#memory-hierarchy

[ follow ]
Artificial intelligence
fromTheregister
6 days ago

d-Matrix aspires to rack scale AI with JetStream I/O cards

d-Matrix’s JetStream I/O cards enable rack-scale distribution of large AI models with custom FPGA NICs achieving ~2µs latency, Corsair AI accelerators, high SRAM and LPDDR5.
[ Load more ]