Skip to main content

Linux setup

Prerequisites​

  • Node.js 18+ — install via NodeSource or your package manager
  • NVIDIA GPU with 10GB+ VRAM
  • CUDA Toolkit

Install CUDA​

Ubuntu/Debian:

sudo apt install nvidia-cuda-toolkit

Fedora:

sudo dnf install cuda-toolkit

Arch:

sudo pacman -S cuda

Verify the installation:

nvcc --version
nvidia-smi

Both commands should work. nvidia-smi should show your GPU and driver version. nvcc should show the CUDA compiler version.

Get your token​

Go to c0mpute.ai/worker, login, and get your worker token from the Native Worker section.

Run the worker​

npx @c0mpute/worker --token <your-token>

On first run:

  1. node-llama-cpp compiles with CUDA support
  2. The Qwen2.5 14B model downloads (~9GB)
  3. A benchmark runs to verify GPU performance
  4. The worker connects to the network and starts accepting jobs

Expected benchmark results:

  • RTX 3060 12GB: ~30-40 tok/s
  • RTX 3080: ~50-60 tok/s
  • RTX 4070 Ti: ~60-80 tok/s
  • RTX 4090: ~100+ tok/s

If you see less than 10 tok/s, CUDA is not being used. See Troubleshooting.

Run as a systemd service​

For unattended operation, create a systemd service:

sudo nano /etc/systemd/system/c0mpute-worker.service
[Unit]
Description=c0mpute Native Worker
After=network.target

[Service]
ExecStart=/usr/bin/npx @c0mpute/worker --token YOUR_TOKEN
Restart=always
RestartSec=10
User=your-username
Environment=PATH=/usr/local/bin:/usr/bin:/bin

[Install]
WantedBy=multi-user.target
sudo systemctl daemon-reload
sudo systemctl enable c0mpute-worker
sudo systemctl start c0mpute-worker

Check status:

sudo systemctl status c0mpute-worker
journalctl -u c0mpute-worker -f

Alternatively, use tmux or screen for a simpler setup:

tmux new -s c0mpute
npx @c0mpute/worker --token <your-token>
# Ctrl+B, D to detach