Questions tagged [cuda]
The cuda tag has no usage guidance.
57
questions
0
votes
0
answers
21
views
A process using CUDA gets stuck, then all others get stuck as well - what do I do?
I’m writing some program using CUDA CUDA 12.1, running on a Linux system (Devuan Daedalus, kernel version 6.1.27).
For some reason (which may be a bug of mine, although I kind of doubt it) - the ...
- 8,852
1
vote
0
answers
768
views
How can I downgrade CUDA from 12.0 to 11.8 via command line?
I use Ubuntu 20.04.5 LTS with an NVIDIA A100 40GB GPU and CUDA 12.0. How can I downgrade CUDA from 12.0 to 11.8 via command line?
- 4,759
0
votes
0
answers
423
views
How to install torch for cuda in Linux?
I would like to install pytorch for cuda, but getting had package conflicts. How to resolve this?
$ uname -r
3.10.0-1160.71.1.el7.x86_64
$ python -c "import torch; print(torch.version.cuda)"...
- 113
0
votes
0
answers
62
views
Programs using CUDA freeze after one of them fails, can't be killed
I'm trying to debug a certain program of mine which uses CUDA (specifically, CUDA graphs). This program (sometimes) triggers an error, then dies - relatively gracefully. However, after it has failed, ...
- 8,852
0
votes
0
answers
259
views
I am trying to install CUDA in a real time kernel
I installed the real time kernel 5.11.4-rt11 and now I'm trying to install CUDA.
When I type the command:
sudo apt-get -y install cuda-11-8
I get the following error:
ERROR (dkms apport): kernel ...
1
vote
0
answers
147
views
How do I install CUDA from the extracted the local runfile?
I've downloaded a runfile of a recent CUDA version (say, 12.0), made it executable, and ran it with:
./cuda_12.0.1_525.85.12_linux.run --extract=/tmp/cuda-install
and suppose I no longer have the ...
- 8,852
0
votes
0
answers
148
views
Nvidia GPU driver won't compile from RPM
I recently update from OpenSuSE 15.3 to 15.4 including a GPU update to be able to use the new drivers (beyond version 490). Since this upgrade the kernel module is not updated by any RPM base install. ...
- 101
0
votes
0
answers
50
views
GPUDirect RDMA on X86 linux PC, driver build issue
I am trying to do GPUDirect RDMA mentioned in the below link. I am using an X86 Linux PC with Quadro M4000 GPU and CUDA 12.0 tool kit. While building the driver and executing CUDA programming, I am ...
0
votes
0
answers
369
views
Change NVIDIA device files via modprobe.d: NVreg_DeviceFileGID not changed
I want to change NVIDIA device files.
have
# nvidia-smi -L
GPU 0: Tesla V100-PCIE-32GB
GPU 1: Tesla V100-PCIE-32GB
# cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module ...
2
votes
1
answer
563
views
Restricting user access to nvidia GPU?
On a server with Tesla Nvidia Card we decide to Restrict user access to GPU. In our server 2 GPU.
# ls -las /dev/nvidia*
0 crw-rw-rw-. 1 root root 195, 0 Dec 2 22:02 /dev/nvidia0
0 crw-rw-rw-. 1 ...
2
votes
1
answer
415
views
Stuck at booting after upgrading
My GPU is NVIDIA - GeForce RTX 3090 Ti, and the OS is Ubuntu 18.04.
As my code didn’t work, I checked the versions of python, pytorch, cuda, and cudnn.
Python: 3.6
torch. version : 1.4.0
torch....
- 23
2
votes
1
answer
4k
views
How to install matching version of nvidia-cuda-toolkit?
I have installed the NVIDIA drivers for my system (Ubuntu 22) as follows:
Download the installer script: https://us.download.nvidia.com/XFree86/Linux-x86_64/515.65.01/NVIDIA-Linux-x86_64-515.65.01....
user542094
4
votes
1
answer
4k
views
How do I use Cuda toolkit nvcc 11.7.1 on Fedora 36?
As of Sept 2022, Nvidia still has not officially supported cuda toolkit on Fedora 36. The particular part missing is support for gcc12, which Fedora 36 defaults to. One solution to use nvcc on fedora ...
- 333
0
votes
1
answer
267
views
low resolution after installing nvidia-cuda-toolkit from Debian 11 nonfree package (not from Nvidia site)
I installed Debian 11 on my desktop, and on first boot, the Noveau drivers worked fine and had the high resolution, though moving windows was slightly laggy as to be expected. I then installed the ...
- 333
2
votes
0
answers
1k
views
why Cuda not detected by hashcat?
I have the following graphices cards installed in my workstation
Thu Aug 18 19:22:51 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.65.01 ...
- 21
0
votes
1
answer
361
views
Docker Ubuntu container on Debian host with custom CUDA/cuDNN
I have access to Debian 11 server, where some arbitrary NVIDIA driver and CUDA versions are installed. However, since many people use the server, different versions of CUDA and cuDNN are required by ...
- 103
0
votes
0
answers
67
views
How do you set up nvidia mps daemon to launch at boot time on arch Linux?
I've tried searching the web, and I can't find anything that just gives a command on how to get systemctl to enable it. I've tried having systemctl list the installed and I couldn't find anything on ...
0
votes
3
answers
6k
views
Ubuntu 22.04, Dual graphic cards, Intel for display, NVidia for GPGPU, how to setup?
My laptop got 2 graphics cards, something very similar to: https://www.linuxbabe.com/desktop-linux/switch-intel-nvidia-graphics-card-ubuntu .
I would like to have Intel graphic card for display ...
- 101
2
votes
1
answer
525
views
Cuda installation, conflicting nvidia drivers seem to not exist
I am a relatively new Linux user getting an error about conflicting nvidia drivers when installing cuda on Centos 7:
Error: nvidia-driver-latest-dkms-libs conflicts with nvidia-x11-drv-libs-515.48.07-...
- 21
0
votes
0
answers
1k
views
CUDA 11.7 failing to install on Debian sid
I'm trying to install CUDA 11.7 (the latest) on Debian sid using the NVIDIA-recommended method, but am running into the following errors:
$ uname -a
Linux foo 5.17.0-1-amd64 #1 SMP PREEMPT Debian 5.17....
- 153
0
votes
1
answer
737
views
How can update my broken CUDA packages?
How I can fix the following problem? I want to update both these CUDA packages.
Check if you are using third party repositories. If so disable them, since they are a common source of problems.
...
- 113
0
votes
1
answer
507
views
Clearing apt-cache policy version table to prevent the installation of newer versions
I am trying to install a specific package , CUDA for Nvidia to be exact. I followed the steps in their developer guide and my question is not about CUDA specifically. When I try to install it with APT ...
1
vote
1
answer
118
views
Installation of iMod on CentOS 7
I'm trying to create a shell script that installs a series of things for me. One such thing is iMod. I've located self-installing shell script for iMod and have run the following commands on my bash ...
- 113
1
vote
0
answers
179
views
how to run cuda-uninstaller without interaction?
the cuda-unintaller does not have any help message. I try to use -s or --silent but it does not work. I do not want to run the uninstaller on all machines manually and select the components to ...
- 1,214
0
votes
1
answer
1k
views
/usr/local/cuda/version.txt does not match with /usr/lib/cuda/version
I realized that the server I am working on has problem with the CUDA versions. WHen I check the CUDA version with nvidia-smi, nvcc --version and cat /usr/local/cuda/version.txt - they all yield ...
Ezgi Çakır
1
vote
4
answers
7k
views
Installing CUDA Toolkit (11.5) on Debian (11)
The instructions to install CUDA Toolkit (11.5) on Debian (11) are nicely organized here: developer.nvidia.com/cuda-downloads.
But I seem to run in a small dependency issue, whether I try the deb (...
- 172
1
vote
0
answers
113
views
Use the GPU on another process when it is available(Nice-like command for GPU)
I have a program that uses a lot of GPU power. During some epochs, for example, it takes a job and processes it, and then waits about 10 to 15 minutes before taking the next job. When the main program ...
- 111
0
votes
0
answers
124
views
Linux source path issue with nvidia graphic driver installation
I'm trying to install nvidia driver on Ubuntu Linux 20.
The issue - I'm getting the following error during installation.
If you are using a Linux 2.6 kernel, please make sure
you have configured ...
4
votes
0
answers
725
views
CUDA not available on Deep Learning AMI (DLAMI) running on an Amazon EC2 P2 instance
I am running the Ubuntu 18.04 Deep Learning AMI (DLAMI) on AWS, and am attempting to run it on a p2.xlarge EC2 instance, but CUDA is not available in my Python interpreter. I was assuming that CUDA ...
- 2,665
3
votes
0
answers
2k
views
"cuda-runtime-11-3: Depends: cuda-drivers but it is not going to be installed" - want to install anyway
Background
I want to build (not run) a program using CUDA 11.3, on an Ubuntu 20.04 machine (a Travis CI machine actually). To do this, I want to install, among other things, the
cuda-runtime-11-3 ...
- 8,852
2
votes
1
answer
384
views
CentOS 8 can't boot after cuda toolkit installation
[Lenovo LEGION laptop with dual boot - win10 and centos 8]
I was trying to run some YOLOv5 example on my laptop Tutorial#1, using pycharm.
to keep it under control I've generated an yolov5 conda ...
- 21
1
vote
0
answers
1k
views
RuntimeError: CUDA unknown error - this may be due to an incorrectly set up environment, e.g. changing env variable CUDA_VISIBLE_DEVICES
I am trying to check the GPU device name but after executing this code. I got this unknown runtime error. please help me to solve this problem and give complete instructions to solve this error. ...
0
votes
1
answer
77
views
Graphic driver versions available for Ubuntu 16
I have Nvidia GTX 1080Ti and according to Nvidia's website the latest driver version is 460. I tried to install this and failed. With the .run file nothing got installed. With ppa and ubuntu-drivers ...
- 1
1
vote
1
answer
502
views
CUDA on kernel 5.9
I tried to install cuda on Manjaro with kernel linux54 and linux59 but did not manage to get it work. I have a 64-bit laptop with Hybrid graphics, my graphic card being a GeForce 950M. My video-driver ...
- 123
0
votes
0
answers
232
views
NVIDIA driver version/pytorch error message on Ubuntu 18.04
in pycharm IDE I get the following error message:
RuntimeError: The NVIDIA driver on your system is too old (found version 9010). Please update your GPU driver by downloading and installing a new ...
2
votes
1
answer
2k
views
Installing CUDA on Debian Machine 10.3
So, I am trying to install Cuda on Debian 10.3 on SageMaker studio instance ml.g4dn.xlarge.
I am trying the instructions from here:
https://developer.nvidia.com/cuda-downloads?target_os=Linux&...
- 85
0
votes
1
answer
2k
views
failure: repodata/repomd.xml from libnvidia-container: [Errno 256] No more mirrors to try
I followed the instructions from https://developer.nvidia.com/cuda-80-ga2-download-archive
sudo rpm -i cuda-repo-rhel7-8-0-local-ga2-8.0.61-1.x86_64-rpm
sudo yum clean all
sudo yum install cuda
Now I ...
- 109
1
vote
1
answer
2k
views
dpkg-shlibdeps fails with “no dependency information found”
I am building a custom debian package for tensorflow. At some point, when I run
dpkg-buildpackage -us -uc
I get:
dpkg-shlibdeps: error: no dependency information found for /usr/local/cuda-9.1/lib64/...
- 111
2
votes
1
answer
3k
views
Should I grant sudo to a user or add CAP_SYS_ADMIN capabilities to use nvprof/ncu? And why?
Since CUDA 10.1, users are required to have sudo privileges to collect advanced metrics with cuda profiling tools such as nvprof or nsight compute (ncu).
Alternatives to get around it are described ...
- 183
1
vote
1
answer
207
views
compiling HPL-2.0_FERMIv15
when i compile xhpl i always get the error message:
./xhpl: error while loading shared libraries: libdgemm.so.1: cannot open shared object file: No such file or directory
when i type ldd xhpl:
linux-...
- 11
2
votes
0
answers
450
views
How to fix broken CUDA installation [ZorinOS/Ubuntu]?
I have some problems installing CUDA on ZorinOS (I know it's not officially supported, but it acts much like Ubuntu). I tried to install it the way it is described in the official documentation, did ...
- 131
9
votes
3
answers
30k
views
libGL error: No matching fbConfigs or visuals found | Glxgears error, Docker, CUDA, VirtualGL
when I run glxgears, I get following error.
libGL error: No matching fbConfigs or visuals found
libGL error: failed to load driver: swrast
Error: couldn't get an RGB, Double-buffered visual
My ...
- 111
0
votes
0
answers
41
views
"Hermetically" override a distribution-installed CUDA with a locally-installed version
In my Devuan GNU/Linux Beowulf distribution (which is essentially Debian Buster without systemd), CUDA 10.1 is installed. I don't want to uninstall it for various reasons; but - I do want to use a ...
- 8,852
4
votes
1
answer
4k
views
CUDA 10 and dkms on RHEL8
I want to install CUDA according to the info on NVIDIA CUDA toolkit site
wget http://developer.download.nvidia.com/compute/cuda/10.2/Prod/local_installers/cuda-repo-rhel8-10-2-local-10.2.89-440.33.01-...
- 203
1
vote
1
answer
2k
views
Unable to find (and uninstall) installed version of CUDA
I am trying to install CUDA version 10.0, but it is telling me a newer version is already installed. I am not able to find the newer version, so I can't run the uninstaller. I am on CentOS 7. Here's ...
- 195
0
votes
1
answer
682
views
Transaction check error during CUDA installation on CentOS 7
I am trying to install CUDA on a Linux CentOS 7 x86_64 AWS instance via the installation guide and running into an error that I cannot resolve.
Here are the steps I took:
I verified that I had gcc ...
- 195
0
votes
1
answer
642
views
Installing CUDA on my Ubuntu 18.04 system
I want to install CUDA on my system to run OpenCV.
I have the NVIDIA 1050 Ti Mobile which supports the driver 390.116. I tried to install CUDA 10 and failed, I think because of the driver ...
- 1
0
votes
1
answer
3k
views
How to install CUDA 9.2 on Linux Mint 19
I would like to install CUDA 9.2. on Linux Mint 19.
It was possible to install CUDA on Linux Mint 18.3 even though Linux Mint is not an officially supported distribution. However,
Linux Mint 18 was ...
- 101
5
votes
1
answer
6k
views
How to install CUDA 9.1 on Mint 18.3?
I'm running Linux Mint 18.3 with the Cinnamon desktop environment, and I want to install CUDA 9.1 and NVIDIA drivers. How can I do that?
0
votes
1
answer
3k
views
Install CUDA 8 on linux mint 18.2
I'm running Linux Mint 18.2 "Sonya" and want to install CUDA 8. When I install it using package manager it install Cuda 7.5 instead:
sudo apt-get install nvidia-cuda-dev nvidia-cuda-toolkit
...
...
- 173