Skip to main navigation Skip to search Skip to main content

Memory Processing Unit for in-memory processing

  • Rotem Ben Hur
  • , Shahar Kvatinsky

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

43 Scopus citations

Abstract

Performance and energy of modern computers, usually built as von Neumann machines, are primarily limited by data transfer from the memory to the CPU and vice versa. Only a true non-von Neumann architecture, where data is processed and stored within the same unit can remove this bottleneck. Using emerging non-volatile resistive memory technologies (namely, memristors) enables the development of Memory Processing Unit (MPU)-a novel non-von Neumann architecture. MPU relies on adding computing capabilities to the memristive memory cells without changing the basic memory array structure, and is compatible with existing computing systems. This paper describes the MPU architecture and examines its controller.

Original languageEnglish
Title of host publicationProceedings of the 2016 IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2016
PublisherPresses Polytechniques et Universitaires Romandes
Pages171-172
Number of pages2
ISBN (Electronic)9781450343305
DOIs
StatePublished - 14 Sep 2016
Externally publishedYes
Event2016 IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2016 - Beijing, China
Duration: 18 Jul 201620 Jul 2016

Publication series

NameProceedings of the 2016 IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2016

Conference

Conference2016 IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2016
Country/TerritoryChina
CityBeijing
Period18/07/1620/07/16

Keywords

  • CPU
  • Crossbar memory
  • MAGIC
  • MPU
  • Memristive systems
  • logic design
  • memory controller
  • memristor

ASJC Scopus subject areas

  • Hardware and Architecture
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Memory Processing Unit for in-memory processing'. Together they form a unique fingerprint.

Cite this