I am on Ubuntu 20, CUDA 11.1 and Pytorch 10.2.
I am trying to compile a Pytorch extension (GitHub - daerduoCarey/PyTorchEMD: PyTorch Wrapper for Earth-Mover-Distance (EMD) for 3D point cloud regression) .
And I got this error with python 3.6, when running the python setup.py install :
The detected CUDA version (11.1) mismatches the version that was used to compile
PyTorch (10.2). Please make sure to use the same CUDA versions.
And when I use another old conda environment it compiles but gives me this error.
cuda/emd_kernel.cu(391): error: no suitable constructor exists to convert from “c10::ScalarType” to “at::Type”
error: command ‘/usr/bin/nvcc’ failed with exit status 1
I tried to upgrade pytorch, but same issue.
I can I install CUDA 11.3 with lambda stack? will I have any issues with code written in Pytorch 10.2? Or should I keep the CUDA 11.1 but use python 3.8?