Abstract
In this paper we present a fast parallel Gaussian algorithm for solving a large sparse system on a distributed memory model(DMM) by designing dynamic local pivoting and proper data distribution schemes which sharply reduce the communication time and achieve a balanced distribution of computation load. We shall show that the computation overhead caused by the local pivoting can be compensated by efficient data processing using those two schemes.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have