A high-performance distributed storage engine for GPU training and LLM inference, offering petabyte-scale S3-compatible object storage and KV-cache offload over io_uring and RDMA.
3095gpu,high-performance,rdmaA high-performance distributed storage engine for GPU training and LLM inference, offering petabyte-scale S3-compatible object storage and KV-cache offload over io_uring and RDMA.