Applications of Computation-In-Memory Architectures based on Memristive Devices
Said Hamdioui1,a, Hoang Anh Du Nguyen1, Mottaqiallah Taouil1, Abu Sebastian2,b, Manuel Le Gallo2, Sandeep Pande3, Siebren Schaafsma3, Francky Catthoor4,c, Shidhartha Das5,d, Fernando G. Redondo5, G. Karunaratne6, Abbas Rahimi6 and Luca Benini6,e
1Computer Engineering, TU Delft, Delft, the Netherlands
aS.Hamdioui@tudelft.nl
2IBM Research - Zurich, Switzerland
bASE@zurich.ibm.com
3IMEC, Eindhoven, Netherlands
cFrancky.Catthoor@imec.be
4IMEC., Leuven, Belgium
5ARM Limited, Cambridge, UK
dShidhartha.Das@arm.com
6Integrated Systems Laboratory, ETH Zurich, Switzerland
elbenini@iis.ee.ethz.ch
ABSTRACT
Today's computing architectures and device technologies are unable to meet the increasingly stringent demands on energy and performance posed by emerging applications. Therefore, alternative computing architectures are being explored that leverage novel post-CMOS device technologies. One of these is a Computation-in-Memory architecture based on memristive devices. This paper describes the concept of such an architecture and shows different applications that could significantly benefit from it. For each application, the algorithm, the architecture, the primitive operations, and the potential benefits are presented. The applications cover the domains of data analytics, signal processing, and machine learning.
