Plain CUDA + SSH

Plain CUDA + SSH

Only port 22 is exposed. Python 3, pip, git and build tools are preinstalled; nvidia-smi works out of the box.

Before you stop

Everything you create lives inside the container on the host's disk and is removed when the rental ends. Push results out before stopping (scp, rsync, git, HF Hub, S3...). From inside the container you can run fairgpu extend 60 to add time, fairgpu mark "epoch 3 done" to annotate the timeline and fairgpu stop when you are done.

Step by step

1. Connect

Only SSH (port 22) is exposed. Use the browser Terminal on the rental page, or ssh -p <port> root@<relay> with the key you stored in your account.

2. Check the toolchain

nvidia-smi
nvcc --version || ls /usr/local/cuda/bin
python3 --version && pip --version

3. Bring your own stack

cd /workspace
python3 -m venv venv && . venv/bin/activate        # the venv lives in /workspace and survives stops
pip install torch --index-url https://download.pytorch.org/whl/cu124

Compile CUDA code with nvcc -o /workspace/app app.cu; expose extra ports by editing the template (e.g. 8000/http).

4. Run

Long jobs: nohup ./run.sh > /workspace/run.log 2>&1 & or the Commands panel so you are emailed on exit.

5. Make it repeatable

Put every apt-get install and build step in a setup script on your own template (Save as template on the rental page) - it runs before you connect, on any host.

Save your results before you stop

• Everything outside /workspace disappears when the rental stops; /workspace survives on the same host (persistent workspace) and anywhere with a FairGPU checkpoint.
• Save checkpoint now on the rental page (or fairgpu snapshot save "label" inside the container) archives /workspace to FairGPU cloud storage; Download workspace gets it to your computer.
• scp -P <port> root@<relay>:/workspace/results ./ copies files out over SSH; rsync and SFTP (WinSCP/FileZilla) work the same way.
• Stop the rental to stop billing. Set an idle auto-stop if you tend to forget, or buy the $0.99 finish alert (SMS + email).

Machines that can run it