Title :
A reconfigurable binary/RNS/LNS architecture for DSP
Author_Institution :
University of Florida
Abstract :
An integrated digital signal processing machine will accept data from a plethera of sensors and/or subsystems which may be running at different speeds and with different precision/dynamic range metrics. Data would be manipulated using a set of algorithms which perform a variety of filtering or transform tasks. The design of such a machine considered in this paper, integrates recent advancements in mesh array synthesis with processor technology. The result is a high-throughput GIPS class DSP machine capable of responding to a wide mix of system and user defined DSP problems.
Keywords :
Arithmetic; Computer architecture; Delay effects; Digital signal processing; Dynamic range; Filtering algorithms; Partitioning algorithms; Signal processing algorithms; Subspace constraints; Systolic arrays;
Conference_Titel :
Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '87.
DOI :
10.1109/ICASSP.1987.1169672