Pytorch on an AMD gpu (frame.work 13)

I have a frame.work laptop. it is really nice! it looks awesome and is easily repairable. I chose an AMD type, which as an integated GPU. the AMD Ryzen 7 7840U

You can actually use this GPU with pytorch!

But you need to perform a few steps, I write them down here for future use. (I’m using ubuntu on this device)

• allocate more VRAM to GPU with a bios setting (go into bios and change setting GPU to gaming mode or something, see this link)
• start a virtual environment in your project
• install the right versions of pytorch packages; go to https://pytorch.org/get-started/locally/ and click the right versions together to make the correct link, I came out to pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7
• set an env var to point pytorch in the right direction. in my case export HSA_OVERRIDE_GFX_VERSION=11.0.0

I don’t think it is particularily fast, but using the gpu is often faster than cpu.

Will it run large deep learning models like LLMs?

According to this benchmark https://llm-tracker.info/howto/AMD-GPUs#bkmrk-amd-apu LLMs could be slightly faster than on cpu

What about torch, for R?

It does not yet work for torch on R, see this issue https://github.com/mlverse/torch/issues/907