On Tuning The Symmetric Sparse Matrix Vector Multiplication With Csr And Tjds
Keywords
CSR; Data structures; Sparse matrix vector product; Storage format; Symmetric sparse matrix; TJDS
Abstract
In this work we present a heuristic to select the appropriate compressed storage format when computing the symmetric SpMV multiplication sequentially. A subset of symmetric sparse matrices were selected from the SPARSITY benchmark suite and extended with other matrices we consider complement them. All matrices were collected from Matrix Market and UF matrix collection. Experimental evidence shows that given a symmetric sparse matrix, predicting what is the more convenient format to use for computing the symmetric SpMV multiplication could be possible. According to our findings, and good rule of thumb, if the average number of non zero coefficients per column (row) is less than 3.5, then the symmetric SpMV multiplication runs up to 1.6× faster using the TJDS format compared to CSR.
Publication Date
1-1-2018
Publication Title
Simulation Series
Volume
50
Issue
4
Number of Pages
36-47
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
Copyright Status
Unknown
Socpus ID
85055258674 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/85055258674
STARS Citation
Aymerich, Edward; Duchateau, Alexandre; Montagne, Euripides; and Plochan, Frank, "On Tuning The Symmetric Sparse Matrix Vector Multiplication With Csr And Tjds" (2018). Scopus Export 2015-2019. 10055.
https://stars.library.ucf.edu/scopus2015/10055