Modal launches serverless GPU clusters with gang scheduling and RDMA

Modal is extending Erik Bernhardsson's pooled compute platform to distributed training and inference, with per-second billing and a single Python decorator.

By · Published

Generic GPU server nodes linked by fiber cables represent Modal’s move to serverless multi-node computing.

Modal is extending Erik Bernhardsson's pooled compute platform to distributed training and inference, with per-second billing and a single Python decorator.