Work

LAcc

Year: 2019

Type: conference-paper

Abstract: PIM (Processing-in-memory)-based CNN (Convolutional neural network) accelerators leverage the characteristics of basic memory cells to enable simple logic and arithmetic operations so that the bandwidth constraint can be effectively alleviated. However, it remains a major challenge to support multiplication operations efficiently on PIM accelerators, in particular, DRAM-based PIM accelerators. This has prevented PIM-based accelerators from being immediately adopted for accurate CNN inference.

Source: ACM/IEEE Design Automation Conference (DAC)

DOI: 10.1145/3316781.3317845

Authors: Quan Deng, Youtao Zhang, Minxuan Zhang, Jun Yang

Institutions: National University of Defense Technology, University of Pittsburgh

Language: English

Topic: Advanced Memory and Neural Computing

Subfield: Electrical and Electronic Engineering

Field: Engineering

Domain: Physical Sciences

Open Access status: bronze

Funders: National Science Foundation, National Natural Science Foundation of China, Division of Computing and Communication Foundations

Awards: G2187124540, G4863315741, SHF: Small: Approximate-Computing Enabled Robust 3D NAND Flash Memories, SHF: Small: Architectural Support for Reliable ReRAM Crossbar Memory