Accelerating ML Recommendation with over a Thousand RISC-V/Tensor Processors on a 7nm Chip

To accelerate Machine Learning Recommendation and other workloads, Esperanto Technologies has implemented over a thousand low-power RISC-V processors on a single chip along with a distributed on-die memory system. This talk will go over the design of the 7nm ET-SoC-1 chip and how Esperanto extended the RISC-V instruction set to provide over 100 TOPS of performance and to be able to run ML recommendation workloads using less than 20 watts of power.

Bio: Dave Ditzel is the founder and Executive Chairman of Esperanto Technologies. His previous positions include the CEO of Esperanto Technologies, founder and CEO of Transmeta Corp., founder and CEO of ThruChip Communications, VP of Hybrid Computing at Intel, CTO of the SPARC Technology Business at Sun Microsystems and MTS at AT&T Bell Laboratories. He helped create both the Hot Chips and ASPLOS conferences, and co-authored “The Case for the Reduced Instruction Set Computer.” He holds an MS EECS from U.C. Berkeley, and B.S degrees in EE and CS from Iowa State University.

Watch the full video.

Accelerating ML Recommendation with over a Thousand RISC-V/Tensor Processors on a 7nm Chip | Stanford Online

Previous PostAccelerating ML Recommendation With Over 1,000 RISC-V/Tensor Processors on Esperanto's ET-SoC-1 Chip | David R. Ditzel, Esperanto Technologies

Next PostExperimental evaluation of neutron-induced errors on a multicore RISC-V platform | Fernando Fernandes dos Santos (TARAN), Angeliki Kritikakou (TARAN), Olivier Sentieys (TARAN)

Stay Connected With RISC-V