2

On a newly built Ubuntu 16.04 machine, running nvidia-smi fails as a regular user

$ nvidia-smi
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

Running as root works

$ sudo nvidia-smi
[sudo] password for hanxue: 
Fri Jul 19 10:05:49 2019       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.130                Driver Version: 384.130                   |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Tesla P100-PCIE...  Off  | 00000000:3B:00.0 Off |                    0 |
| N/A   38C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  Tesla P100-PCIE...  Off  | 00000000:5E:00.0 Off |                    0 |
| N/A   33C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  Tesla P100-PCIE...  Off  | 00000000:86:00.0 Off |                    0 |
| N/A   31C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  Tesla P100-PCIE...  Off  | 00000000:AF:00.0 Off |                    0 |
| N/A   31C    P0    28W / 250W |      0MiB / 16276MiB |      4%      Default |
+-------------------------------+----------------------+----------------------+

And subsequently running as a regular user works

$ nvidia-smi
Fri Jul 19 10:09:00 2019       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.130                Driver Version: 384.130                   |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Tesla P100-PCIE...  Off  | 00000000:3B:00.0 Off |                    0 |
| N/A   40C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  Tesla P100-PCIE...  Off  | 00000000:5E:00.0 Off |                    0 |
| N/A   35C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  Tesla P100-PCIE...  Off  | 00000000:86:00.0 Off |                    0 |
| N/A   33C    P0    31W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  Tesla P100-PCIE...  Off  | 00000000:AF:00.0 Off |                    0 |
| N/A   33C    P0    27W / 250W |      0MiB / 16276MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

Is there a mis-configuration that needs nvidia-smi to be run by root users first, and is there a solution for it? e.g. manually load the NVIDIA kernel modules

3
  • Hm. If you were able to run compute jobs then the correct modules should have been loaded already. But you don't have CUDA installed yet? The CUDA version should be showing. Jul 19, 2019 at 4:27
  • CUDA is already installed. I wonder if it is a bug - on a newly built machine, nvidia-smi must be run as root first
    – hanxue
    Jul 21, 2019 at 7:13
  • I have the same issue on my laptop: It does not detect my eGpu unless I run it in privileged mode (then, once detected, my regular user can detect it with $ nvidia-smi). yesterday

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.