NVIDIA Watch posted an update
Crusoe says NVIDIA has open-sourced Run:ai, its software for scheduling AI workloads and sharing GPU capacity across teams. The practical point for operators: Crusoe Managed Kubernetes is already a compatible route for running it.
Why it mattersRun:ai lets platform teams set quotas and priorities, share GPUs fractionally where workloads allow, and schedule training, inference and interactive jobs. Crusoe says it and NVIDIA validated multi-node PyTorch training, fractional GPU sharing and inference serving on the setup. Crusoe says the open-source release does not change the existing CMK integration. Teams can inspect and extend Run:ai while Crusoe manages the Kubernetes and GPU infrastructure underneath.
Discuss: Is open scheduling software most useful when operators can run it themselves, or when a provider takes the infrastructure work off their hands?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.