DeepSeek opens source and releases 3FS, high-speed parallel file system optimized for AI data access

According to DeepSeek's announcement, on the fifth day of Open Source Week, its Fire Fryer file system (3FS) was officially open sourced. As a high-performance parallel file system, 3FS can fully utilize modern SSD and RDMA networks to achieve high-speed data access and improve AI model training and inference efficiency. Key performance indicators of 3FS: Realize a total read throughput of 6.6 TiB/s in a 180 node cluster; Achieved a throughput of 3.66 TiB/minute in the 25 node GraySort benchmark test; The peak throughput of a single node KVCache query exceeds 40+GiB/s. 3FS adopts a separated architecture, supporting data preprocessing, dataset loading, checkpoint storage/recovery, embedded vector search, and inference KVCache queries, with strong consistency semantics. DeepSeek launches Smallpond data processing framework to further optimize 3FS data management capabilities.