We launched a bare-bones version of gpudeploy.com a few months back.
Basically, it's a site for renting GPU-enabled servers and the traffic gets routed to idle compute resources (= great price + high availability).
We aim for a smooth user experience by being very opinionated with configurations and software and sorting out bad machines immediately.
Since then, we've thrown out the air beds.
We had a massive influx of users and people renting out their GPUs (tons of H100s, A100s).
We also improved the user experience and introduced major new features, such as default Jupyter Notebook support and a control panel for operators: it allows you to monitor your machine and its usage, schedule downtime (highly requested), and get pricing stats.
So, if you don't have any plans for this weekend, you should try plagiarizing Shakespeare: setting up nanoGPT on one of our machines literally takes < 5 minutes.
We launched a bare-bones version of gpudeploy.com a few months back. Basically, it's a site for renting GPU-enabled servers and the traffic gets routed to idle compute resources (= great price + high availability). We aim for a smooth user experience by being very opinionated with configurations and software and sorting out bad machines immediately.
Since then, we've thrown out the air beds. We had a massive influx of users and people renting out their GPUs (tons of H100s, A100s). We also improved the user experience and introduced major new features, such as default Jupyter Notebook support and a control panel for operators: it allows you to monitor your machine and its usage, schedule downtime (highly requested), and get pricing stats.
So, if you don't have any plans for this weekend, you should try plagiarizing Shakespeare: setting up nanoGPT on one of our machines literally takes < 5 minutes.