[FEATURE] AMD GPU support

[Notnaton]

Discoveries making docker work with AMD GPU

Prerequisites:
You need amdgpu-dkms on host. (https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/quick-start.html#rocm-install-quick)

Add this to docker-compose.yaml(https://rocm.docs.amd.com/projects/install-on-linux/en/latest/how-to/docker.html#docker-compose)

    devices:
      - /dev/kfd:/dev/kfd
      - /dev/dri:/dev/dri
    environment:
      - PATH=/opt/rocm/bin:/opt/rocm/hip/bin:/opt/rocm/opencl/bin:/opt/rocm/hsa/bin:/opt/rocm/llvm/bin:$PATH
      - HIP_PATH=/opt/rocm/
      - HIP_CLANG_PATH=/opt/rocm/llvm/bin
      - HIP_DEVICE_LIB_PATH=/opt/rocm/amdgcn/bitcode

run:
sudo docker compose up -d
and
sudo docker exec -it pixelarch_os-pixelarch-os-1 /bin/bash

Installing amd stuff:
yay -Syu rocm-hip-sdk rocminfo

Note:
Running things on GPU requires sudo (ROCm/ROCm-docker#90 (comment))
Optional compile llama.cpp:

git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp

HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \
    cmake -S . -B build -DGGML_HIP=ON -DAMDGPU_TARGETS=gfx1100 -DCMAKE_BUILD_TYPE=Release \
    && cmake --build build --config Release -- -j 16

[Notnaton]

┌─[midori-ai][b15ade670966][~]
└─➞ ls -l /dev/kfd /dev/dri
crw-rw---- 1 root 44 235, 0 Feb 24 20:44 /dev/kfd

/dev/dri:
total 0
crw-rw---- 1 root 44 226, 1 Feb 24 20:44 card1
crw-rw---- 1 root 110 226, 128 Feb 24 20:44 renderD128

[lunamidori5]

@Notnaton cleaned up your post for you