hello-rocm

English

简体中文

English

简体中文

Appearance

Ubuntu 24.04 / Windows 11 Environment Setup: ROCm 7.13 + PyTorch + vLLM (gfx1151 Example) ​

ROCm 7.13.0-preview environment setup guide for Gemma 4 deployment.

This guide uses Ryzen AI Max / Ryzen AI Max+ (gfx1151) as the reference GPU architecture and summarizes the key setup steps under ROCm 7.13 / TheRock.

Official references:

1. ROCm 7.13 / TheRock Notes ​

ROCm 7.13 moves into the TheRock / Core SDK packaging model:

ItemLegacy ROCmROCm 7.13
Core path/opt/rocm//opt/rocm/core
Package prefixrocm-*, hip*, roc*amdrocm-*
CompatibilityLegacy ROCm layoutCore SDK with ABI / API compatibility and common symlinks

When installed through package managers, common /opt/rocm symlinks are preserved. With tarball or custom installs, check PATH, LD_LIBRARY_PATH, and ROCM_PATH manually.

2. Clean Existing ROCm / AMD Components ​

If an older ROCm stack, HIP SDK, or amdgpu-dkms has been installed, clean it first to avoid conflicts with ROCm 7.13 / TheRock components:

bash
sudo apt remove 'rocm*' 'amdrocm*' 'amdgpu-dkms*' -y
sudo apt autoremove -y

Also check ~/.bashrc, ~/.zshrc, and /etc/profile.d/ for stale ROCm environment variables.

3. Ubuntu 24.04 + gfx1151 Setup ​

2.1 Install OEM kernel 6.14 ​

bash
sudo apt update
sudo apt install -y linux-image-6.14.0-1018-oem
sudo reboot

2.2 Install basic dependencies ​

bash
sudo apt update
sudo apt install -y \
  python3.13 python3.13-venv \
  libatomic1 libquadmath0 \
  build-essential git curl wget jq pciutils

2.3 Configure GPU permissions ​

bash
sudo usermod -a -G render,video $LOGNAME
sudo reboot

Or use udev rules:

bash
sudo tee /etc/udev/rules.d/70-amdgpu.rules <<'EOF'
KERNEL=="kfd", GROUP="render", MODE="0666"
SUBSYSTEM=="drm", KERNEL=="renderD*", GROUP="render", MODE="0666"
EOF

sudo udevadm control --reload-rules
sudo udevadm trigger
sudo reboot

4. Install PyTorch 2.11.0 (ROCm 7.13 / gfx1151) ​

The project recommends using uv to manage Python environments and dependencies instead of the traditional pip + venv flow.

bash
# Install uv if needed
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install Python 3.13
uv python install 3.13

# Create and activate the virtual environment
uv venv --python 3.13
source .venv/bin/activate

# Fallback: Python standard-library venv
# python3.13 -m venv .venv
# source .venv/bin/activate
# python -m pip install --upgrade pip
bash
uv pip install --index-url https://repo.amd.com/rocm/whl/gfx1151/ \
  "torch==2.11.0+rocm7.13.0" \
  "torchvision==0.26.0+rocm7.13.0" \
  "torchaudio==2.11.0+rocm7.13.0"

Verify:

bash
python - <<'PY'
import torch
print("torch:", torch.__version__)
print("HIP available:", torch.cuda.is_available())
if torch.cuda.is_available():
    print("device:", torch.cuda.get_device_name(0))
PY

5. Windows 11 + pip Path (ROCm 7.13) ​

On Windows 11, ROCm 7.13 uses the pip / TheRock flow. Before installation, uninstall any existing HIP SDK, disable WDAG / SAC, and install AMD Software: Adrenalin Edition 26.5.1 or newer.

powershell
irm https://astral.sh/uv/install.ps1 | iex
uv python install 3.13
uv venv --python 3.13
.venv\Scripts\activate

# Fallback:
# py -3.13 -m venv .venv
# .venv\Scripts\activate

uv pip install --index-url https://repo.amd.com/rocm/whl/gfx1151/ `
  "torch==2.11.0+rocm7.13.0" `
  "torchvision==0.26.0+rocm7.13.0" `
  "torchaudio==2.11.0+rocm7.13.0"

python -c "import torch; print(torch.cuda.is_available())"

6. vLLM Environment Check (Docker) ​

The official ROCm 7.13 vLLM image for gfx1151 is:

bash
docker pull rocm/vllm:rocm7.13.0_gfx1151_ubuntu24.04_py3.13_pytorch_2.10.0_vllm_0.19.1

Note: the vLLM 0.19.1 Docker image includes PyTorch 2.10.0. PyTorch 2.11.0 belongs to the pip installation path above.

bash
docker run -it --rm \
  --device /dev/kfd \
  --device /dev/dri \
  --network=host \
  --ipc=host \
  --group-add=video \
  --cap-add=SYS_PTRACE \
  --security-opt seccomp=unconfined \
  -v ~/models:/app/models \
  -e HF_HOME="/app/models" \
  rocm/vllm:rocm7.13.0_gfx1151_ubuntu24.04_py3.13_pytorch_2.10.0_vllm_0.19.1 \
  bash

Inside the container:

bash
python -c "import vllm; print('vLLM:', vllm.__version__)"
python -c "import torch; print('PyTorch:', torch.__version__, 'HIP:', torch.cuda.is_available())"

7. Next Steps ​