How can i make it works when pytorch.dataloader num_workers > 0 with gpu_num > 1?

[hopkin-ghp]

Hello, David

GPURIR works on Pytorch online training with one GPU when I set the num_worker of dataloader is 0 through adding following code:

multiprocessing.set_start_method('spawn')
frezze_support

But it don't works when GPU num >1, do you have any good suggestions?

Thanks,
hopkin

[DavidDiazGuerra]

Hello Hopkin,

Being able to use gpuRIR in a GPU different form the one used for training or even using it to perform different simulations in different GPUs would be indeed an interesting thing. However, unfortunately, I have no idea about how to do this and I guess it would need some (minor) changes in the library code.

I have no time to work on this right now, but if someone has some ideas about how to do it that would be an awesome contribution to the library.

Best,
David