Invention Application
- Patent Title: OPTIMIZING OFF-CHIP MEMORY ACCESSES ON A NEURAL NETWORK HARDWARE ACCELERATOR
-
Application No.: PCT/US2022/023739Application Date: 2022-04-06
-
Publication No.: WO2022216878A1Publication Date: 2022-10-13
- Inventor: ZHANG, Dan , HUDA, Safeen , MIRHOSEINI, Azalia , GOLDIE, Anna Darling , SONGHORI, Ebrahim
- Applicant: GOOGLE LLC
- Applicant Address: 1600 Amphitheatre Parkway
- Assignee: GOOGLE LLC
- Current Assignee: GOOGLE LLC
- Current Assignee Address: 1600 Amphitheatre Parkway
- Agency: PORTNOV, Michael
- Priority: US63/171,526 2021-04-06
- Main IPC: G06N3/063
- IPC: G06N3/063 ; G06N3/04 ; G06N5/00 ; G06N3/10
Abstract:
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining a hardware datapath for a hardware accelerator computer chip.
Information query
