CIM100x

Computation in-Memory Architecture Based on Resistive Devices

Conference Paper (2016)
Author(s)

Said Hamdioui (TU Delft - Computer Engineering)

Mottaqiallah Taouil (TU Delft - Computer Engineering)

Hoang Anh Du Nguyen (TU Delft - Computer Engineering)

Adib Haron (TU Delft - Computer Engineering)

Lei Xie (TU Delft - Computer Engineering)

Koen Bertels (TU Delft - FTQC/Bertels Lab, TU Delft - Quantum & Computer Engineering)

Research Group
Computer Engineering
URL related publication
http://ieeexplore.ieee.org/document/7827975/
More Info
expand_more
Publication Year
2016
Language
English
Research Group
Computer Engineering
Pages (from-to)
95-96
Publisher
VDE
ISBN (electronic)
978-3-8007-4252-3
Event
15th International Workshop on Cellular Nanoscale Networks and their Applications (CNNA 2016) (2016-08-23 - 2016-08-25), Dresden, Germany
Downloads counter
190

Abstract

In this paper, we briefly discuss a new proposed architecture, Computation-In-Memory (CIM) architecture, that targets specifically data-intensive applications. The architecture consists of the interwoven placement of computing and storage units, which are physically tightly integrated together inside a non-volatile memristor crossbar. The architecture has the potential of improving the energy-delay product, computing efficiency and performance per area by at least two orders of magnitude with respect to conventional CMOS architectures.