SRSparse: Generating Codes for High-Performance Sparse Matrix-Vector Semiring Computations

doi:10.1145/3722114

	SRSparse: Generating Codes for High-Performance Sparse Matrix-Vector Semiring Computations
	Du, Zhen 1; Li, Ying 1; Sun, Ninghui 1; Cui, Huimin 1,2; Feng, Xiaobing 1; Li, Jiajia 3
	2025-06-01
发表期刊	ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION
ISSN	1544-3566
卷号	22 期号:2 页码:26
摘要	Sparse matrix-vector semiring computation is a key operation in sparse matrix computations, with performance strongly dependent on both program design and the features of the sparse matrices. Given the diversity of sparse matrices, designing a tailored program for each matrix is challenging. To address this, we propose SRSparse,1 a program generator that creates tailored programs by automatically combining program designing methods to fit specific input matrices. It provides two components: the problem definition configuration, which declares the computation, and the scheduling language, which can be leveraged by an auto-tuner to specify the program designs. The two are lowered to the intermediate representations of SRSparse, the Format IR and Kernel IR, which respectively generate format conversion routine and kernel code. We evaluate SRSparse on four representative sparse kernels and three format conversion routines. For sparse kernels, SRSparse achieves median speedups over handwritten programs: COO (3.50x), CSR-Adaptive (5.36x), CSR5 (2.06x), ELL (1.63x), Gunrock (1.57x), and GraphBLAST (1.96x); over an auto-tuner: AlphaSparse (1.16x); and over a compiler: TACO (1.71x). For format conversion routines, SRSparse achieves median speedups over handwritten implementations: Intel MKL (7.60x), SPARSKIT (2.61x), CUSP (2.77x), and Ginkgo (1.74x); and over a compiler: TACO (4.04x).
关键词	High performance computing sparse matrix computation auto-tuning code generator semiring computation
DOI	10.1145/3722114
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2023YFB3001502] ; National Natural Science Foundation of China[U23B2020] ; National Natural Science Foundation of China[62090024] ; National Natural Science Foundation of China[62232015] ; National Natural Science Foundation of China[62302479] ; Innovation Funding of ICT, CAS[E361010] ; Innovation Funding of ICT, CAS[E261110]
WOS研究方向	Computer Science
WOS类目	Computer Science, Hardware & Architecture ; Computer Science, Theory & Methods
WOS记录号	WOS:001533499400001
出版者	ASSOC COMPUTING MACHINERY
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/42087
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Cui, Huimin
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp & Control Engn, Beijing, Peoples R China 3.North Carolina State Univ NCSU, Dept Comp Sci, Raleigh, NC USA
推荐引用方式 GB/T 7714	Du, Zhen,Li, Ying,Sun, Ninghui,et al. SRSparse: Generating Codes for High-Performance Sparse Matrix-Vector Semiring Computations[J]. ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,2025,22(2):26.
APA	Du, Zhen,Li, Ying,Sun, Ninghui,Cui, Huimin,Feng, Xiaobing,&Li, Jiajia.(2025).SRSparse: Generating Codes for High-Performance Sparse Matrix-Vector Semiring Computations.ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,22(2),26.
MLA	Du, Zhen,et al."SRSparse: Generating Codes for High-Performance Sparse Matrix-Vector Semiring Computations".ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION 22.2(2025):26.