Reputation: 19
I am trying to run a downloaded deep learning algorithm, but get the following error message:
THCudaCheck FAIL file=/pytorch/torch/csrc/cuda/Module.cpp line=34 error=35 : CUDA driver version is insufficient for CUDA runtime version
Traceback (most recent call last):
File "train_nli.py", line 61, in <module>
torch.cuda.set_device(params.gpu_id)
File "home/jakob/.local/lib/python3.6/site-packages/torch/cuda/__init__.py", line 264, in set_device
torch._c._cuda_setDevice(device)
RuntimeError: cuda runtime error (35) : CUDA driver version is insufficient for CUDA runtime version at /pytorch/torch/csrc/cuda/Module.cpp:34
I already updated the GPU driver but it didn't do anything. If I execute the same in the windows console I get a different error:
Traceback (most recent call last):
File "train_nli.py", line 61, in <module>
torch.cuda.set_device(params.gpu_id)
File "C:\Users\Jaki\Anaconda3\lib\site-packages\torch\cuda\__init__.py", line 264, in set_device
torch._C._cuda_setDevice(device)
AttributeError: module 'torch._C' has no attribute '_cuda_setDevice'
Upvotes: 1
Views: 1375
Reputation: 15943
The Windows subsystem for Linux (on top of which the Windows 10 Ubuntu Shell is built) is basically a separate kernel interface that allows Linux userspace programs to run natively on top of the Windows kernel. Think of it as an adapter that maps Linux syscalls to Windows syscalls. While it's possible to run many Linux applications on top of Windows like this, you are still running on top of the Windows kernel. There is no Linux kernel underneath. You can't simply install stuff like Linux kernel modules for device drivers in there. It's just not possible at this time to access the GPU in there…
As for that error you get when running in the Windows console, that would seem like some issue with your installation of this package (does it even support Windows?)…
Upvotes: 2