CUDA pytorch not working on a fresh install RTX 3070 - Technical Help - DeepTalk - Deep Learning Community

CUDA pytorch not working on a fresh install RTX 3070

post by novice1 on Mar 9, 2021

Hello,

I did a fresh install of ubuntu 20.04 and immediately installed lambda stack following the instructions as I want to use pytorch on my GPU. The installation went without errors nvndia-smi shows 460.39 and CUDA 11.2. Kernel version is 5.8.0-44-generic.

If i try to run torch in a lambda stack virtual environment however i get errors.

torch.cuda.is_available() throws

/usr/lib/python3/dist-packages/torch/cuda/**init**.py:52: UserWarning: CUDA initialization: CUDA unknown error - this may be due to an incorrectly set up environment, e.g. changing env variable CUDA_VISIBLE_DEVICES after program start. Setting the available devices to be zero. (Triggered internally at …/c10/cuda/CUDAFunctions.cpp:100.)

return torch._C._cuda_getDeviceCount() > 0

If i do print(torch.version.cuda) it shows 11.1 .

I’ve tried everything relating to similar errors i’ve found on the internet including rebooting my pc and nothing seems to fix it. Especially confusing is that it is a fresh install which should work without any issues. Any help would be very appreciated.

1.8k views

post by novice1 on Mar 10, 2021

I reinstalled ubuntu again and did the exact same thing and it appears to be working. No clue what the issue was.