On solving separable block tridiagonal linear systems using a GPU implementation of radix-4 PSCR method

Open Webpage

Mirko Myllykoski, Tuomo Rossi, Jari Toivanen

Published: 01 Jan 2018, Last Modified: 06 Nov 2025J. Parallel Distributed Comput. 2018EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights •A generalized GPU implementation of the radix-4 PSCR method is presented.•Applicable to real and complex valued separable block tridiagonal linear systems.•Up to 24-fold speedups when compared to a single-threaded CPU implementation.