{"title":"经验椭圆型偏微分方程矢量化软件","authors":"W. Wand, Y. Li","doi":"10.1145/98949.99048","DOIUrl":null,"url":null,"abstract":"ELLPACK is a FORTRAN-based package for solv ing elliptic partial differential equations; this package contains over 100 modules, each typically consisting of a number of subroutines. The enhancement of six ELLPACK modules to take advantage of the vector pro cessing capabilities of Cray supercomputers is discussed and variations in performance due to the use of three different compilers are also noted. Speedup factors arc computed which quantify improvements due to program optimization as well as the compiler effects. Graphic per formance profiles showing CPU time versus grid size arc presented. 1. Statement of the Problem One of the most important models of physical phenomena is the boundary-value problem. Examples of such phenomena include heat flow and diffusion, gravita tional and electrostatic potentials, fluid dynamics, and loading of beams, plates, and other objects. Because of this, much effort has been devoted to the development of reliable methods for numerically solving these equations. The underlying mathematical problem addressed by these methods is a second-order elliptic boundary-value prob lem consisting of a partial differential equation (PDE) Lu = auu + Ibiiry + ctiyy + duz + euy + fu = g (1) defined on a domain R and its accompanying boundary conditions","PeriodicalId":409883,"journal":{"name":"ACM-SE 28","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1990-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Experiences vectorizing software for elliptic partial differential equations\",\"authors\":\"W. Wand, Y. Li\",\"doi\":\"10.1145/98949.99048\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"ELLPACK is a FORTRAN-based package for solv ing elliptic partial differential equations; this package contains over 100 modules, each typically consisting of a number of subroutines. The enhancement of six ELLPACK modules to take advantage of the vector pro cessing capabilities of Cray supercomputers is discussed and variations in performance due to the use of three different compilers are also noted. Speedup factors arc computed which quantify improvements due to program optimization as well as the compiler effects. Graphic per formance profiles showing CPU time versus grid size arc presented. 1. Statement of the Problem One of the most important models of physical phenomena is the boundary-value problem. Examples of such phenomena include heat flow and diffusion, gravita tional and electrostatic potentials, fluid dynamics, and loading of beams, plates, and other objects. Because of this, much effort has been devoted to the development of reliable methods for numerically solving these equations. The underlying mathematical problem addressed by these methods is a second-order elliptic boundary-value prob lem consisting of a partial differential equation (PDE) Lu = auu + Ibiiry + ctiyy + duz + euy + fu = g (1) defined on a domain R and its accompanying boundary conditions\",\"PeriodicalId\":409883,\"journal\":{\"name\":\"ACM-SE 28\",\"volume\":\"1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1990-04-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"ACM-SE 28\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1145/98949.99048\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM-SE 28","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/98949.99048","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
ELLPACK是一个基于fortran的求解椭圆型偏微分方程的程序包;该包包含100多个模块,每个模块通常由许多子例程组成。讨论了六个ELLPACK模块的增强,以利用Cray超级计算机的矢量处理能力,并指出了由于使用三种不同的编译器而导致的性能变化。计算加速因子,量化由于程序优化和编译器效果而带来的改进。图形性能配置文件显示CPU时间与网格大小。1. 边值问题是研究物理现象最重要的模型之一。这种现象的例子包括热流和扩散、重力和静电势、流体动力学以及梁、板和其他物体的载荷。正因为如此,人们投入了大量的精力来开发可靠的数值解这些方程的方法。这些方法所解决的基本数学问题是一个二阶椭圆型边值问题,由定义在域R上的偏微分方程(PDE) Lu = auu + ibiity + ctiy + duz + euy + fu = g(1)及其伴随的边界条件组成
Experiences vectorizing software for elliptic partial differential equations
ELLPACK is a FORTRAN-based package for solv ing elliptic partial differential equations; this package contains over 100 modules, each typically consisting of a number of subroutines. The enhancement of six ELLPACK modules to take advantage of the vector pro cessing capabilities of Cray supercomputers is discussed and variations in performance due to the use of three different compilers are also noted. Speedup factors arc computed which quantify improvements due to program optimization as well as the compiler effects. Graphic per formance profiles showing CPU time versus grid size arc presented. 1. Statement of the Problem One of the most important models of physical phenomena is the boundary-value problem. Examples of such phenomena include heat flow and diffusion, gravita tional and electrostatic potentials, fluid dynamics, and loading of beams, plates, and other objects. Because of this, much effort has been devoted to the development of reliable methods for numerically solving these equations. The underlying mathematical problem addressed by these methods is a second-order elliptic boundary-value prob lem consisting of a partial differential equation (PDE) Lu = auu + Ibiiry + ctiyy + duz + euy + fu = g (1) defined on a domain R and its accompanying boundary conditions