arrow
返回

Level-Based Blocking for Sparse Matrices: Sparse Matrix-Power-Vector Multiplication

delete2023-02-01
delete3
delete
OA
AI
C
Christie Louis Alappat *
G
Georg Hager
O
Olaf Schenk
G
Gerhard Wellein
DOI:10.1109/TPDS.2022.3223512delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
The multiplication of a sparse matrix with a dense vector (SpMV) is a key component in many numerical schemes and its performance is known to be severely limited by main memory access. Several numerical schemes require the multiplication of a sparse matrix polynomial with a dense vector which is typically implemented as a sequence of SpMVs. This results in low performance and ignores the potential to increase the arithmetic intensity by reusing the matrix data from cache. In this work we use the recursive algebraic coloring engine (RACE) to enable blocking of sparse matrix data across the polynomial computations. In the graph representing the sparse matrix we form levels using a breadth-first search. Locality relations of these levels are then used to improve spatial and temporal locality when accessing the matrix data and to implement an efficient multithreaded parallelization. Our approach is independent of the matrix structure and avoids shortcomings of existing blocking strategies in terms of hardware efficiency and parallelization overhead. We quantify the quality of our implementation using performance modelling and demonstrate speedups of up to 3x and 5x compared to an optimal SpMV-based baseline on a single multicore chip of recent Intel and AMD architectures. Various numerical schemes like $s$s-step Krylov solvers, polynomial preconditioners and power clustering algorithms will benefit from our development.
Keyword:
Algorithm design and analysis
computer architecture
graph algorithms
kernel optimization
memory hierarchies
performance evaluation
sparse matrices

期刊

IEEE Transactions on Parallel and Distributed Systems 封面图
IEEE Transactions on Parallel and Distributed Systems
IF:
6
论文数:
5.2K
被引数:
1.1W

机构

U
Universita della Svizzera Italiana
学者数:
3.3K
论文数: 2.8K
被引数: 3
U
University of Erlangen Nuremberg
学者数:
3.2W
论文数: 2.6W
被引数: 29
引用论文

引用论文

Author response: Preserving neuromuscular synapses in ALS by stimulating MuSK with a therapeutic agonist antibody
err
IF0
err2018-02-02
err0
errOAAI
errSarah Cantor; Wei Zhang; Nicolas Delestrée; Leonor Remédio; George Z Mentis; Steven J Burden
err分享
err收藏
Peptide analysis by capillary (zone) electrophoresis
err1997-10-01
err0
PREAI
errIrene Messana; Diana Valeria Rossetti; Loredana Cassiano; Francesco Misiti; Bruno Giardina; Massimo Castagnola
err分享
err收藏
Observations on soil fungistasis
err1973-04-01
err0
PREAI
errR. R. Mishra; R. S. Kanaujia
err分享
err收藏
Surface glycoprotein, gp24, involved in early adhesion of Dictyostelium discoideum
err1987-05-01
err0
PREAI
errDavid A. Knecht; Danny L. Fuller; William F. Loomis
err分享
err收藏
err分享
err收藏
学者 查看更多内容