Untether AI in Canada has developed a AI device with over 1400 RISC-V processors called Boqueria for ‘at memory’ computing.
Boqueria, discussed at the HotChips Conference today, is built on TSMC’s 7nm process with 238MB of SRAM. The device has a performance of 2 PetaFlops for FP8 8bit AI data types with a power figure of 30 TFLOPs/W that comes from keeping processing closer to the AI cores with 729 dual RISC-V memory banks.
Because at-memory compute is significantly more energy efficient than traditional von Neumann architectures, more TFlops can be performed for a given power envelope. With the introduction of the runAI devices in 2020, Untether AI saw an energy efficiency level at 8 TOPs/W for the INT8 datatype.