Run.ai project service provides access to Nvidia Run.ai project on high-performance clusters based on DGX B200. Nodes are interconnected using an NVIDIA Quantum-2 400 Gbit/s Infiniband and equipped with 2x100 Gb/s Ethernet for IP communication and 2x100 Gb/s internet access.
As a part of the Run.ai project service, you reserve and receive resources equal to the configuration of a node or multiple DGX nodes.
Your environment can be directly accessed using Run.ai webui or Run.ai cli.
Along with Run.ai project service you can reserve dedicated storage for AI from pure flash blade which is directly integrated with our DGX clusters and provided to Run.ai project via NFS with RDMA option and connected on Kubernetes level with PVC.
There is an option to connect external data sources to your Run.ai project like S3 storage or GitHub.