https://docs.nvidia.com/pdf/CUDA_Installation_Guide_Linux.pdf ,CUDA Installation Guide for Linux
If your intention is to run cuda in a docker container, why would you even bother with compiling unless you wanted some special, really oddball version?
If you’re just interested in running the latest, then the image should be pre-built and availaible for you… you only have to download and run it.
Here is the web search for cuda builds on Docker Hub
https://hub.docker.com/search?q=cuda&type=image
Once you’ve downloaded it, you would then create a dockerfile to customize it for your own purposes…
If you haven’t used docker before, I have some easy, quick articles in my Wiki that can get your toes wet… the articles are the standard docker tutorials long ago, and modified for openSUSE tools and common methods. Nowadays, docker is more complex but you still have to know the basics of how to find an image on Docker Hub, download and run it, and create a dockerfile. The article on how to access a console in the container will probably still work, but is superceded by a different method docker introduced later after the article was written. Skim down to the docker section in my Wiki ToC, and concentrate on the Installation and dockerfile articles
https://en.opensuse.org/User:Tsu2
Keep in mind that a container image downloaded from Docker Hub may contain a stub of a specific distro which makes it portable, and can run on any distro. This means that if the image you choose is for instance build on a Debian stub instead of openSUSE, it will mean that your dockerfile modifications may require installing from Debian sources. If you want to customize your container using openSUSE sources, you’ll need to find a container based on openSUSE.
If a cuda on openSUSE container image doesn’t exist, I’m sure that other openSUSE Users who follow you would appreciate your effort if you create an image based on openSUSE successfully and contribute to Docker Hub (That’s very easy to do).
HTH,
TSU
I am making some modifications and will publish a howto when it is done. I’ve already built one ELK container for forwarding host logs to it with filebeat.
Hi
Installs here fine, I don’t use the included driver and skip the checks as it doesn’t like gcc8…
./cuda_10.0.130_410.48_linux.run --override
Logging to /tmp/cuda_install_14569.log
Using less to view the EULA.
Do you accept the previously read EULA?
accept/decline/quit: accept
You are attempting to install on an unsupported configuration. Do you wish to continue?
(y)es/(n)o default is no ]: yes
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?
(y)es/(n)o/(q)uit: no
Install the CUDA 10.0 Toolkit?
(y)es/(n)o/(q)uit: yes
Enter Toolkit Location
default is /usr/local/cuda-10.0 ]:
Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: yes
Install the CUDA 10.0 Samples?
(y)es/(n)o/(q)uit: n
Installing the CUDA Toolkit in /usr/local/cuda-10.0 ...
===========
= Summary =
===========
Driver: Not Selected
Toolkit: Installed in /usr/local/cuda-10.0
Samples: Not Selected
Please make sure that
- PATH includes /usr/local/cuda-10.0/bin
- LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin
Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/doc/pdf for detailed information on setting up CUDA.
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 384.00 is required for CUDA 10.0 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
sudo <CudaInstaller>.run -silent -driver
Logfile is /tmp/cuda_install_14569.log
I’m only using it (cuda) for rebuilding lc0 to add nvidia support, I’m currently using 418.43 with Tumbleweed.
Awesome, I will try that now. Thank you!
It worked, thank you.