Installing/Compiling pointops on recent pytorch and cuda versions

[bjoukovs]

Here are some indicators on how to install pointops using more recent versions of pytorch and cuda. I was initially unable to install the binaries using the recommended configuration of the repository. I couldn't find the exact solution in the existing issues.

First, here is my configuration for reference:
Ubuntu 22.04.3
GPU RTX 2080 Ti
Pytorch 2.1.0
nvcc 11.5
cudatoolkit 11.8

The key to solve the installation issues was to

• remove all "#include <THC/THC.h>" lines in the *.cpp files from the source code
• when installing pointops, make sure to use gcc-10 and g++-10 instead of 11. You can use the following command:
  export CMAKE_CUDA_HOST_COMPILER=/usr/bin/g++-10
CC=gcc-10 CXX=g++-10 python setup.py install

I hope this helps other people struggling to install pointops on more recent configurations (I think this should also work for earlier versions of pytorch)