I had an issue with my computer and had to completely reinstall windows on my machine. I previously had CUDA working with WSL and could run PyTorch models using my GPU. However it seems that WSL can't find the GPU anymore, and I am not sure why. Running nvidia-smi
in powershell shows the GPU, but I can't even get nvidia-smi to run properly in WSL.
I am running Windows 10 Pro v 22H2 Build 19045.4780, with Nvidia 2070 Super, with GeForce Game Ready Driver 560.81, and AMD Ryzen 7 3700X 8-Core Processor
Below are outputs from relevant commands in the WSL shell.
$ nvcc --versionnvcc: NVIDIA (R) Cuda compiler driverCopyright (c) 2005-2024 NVIDIA CorporationBuilt on Fri_Jun_14_16:34:21_PDT_2024Cuda compilation tools, release 12.6, V12.6.20Build cuda_12.6.r12.6/compiler.34431801_0$ lspcipcilib: Cannot open /proc/bus/pcilspci: Cannot find any working access method.$ nvidia-smiCommand 'nvidia-smi' not found, but can be installed with:sudo apt install nvidia-utils-390 # version 390.157-0ubuntu0.22.04.2, orsudo apt install nvidia-utils-418-server # version 418.226.00-0ubuntu5~0.22.04.1sudo apt install nvidia-utils-450-server # version 450.248.02-0ubuntu0.22.04.1sudo apt install nvidia-utils-470 # version 470.256.02-0ubuntu0.22.04.1sudo apt install nvidia-utils-470-server # version 470.256.02-0ubuntu0.22.04.1....$ ./cuda-samples/Samples/1_Utilities/deviceQuery/deviceQuery./deviceQuery Starting... CUDA Device Query (Runtime API) version (CUDART static linking)cudaGetDeviceCount returned 100-> no CUDA-capable device is detectedResult = FAIL
I have in Windows Features settings: Hyper-V, Virtual Machine Platform, Windows Hypervisor Platform, Windows Powershell 2.0, Windows Subsystem for Linux all turned on.
My only thought is that I need to downgrade the driver version, but I have tried a few previous versions of the driver, as well as using CUDA 11.5 and 11.6.