SPC5: An efficient SpMV framework vectorized using ARM SVE and x86 AVX-512-Reference-Cited by-同舟云学术

SPC5: An efficient SpMV framework vectorized using ARM SVE and x86 AVX-512

Published:2024 Issue:1 Volume:21 Page:203-221
ISSN:1820-0214
Container-title:Computer Science and Information Systems
language:en
Short-container-title:ComSIS

Author:

Regnault Evann¹,Bramas Bérenger²

Affiliation:

1. Strasbourg University UFR de Mathématique et d’Informatique, Strasbourg, France

2. Inria Nancy CAMUS Team, Villers-lès-Nancy, France + ICube laboratory ICPS Team, Illkirch Cedex, France

Abstract

The sparse matrix/vector product (SpMV) is a fundamental operation in scientific computing. Having access to an efficient SpMV implementation is therefore critical, if not mandatory, to solve challenging numerical problems. The ARMbased AFX64 CPU is a modern hardware component that equips one of the fastest supercomputers in the world. This CPU supports the Scalable Vector Extension (SVE) vectorization technology, which has been less investigated than the classic x86 instruction set architectures. In this paper, we describe how we ported the SPC5 SpMV framework on AFX64 by converting AVX512 kernels to SVE. In addition, we present performance results by comparing our kernels against a standard CSR kernel for both Intel-AVX512 and Fujitsu-ARM-SVE architectures.

Publisher

National Library of Serbia

Reference25 articles.

1. Alappat, C., Meyer, N., Laukemann, J., Gruber, T., Hager, G., Wellein, G., Wettig, T.: Ecm modeling and performance tuning of spmv and lattice qcd on a64fx. arXiv preprint arXiv:2103.03013 (2021)

2. AOKI, R., MURAO, H.: Optimization of x265 encoder using arm sve

3. ARM: Arm architecture reference manual supplement, the scalable vector extension (sve), for armv8-a. https://developer.arm.com/documentation/ddi0584/ag/, accessed: July 2020 (version Beta)

4. ARM: Arm c language extensions for sve. https://developer.arm.com/documentation/100987/0000, accessed: July 2020 (version 00bet1)

5. Bramas, B.: Optimization and parallelization of the boundary element method for the wave equation in time domain. Ph.D. thesis, Université de Bordeaux (2016)