Question about Train environment

[YUHANG-Ma]

Hi wenbo, Thanks for your code sharing. I am trying to use your code to train my own model. Could I ask what typr of GPU and CPU are you based on when you are training your model? Like how many kernals and memory of cpu and GPUtpe(1080,2060 or 3090?) Also, the type of my GPU is 3060 and it is not suit for your image env(cuda10). I couldn't run model.cuda( ) on my computer. So I changed the cudatoolkit to 11.0. In this way model.cuda() is sucessful but when I run build.sh, pg_op can't be built successfully. I gusess it is because the dismatch of cuda. Do you have any sugesstion on how to import pg_op on cuda11.0? Looking forward to your reply : )

[wenbowen123]

Hi, we were using 2080Ti and the CPU memory was 32G. Did you replace the cuda version inside docker? If so, can you put the screenshot of the build error here? That'd help to check the problem.

[YUHANG-Ma]

I just change the cuda env in python and not in docker. Here is the error when building.

Here is the cuda version in python. Cuda in docker is still 10.1. Here is the error when building.

[wenbowen123]

Outside of docker, there will be many overhead (more than just cuda related things) and thus it's hard to support. One possible solution is to replace the cuda version inside the docker, then you can still have all the other dependencies for the environments.

[waiyc]

@YUHANG-Ma I am having the same GPU 3060 as you and would like to upgrade the CUDA version in the docker to test this repo. Do you mind to tell me how you change the CUDA and cuDNN version in the docker container?

[wenbowen123]

@waiyc @YUHANG-Ma we just updated our docker for newer GPU, e.g. 3090 wenbowen123/catgrasp:cuda11