如何删除 cuInit 失败：CUDA 中的未知错误 (PyCuda)

Question

I have CUDA installed then I tried the following:

In [1]: import pycuda.driver as cuda

In [2]: cuda.init()
---------------------------------------------------------------------------
Error                                     Traceback (most recent call last)
<ipython-input-2-2845c9c0ab3c> in <module>()
----> 1 cuda.init()

Error: cuInit failed: unknown error

How can resolve the error?

I already have this installed:

$ which nvidia-modprobe
/usr/bin/nvidia-modprobe

My other info:

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2015 NVIDIA Corporation
Built on Tue_Aug_11_14:27:32_CDT_2015
Cuda compilation tools, release 7.5, V7.5.17

And this:

$ uname -a
Linux foobar1 4.10.0-28-generic #32~16.04.2-Ubuntu SMP Thu Jul 20 10:19:48 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

And this is the information for the nvidia-smi

$ nvidia-smi
Mon Nov 19 15:56:43 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.48                 Driver Version: 410.48                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Quadro P400         Off  | 00000000:AF:00.0 Off |                  N/A |
| 34%   45C    P0    N/A /  N/A |      0MiB /  1999MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Answer 1

cuInit failed: unknown error is often the result of nvidia-uvm kernel module not being loaded. I've been periodically running into this issue on Ubuntu.

sudo nvidia-modprobe -u should fix the problem. That is, until you reboot. Then you will need to do it again.

Another workaround is to run your failing application as root once . AFAICT it in this case CUDA runtime will attempt to load the missing module (and will most likely succeed at that, because we're running as root).