Windows AMD Setup

Run GPUFlow on Windows with AMD GPUs using Windows Subsystem for Linux 2 (WSL2) and ROCm drivers. This setup is experimental but allows Windows users with AMD hardware to participate in the GPUFlow network.

Caution

AMD GPU support on Windows through WSL2 is experimental. NVIDIA GPUs have more mature support. Expect occasional issues and limited software compatibility.

Prerequisites

Verify your system meets these requirements:

• Windows 10 version 2004+ or Windows 11 (any version)
• AMD RX 6000/7000 series GPU with 12GB+ VRAM
• 32GB system RAM (minimum due to WSL2 overhead with AMD)
• Administrator access to install software
• Stable internet connection (25+ Mbps upload)

Step 1: Check hardware compatibility

Verify your AMD GPU is supported:

Check your GPU model:

1. Right-click desktop, select “AMD Software”
2. Go to “System” tab
3. Note your GPU model under “Graphics Hardware”

Supported AMD GPUs:

• RX 6600, 6700 XT, 6800 XT, 6900 XT
• RX 7600, 7700 XT, 7800 XT, 7900 XT, 7900 XTX
• Radeon Pro W6000+ series

Step 2: Install Windows AMD drivers

Install the latest AMD Adrenalin drivers:

1. Visit AMD’s driver download page
2. Select your GPU model
3. Download and install AMD Software: Adrenalin Edition
4. Restart your computer

Step 3: Enable WSL2

Install Windows Subsystem for Linux 2:

# Run PowerShell as Administrator
wsl --install

# Restart computer when prompted

After restart, verify installation:

wsl --version

Step 4: Install Ubuntu 22.04 on WSL2

# Install Ubuntu
wsl --install -d Ubuntu-22.04

# Set as default distribution
wsl --set-default Ubuntu-22.04

Configure Ubuntu:

1. Launch Ubuntu from Start menu
2. Create username and password
3. Complete initial setup

Step 5: Install ROCm in WSL2

ROCm provides AMD GPU compute support in Linux containers.

Launch WSL2 and install ROCm:

# Update Ubuntu
sudo apt update && sudo apt upgrade -y

# Install prerequisites
sudo apt install wget gnupg2

# Add ROCm repository
wget -q -O - https://repo.radeon.com/rocm/rocm.gpg.key | sudo apt-key add -
echo 'deb [arch=amd64] https://repo.radeon.com/rocm/apt/5.7/ ubuntu main' | sudo tee /etc/apt/sources.list.d/rocm.list

# Install ROCm
sudo apt update
sudo apt install rocm-dev rocm-libs rocm-utils

# Add user to required groups
sudo usermod -aG render,video $USER

Logout and login to WSL2 for group changes:

exit
# Then restart WSL2 from PowerShell: wsl

Step 6: Verify GPU access

Test that WSL2 can see your AMD GPU:

# Check if GPU is detected
lspci | grep -i amd

# Test ROCm
/opt/rocm/bin/rocminfo

# Check device files
ls -la /dev/dri/

You should see your AMD GPU listed and device files like card0 and renderD128.

Step 7: Install Docker with AMD support

Install Docker in your WSL2 environment:

# Install Docker
curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh

# Add user to docker group
sudo usermod -aG docker $USER

# Start Docker service
sudo service docker start

Configure Docker for AMD GPUs:

# Create Docker daemon configuration
sudo mkdir -p /etc/docker
sudo tee /etc/docker/daemon.json > /dev/null <<EOF
{
    "runtimes": {
        "rocm": {
            "path": "/usr/bin/rocm-runtime",
            "runtimeArgs": []
        }
    }
}
EOF

# Restart Docker
sudo service docker restart

Step 8: Test AMD GPU in container

Test GPU access with a ROCm container:

# Test basic GPU access
docker run --rm -it \
  --device=/dev/dri \
  --device=/dev/kfd \
  --security-opt seccomp=unconfined \
  --group-add video \
  --group-add render \
  rocm/pytorch:latest \
  rocm-smi

If successful, you’ll see your AMD GPU information displayed.

Step 9: Deploy GPUFlow provider

Create data directory:

sudo mkdir -p /opt/gpuflow
sudo chown $USER:$USER /opt/gpuflow

Run GPUFlow provider with AMD GPU support:

docker run -d \
  --name gpuflow-provider \
  --restart=unless-stopped \
  --device=/dev/dri \
  --device=/dev/kfd \
  --security-opt seccomp=unconfined \
  --group-add video \
  --group-add render \
  --network=host \
  -v /opt/gpuflow:/data \
  -e GPUFLOW_API_KEY="get-from-dashboard" \
  -e GPU_VENDOR="amd" \
  ghcr.io/gpuflow/provider:latest

Verify provider startup:

docker ps
docker logs gpuflow-provider

Look for:

• AMD GPU detected in logs
• ROCm initialization successful
• Network registration complete

Step 10: Windows startup automation

Create startup script:

cat > ~/start-gpuflow-amd.sh << 'EOF'
#!/bin/bash
sudo service docker start
sleep 10  # AMD needs longer startup time
docker start gpuflow-provider
EOF

chmod +x ~/start-gpuflow-amd.sh

Windows startup batch file:

1. Press Win + R, type shell:startup
2. Create GPUFlow-AMD.bat:

@echo off
wsl -d Ubuntu-22.04 -u your-username /home/your-username/start-gpuflow-amd.sh

Managing your AMD provider

Provider management commands:

# Check status
docker ps

# Monitor logs for AMD-specific messages
docker logs gpuflow-provider -f | grep -i rocm

# Restart provider (AMD needs clean restart)
docker stop gpuflow-provider
sleep 5
docker start gpuflow-provider

# Check GPU utilization
rocm-smi

Performance considerations

AMD-specific limitations:

• ROCm has less software support than CUDA
• Some ML frameworks may not work optimally
• Gaming impact may be higher than NVIDIA setups
• Longer container startup times (30-60 seconds)

Resource usage:

• WSL2 overhead: 2-4GB RAM for AMD setup
• ROCm drivers use more system resources
• Gaming performance impact: 5-10% when provider running

Troubleshooting AMD issues

GPU not detected:

# Check ROCm installation
/opt/rocm/bin/rocminfo

# Verify device permissions
ls -la /dev/dri/
ls -la /dev/kfd

# Fix permissions if needed
sudo chmod 666 /dev/kfd
sudo chmod 666 /dev/dri/*

Container startup failures:

# Check ROCm service
sudo systemctl status rocm

# Restart ROCm if needed
sudo systemctl restart rocm

# Clear Docker containers and retry
docker system prune -f

Poor performance:

# Check GPU clocks
rocm-smi --showclocks

# Monitor temperature
rocm-smi --showtemp

Known limitations

Current AMD Windows setup limitations:

• Software compatibility: Many AI frameworks prefer CUDA
• Performance: 15-20% slower than native Linux ROCm
• Stability: ROCm on WSL2 is less stable than NVIDIA CUDA
• Support: Limited community knowledge for troubleshooting

Next steps

Your Windows AMD provider is configured:

1. Create your GPUFlow account
2. Link your AMD hardware in dashboard
3. Configure AMD-optimized listings
4. Monitor performance and earnings

Note

Consider dual-booting Linux for better AMD GPU performance and stability if you’re serious about GPU providing with AMD hardware.