Quickstart

C3 brings your Slurm workflow to the cloud. Write the same #SBATCH scripts you already use, and C3 handles provisioning GPUs from multiple data centers—so you get compute when you need it, at competitive prices.

Install the CLI

curl -fsSL https://raw.githubusercontent.com/samleeney/c3/main/install.sh | sh

Then authenticate:

c3 login

This opens your browser to sign in. Credentials are stored locally in ~/.c3/.

Run your first job

Clone the examples repo and submit a GPU benchmark:

git clone https://github.com/samleeney/c3-examples
cd c3-examples/jax-matmul
c3 deploy job.sbatch

The job script looks familiar if you've used Slurm:

#!/bin/bash
#SBATCH --job-name=jax-matmul
#SBATCH --gres=gpu:1
#SBATCH --time=00:10:00

#C3 OUTPUT ./results

python train.py

C3 reads your #SBATCH directives, provisions a GPU from the marketplace, runs your script, and uploads the results.

Check status and download results

Monitor your job:

c3 squeue

JOB ID              STATUS      SUBMITTED
job_abc123          RUNNING     2024-01-15 10:30:00

Once complete, pull the results:

c3 pull job_abc123

Your output files are downloaded to ./job_abc123/.

---

Next steps: Learn about submission scripts for advanced options, data management for working with datasets, or browse the marketplace to see available GPUs and pricing.