Matrix inversion algorithm for linear array processor
The paper presents parallel algorithm for computing the inversion of a dense matrix based on Gauss-Jordan elimination. The algorithm is proposed for the implementation on the linear array at a processor level which operate in a pipeline fashion. Two types of architectures are considered. One which uses serial data transfer (AP/S) and another which uses parallel data transfer (AP/P) between neighboring processors. The speed up of AP/S and AP/P are O(n/2) and O(4n/5), respectively.
Unable to display preview. Download preview PDF.
- 1.L.G. Aleksandrov and H.N. Djidjev, Solving systems of linear equations on a systolic processor, Technical Report, Center of Informatics And Computer Technology, BAN-Sofia, 1–8 (1989).Google Scholar
- 2.A. El-Amawy, A Systolic Architecture for Fast Dense Matrix Inversion, I.E.E.E. Transactions on Computers 38, 449–455 (1989).Google Scholar
- 3.A. El-Amawy and K.R. Dharmarajan, Parallel VLSI algorithm for stable inversion of dense matrices, IEE Proc.,Pt E. 136, 575–580 (1989).Google Scholar