Problem with installing Nvidia HPC-SDK on Ubuntu: No accelerators found. Check the permissions on your CUDA device

Dear all,

I am not sure if I landed in the right forum category, please redirect me somewhere else if needed. I would like to use Nvidia Fortran compiler with OpenACC for development and profiling of my programs. I have downloaded Nvidia HPC-SDK and installed it on Ubuntu 18.04 workstation.

Command:

nvidia-smi

gives me this output:

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.66       Driver Version: 410.66       CUDA Version: 10.0     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce RTX 2070    On   | 00000000:02:00.0  On |                  N/A |
| 34%   48C    P0    59W / 185W |   1019MiB /  7951MiB |      4%      Default |
+-------------------------------+----------------------+----------------------+
                                                                             
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      4077      G   /usr/lib/paraview/paraview                    35MiB |
|    0      6967      G   /usr/lib/xorg/Xorg                           415MiB |
|    0      7055      G   /usr/bin/gnome-shell                         353MiB |
+-----------------------------------------------------------------------------+

However, if I run the command:

nvaccelinfo -v

I get the following output:

CUDA Driver Version: 10000 
NVRM version: NVIDIA UNIX x86_64 Kernel Module 410.66 Wed Oct 10 12:01:53 CDT 2018 
could not initialize CUDA runtime, error code=999 
No accelerators found. 
Check the permissions on your CUDA device

I am doing all of this as a root, and I do have the /dev/nvidia* files with proper permissions:

crw-rw-rw- 1 root root 195,   0 Jul 13 16:18 /dev/nvidia0
crw-rw-rw- 1 root root 195, 255 Jul 13 16:18 /dev/nvidiactl
crw-rw-rw- 1 root root 195, 254 Jul 13 16:18 /dev/nvidia-modeset
crw-rw-rw- 1 root root 511,   0 Jul 13 16:57 /dev/nvidia-uvm
crw-rw-rw- 1 root root 511,   1 Jul 13 16:57 /dev/nvidia-uvm-tools

I do not, however, have the file:

/usr/lib64/libcuda.so

which, although worrisome, is probably a separate issue.

The card I have on my system Ge-Force RTX 2070. According to this page: https://developer.nvidia.com/cuda-gpus#compute it should be possible to use it for CUDA computing.

I am pretty new to the whole thing, and am baffled. I spent some time googling for potential solution but it was to no avail. Any advice would be appreciated.

Thanks

1 Like