| | |
Summary: Approved for Public Release, Distribution Unlimited.
CEARCH: Cognition Enabled Architecture
Stephen P. Crago, USC/ISI, crago@isi.edu
Janice Onanian McMahon, USC/ISI, jmcmahon@isi.edu
Chris Archer (Northrop Grumman), Krste Asanovic (MIT), Richard Chaung (US Army I2WD), Keith Goolsbey (Cycorp),
Mary Hall (USC/ISI), Christos Kozyrakis (Stanford University), Kunle Olukotun (Stanford University), Una-May O'Reilly
(MIT), Rick Pancoast (Lockheed Martin), Viktor Prasanna (USC), Rodric Rabbah (MIT), Steve Ward (MIT), Donald
Yeung (University of Maryland)
Introduction
The goal of the CEARCH project is to develop a new
computer architecture framework that will enable the
deployment of cognitive systems into DoD platforms. In the
past, cognitive systems have been limited because many
reasoning and learning techniques could not keep up with
real-time data, especially in embedded systems. Cognitive
algorithms have traditionally been developed on commodity
computer architectures, which have been based on
sequential instruction streams with coarse-grain parallelism
available on cluster-based multiprocessors. The CEARCH
architecture will not be constrained by legacy sequential
|
