colab-tricks icon indicating copy to clipboard operation
colab-tricks copied to clipboard

nvidia-smi problem after SSH into Colab

Open vochicong opened this issue 4 years ago • 1 comments

Thank you for great tips about SSH into Colab at ngrok-tricks.ipynb.

I encountered one problem. For the command nvidia-smi, in the Colab notebook:

Fri Mar 20 00:58:40 2020       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.64.00    Driver Version: 418.67       CUDA Version: 10.1     |
|-------------------------------+----------------------+----------------------+
| 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  | 00000000:00:04.0 Off |                    0 |
| N/A   40C    P0    28W / 250W |      0MiB / 16280MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+
Your runtime has 27.4 gigabytes of available RAM

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sat_Aug_25_21:08:01_CDT_2018
Cuda compilation tools, release 10.0, V10.0.130

But on my Mac, after SSH into Colab

$ ssh -p 10055 [email protected] -i ~/.ssh/id_rsa_colab
Welcome to Ubuntu 18.04.3 LTS (GNU/Linux 4.14.137+ x86_64)
.....
root@118ff34f0bd4:~# nvidia-smi; /usr/local/cuda/bin/nvcc --version
Failed to initialize NVML: Driver/library version mismatch
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sat_Aug_25_21:08:01_CDT_2018
Cuda compilation tools, release 10.0, V10.0.130
root@118ff34f0bd4:~#

vochicong avatar Mar 20 '20 01:03 vochicong

Just found that even on the Colab

!which nvidia-smi
!sudo nvidia-smi
/usr/bin/nvidia-smi
Failed to initialize NVML: Driver/library version mismatch

So the problem came from using root for SSH login.

vochicong avatar Mar 20 '20 01:03 vochicong