Logo image
A 55nm 1-to-8 bit Configurable 6T SRAM based Computing-in-Memory Unit-Macro for CNN-based AI Edge Processors
會議論文

A 55nm 1-to-8 bit Configurable 6T SRAM based Computing-in-Memory Unit-Macro for CNN-based AI Edge Processors

Zhixiao Zhang, Jia-Jing Chen, Xin Si, Yung-Ning Tu, Jian-Wei Su, Wei-Hsing Huang, Jing-Hong Wang, Wei-Chen Wei, Yen-Cheng Chiu, Je-Min Hong, …
Proceedings - 2019 IEEE Asian Solid-State Circuits Conference, A-SSCC 2019, 頁碼.217-218
11/2019

摘要

AI edge processor CIM CNN SRAM Computer Networks and Communications Hardware and Architecture Electrical and Electronic Engineering Risk Reliability and Quality Electronic Optical and Magnetic Materials Safety
This work presents a 1-to-8 bit configurable SRAM CIM unit-macro using (1) a hybrid structure combining 6T-SRAM in-memory binary product-sum (IMBPS) with digital near-memory-computing multi-bit-PS accumulation (DNMPSA) to achieve high read accuracy and compact area, (2) a column-based place-value-grouped weight mapping (CPVGWM) and serial-bit input mapping (SBIN) scheme to facilitate reconfiguration and increase array efficiency for various input/weight precision. A 4Kb configurable 6T-SRAM CIM unit-macro was fabricated using a 55nm CMOS process and foundry 6T cells. The resulting macro achieved 3.5ns access time with 3b PS outputs and gained a 3.91-50x improvement in figure-of-merit, compared to previous SRAM-CIMs.

相關連結

指標

1 檢視次數

詳細資料

Logo image