NVlabs / nvdiffmodeling

Differentiable rasterization applied to 3D model simplification tasks
Other
455 stars 30 forks source link

Is it possible to run the code on Linux? #30

Closed X1aoyueyue closed 1 year ago

X1aoyueyue commented 1 year ago

Hello, I want to run the code on Supercomputer and do some parallel rendering. I woule like to know if it is ok to run on a Linux os. And will it possible work for my idea? I want to let one GPU to optimize the base mesh and let other GPUs to render the ref mesh. I think this way will help to load big mesh data.

Thanks.

jmunkberg commented 1 year ago

Yes, linux should work out of the box. We included docker instructions in the readme: https://github.com/NVlabs/nvdiffmodeling#server-usage-through-docker

For multiGPU, we did not add support in this project, but a later, very similar code base does have mGPU support: https://github.com/NVlabs/nvdiffrec#examples so hopefully you can adapt a similar strategy (we used PyTorch DDP

Also see my reply at https://github.com/NVlabs/nvdiffmodeling/issues/16

X1aoyueyue commented 1 year ago

Thanks a lot. I will try.