Troubleshoot why installed nvidia driver and cuda already, why PyTorch still cant use cuda

pytorch.md

Troubleshoot why installed nvidia driver and cuda already, why PyTorch still cant use cuda

My environment: Ubuntu 22.04 LTS on Win11 WSL 2, RTX 2060 on Laptop

I have checked nvidia and cuda already is latest by running below command to see installed version, and comparing the result of googling latest nvdia driver and cuda version online

nvidia-smi

Nvidia driver link

nvcc release notes

And i found out what was the problem

when i tried to pip upgrade all the dependencies, it installed cpu version of pytorch...

you can verify it by

python -m torch.utils.collect_env

I got the result, notice all the cpu in the package description

[pip3] numpy==1.24.3
[pip3] pytorch-lightning==1.7.7
[pip3] torch==1.12.1
[pip3] torchaudio==0.12.1
[pip3] torchdiffeq==0.2.3
[pip3] torchmetrics==0.11.4
[pip3] torchsde==0.2.5
[pip3] torchvision==0.15.1a0+60a3e72
[conda] cudatoolkit               10.2.89             h713d32c_11    conda-forge
[conda] numpy                     1.24.3          py310ha4c1d20_0    conda-forge
[conda] pytorch                   1.12.1          cpu_py310h9dbd814_1  
[conda] pytorch-lightning         1.7.7                    pypi_0    pypi
[conda] pytorch-mutex             1.0                        cuda    pytorch
[conda] torchaudio                0.12.1              py310_cu102    pytorch
[conda] torchdiffeq               0.2.3                    pypi_0    pypi
[conda] torchmetrics              0.11.4                   pypi_0    pypi
[conda] torchsde                  0.2.5                    pypi_0    pypi
[conda] torchvision               0.15.1          cpu_py310h0397dac_0    conda-forge

Then i remove pytorch packages

pip uninstall torch torchvision torchaudio

and install the cuda version as suggested by official doc

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

and then run the package check command again, we can notice now the package are cuda version

[pip3] numpy==1.24.3
[pip3] pytorch-lightning==1.7.7
[pip3] torch==2.0.1+cu118
[pip3] torchaudio==2.0.2+cu118
[pip3] torchdiffeq==0.2.3
[pip3] torchmetrics==0.11.4
[pip3] torchsde==0.2.5
[pip3] torchvision==0.15.2+cu118
[conda] cudatoolkit               10.2.89             h713d32c_11    conda-forge
[conda] numpy                     1.24.3          py310ha4c1d20_0    conda-forge
[conda] pytorch-lightning         1.7.7                    pypi_0    pypi
[conda] pytorch-mutex             1.0                        cuda    pytorch
[conda] torch                     2.0.1+cu118              pypi_0    pypi
[conda] torchaudio                2.0.2+cu118              pypi_0    pypi
[conda] torchdiffeq               0.2.3                    pypi_0    pypi
[conda] torchmetrics              0.11.4                   pypi_0    pypi
[conda] torchsde                  0.2.5                    pypi_0    pypi
[conda] torchvision               0.15.2+cu118             pypi_0    pypi