ℋ-LU Factorization on Many-Core Systems
Contact the author: Please use for correspondence this email.
Submission date: 16. Jan. 2014 (revised version: November 2014)
published in: Computing and visualization in science, 16 (2013) 3, p. 105-117
DOI number (of the published article): 10.1007/s00791-014-0226-7
MSC-Numbers: 65F05, 65Y05, 65Y20, 68W10, 68W40
Keywords and phrases: hierarchical matrices, parallel algorithms, many-core processors
Download full preprint: PDF (691 kB)
A version of the ℋ-LU factorization is introduced, based on the individual computational tasks occurring during the block-wise ℋ-LU factorization. The dependencies between these tasks form a directed acylic graph, which is used for eﬃcient scheduling on parallel systems. The algorithm is especially suited for many-core processors and shows a much improved parallel scaling behavior compared to previous ℋ-LU factorization algorithms.