Design of simple systolic arrays using geometric projections

A systematic and computationally less-intensive method for designing "systolic arrays with simple cells" is described. Recursive convolution is used as an example. The valid schedule and projection vectors are determined from the dependence graph of the algorithm. A projection vector gener...

Full description

Saved in:

Bibliographic Details
Published in:	IEEE transactions on signal processing Vol. 42; no. 4; pp. 985 - 988
Main Authors:	Hussain, S.A., Gupta, S.
Format:	Journal Article
Language:	English
Published:	New York, NY IEEE 01-04-1994 Institute of Electrical and Electronics Engineers
Subjects:	Applied sciences Array signal processing Convolution Design methodology Digital filters Exact sciences and technology Information, signal and communications theory Mathematical methods Pipeline processing Processor scheduling Scheduling algorithm Signal processing algorithms Speech processing Systolic arrays Telecommunications and information theory Geometric transformation Convolution Geometrical projection Recurrence method Systolic network
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	A systematic and computationally less-intensive method for designing "systolic arrays with simple cells" is described. Recursive convolution is used as an example. The valid schedule and projection vectors are determined from the dependence graph of the algorithm. A projection vector generally corresponds to a geometric transformation matrix. The designs of the systolic arrays are obtained by geometric projections.< >
Bibliography:	ObjectType-Article-2 SourceType-Scholarly Journals-1 ObjectType-Feature-1 content type line 23
ISSN:	1053-587X 1941-0476
DOI:	10.1109/78.285670