Using Arm’s scalable vector extension on stencil codes
Using Arm’s scalable vector extension on stencil codes
复制标题
在模板代码上使用 Arm 的可扩展矢量扩展
DOI：
--
复制
发表时间：
2019
影响因子：
3.3
通讯作者：
Miquel Moretó
中科院分区：
文献类型：
--
作者：
Adrià Armejach;Helena Caminal;J. M. Cebrian;Rubén Langarita;Rekai González;Chris Adeniyi;M. Valero;Marc Casas;Miquel Moretó
Data-level parallelism is frequently ignored or underutilized. Achieved through vector/SIMD capabilities, it can provide substantial performance improvements on top of widely used techniques such as thread-level parallelism. However, manual vectorization is a tedious and costly process that needs to be repeated for each specific instruction set or register size. In addition, automatic compiler vectorization is susceptible to code complexity, and usually limited due to data and control dependencies. To address some of these issues, Arm recently released a new vector ISA, the scalable vector extension (SVE), which is vector-length agnostic (VLA). VLA enables the generation of binary files that run regardless of the physical vector register length. In this paper, we leverage the main characteristics of SVE to implement and optimize stencil computations, ubiquitous in scientific computing. We show that SVE enables easy deployment of textbook optimizations like loop unrolling, loop fusion, load trading or data reuse. Our detailed simulations using vector lengths ranging from 128 to 2048 bits show that these optimizations can lead to performance improvements over straightforward vectorized code of up to 1.57×documentclass[12pt]{minimal} usepackage{amsmath} usepackage{wasysym} usepackage{amsfonts} usepackage{amssymb} usepackage{amsbsy} usepackage{mathrsfs} usepackage{upgreek} setlength{oddsidemargin}{-69pt} egin{document}$$ imes$$end{document}. In addition, we show that certain optimizations can hurt performance due to reduced arithmetic intensity and instruction overheads, and provide insight useful for compiler optimizers.
Data-level parallelism is frequently ignored or underutilized. Achieved through vector/SIMD capabilities, it can provide substantial performance improvements on top of widely used techniques such as thread-level parallelism. However, manual vectorization is a tedious and costly process that needs to be repeated for each specific instruction set or register size. In addition, automatic compiler vectorization is susceptible to code complexity, and usually limited due to data and control dependencies. To address some of these issues, Arm recently released a new vector ISA, the scalable vector extension (SVE), which is vector-length agnostic (VLA). VLA enables the generation of binary files that run regardless of the physical vector register length. In this paper, we leverage the main characteristics of SVE to implement and optimize stencil computations, ubiquitous in scientific computing. We show that SVE enables easy deployment of textbook optimizations like loop unrolling, loop fusion, load trading or data reuse. Our detailed simulations using vector lengths ranging from 128 to 2048 bits show that these optimizations can lead to performance improvements over straightforward vectorized code of up to 1.57×documentclass[12pt]{minimal} usepackage{amsmath} usepackage{wasysym} usepackage{amsfonts} usepackage{amssymb} usepackage{amsbsy} usepackage{mathrsfs} usepackage{upgreek} setlength{oddsidemargin}{-69pt} egin{document}$$ imes$$end{document}. In addition, we show that certain optimizations can hurt performance due to reduced arithmetic intensity and instruction overheads, and provide insight useful for compiler optimizers.
影响因子：
4.1
作者：
Komatitsch, Dimitri;Erlebacher, Gordon;Michea, David
通讯作者：
Michea, David