DSP coprocessor cell for systolic arrays

Implementation of systolic arrays has been hindered in the past due to a lack of efficient building blocks, or cells, on silicon. The authors present a DSP coprocessor cell for rapid computation of elementary functions. For signal and image processing systolic arrays, several elementary functions typically need to be computed while the interconnection considerations as well as development costs warrant the use of as few types of cells as possible. With the present approach, all of the desired elementary functions can be realized in hardware on a single cell. A 16 bit four-function VLSI chip and an application example-a tracking version of singular-value decomposition, are presented.<<ETX>>

[1]  Vijay K. Jain,et al.  Novel reciprocal and square-root VLSI cell: architecture and application to signal processing , 1991, [Proceedings] ICASSP 91: 1991 International Conference on Acoustics, Speech, and Signal Processing.

[2]  Gerald E. Sobelman,et al.  Design and programming of a flexible, cost-effective systolic array cell for digital signal processing , 1990, IEEE Trans. Acoust. Speech Signal Process..

[3]  Earl E. Swartzlander,et al.  Arithmetic error analysis of a new reciprocal cell , 1992, Proceedings 1992 IEEE International Conference on Computer Design: VLSI in Computers & Processors.

[4]  Per Christian Hansen Reducing the number of sweeps in Hestenes' method , 1989 .

[5]  Granino A. Korn,et al.  Table-Lookup/Interpolation Function Generation for Fixed-Point Digital Computations , 1969, IEEE Transactions on Computers.