If you want to use this please build the images yourself.
podman build -t repo.despera.space/iru/rocm:5.4.3 -f Dockerfile.5.4.3 .
podman run -it --privileged -v /dev/dri:/dev/dri -v /dev/kfd:/dev/kfd repo.despera.space/iru/rocm:5.4.3 /bin/bash
podman build -t repo.despera.space/iru/rocm:5.4.3-torch5.4.2 -f torch/Dockerfile.5.4.3 .
podman run -it --privileged -v /dev/dri:/dev/dri -v /dev/kfd:/dev/kfd repo.despera.space/iru/rocm:5.4.3-torch5.4.2 /bin/bash
podman build -t repo.despera.space/iru/rocm:5.2.3 -f Dockerfile.5.2.3 .
podman run -it --privileged -v /dev/dri:/dev/dri -v /dev/kfd:/dev/kfd repo.despera.space/iru/rocm:5.2.3 /bin/bash
podman build -t repo.despera.space/iru/rocm:5.2.3-torch-1.13.1 -f torch/Dockerfile.5.2.3 .
podman run -it --privileged -v /dev/dri:/dev/dri -v /dev/kfd:/dev/kfd repo.despera.space/iru/rocm:5.2.3-torch-1.13.1 /bin/bash
Run rocminfo
or rocm-smi
.
Run Python3.
import torch
torch.cuda.is_available()
torch.cuda.device_count()
torch.cuda.get_device_name(0)
The libraries installed are quite big. I have no idea why