Investigate running Pytorch on GPU

Question

Investigate running Pytorch on GPU

Opened this issue 5 years ago · 5 comments

Hi @TinyBugBigProblem I'm opening this on your behalf. I understand you are working on GPU acceleration for Pytorch. Depending on your environment the setup and installation of the stack can be non-trivial.

How are you currently working on this? Can you detail? I've tried to get the basic installation running on macOS 10.14.6 (18G1012) and was unable. It turns out that MacOS binaries do not support CUDA. I was required to install Pytorch from source... which simply didn't work as expected.

Answer 1 · 2019-11-21T06:04:58.000Z

It turns out that the version of macOS I am running is imply too new for the CUDA toolchain. I have a GPU server in work which I'll work on. I'll update this ticket tomorrow with that environment.

Answer 2 · 2019-11-22T06:54:28.000Z

@TinyBugBigProblem PING

Answer 3 · 2019-11-26T05:07:15.000Z

Hi @TinyBugBigProblem please post whatever code you have so I can take a look. Thank you

Answer 4 · 2019-11-27T17:37:55.000Z

@TinyBugBigProblem PING

Answer 5 · 2019-11-30T22:21:08.000Z

@TinyBugBigProblem please post your update by end of business Monday morning as this signifies the end of the project. Thank you