5

I have been having trouble with my GPUs lately. I just did a clean wipe of my machine and installed Ubuntu 16.04 and CUDA 8.0 on it, hoping that would fix the problem. However, I still get errors. Trying to run things like deviceQuery just gives back errors. It seems one of my GPUs is failing; my guess is at the hardware level. Here is the output from nvidia-smi:

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 367.57                 Driver Version: 367.57                    |
|-------------------------------+----------------------+----------------------+
| 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 GTX TIT...  Off  | 0000:05:00.0     Off |                  N/A |
| 32%   46C    P8    28W / 189W |      1MiB /  6079MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX TIT...  Off  | 0000:06:00.0     Off |                  N/A |
|ERR!   58C    P0   ERR! / 189W |      0MiB /  6082MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce GTX TIT...  Off  | 0000:09:00.0      On |                  N/A |
| 31%   50C    P0    73W / 189W |    322MiB /  6079MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  GeForce GTX TIT...  Off  | 0000:0A:00.0     Off |                  N/A |
| 28%   38C    P8    32W / 189W |      0MiB /  6082MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID  Type  Process name                               Usage      |
|=============================================================================|
|    2      4712    G   /usr/lib/xorg/Xorg                             255MiB |
|    2      5314    G   compiz                                          66MiB |
+-----------------------------------------------------------------------------+

I want to disable GPU 1 so that I can just use the other GPUs to run things. The GPUs are Titan Zs, so trying to just unplug a GPU would cut my computational power in half instead of just by 1/4th. So if that is the only option I will do it but I would prefer something else. For some background, I am doing Neural Net research and thus I need to use GPUs a lot.

0

You must log in to answer this question.

Browse other questions tagged .