Keynote 4: Can Parallel Software Catch up with Parallel Hardware? Trends in Automatic Parallelization

International Conference in IT & Application Pub Date : 2005-07-04 DOI:10.1109/ICITA.2005.170

M. Guo

{"title":"Keynote 4: Can Parallel Software Catch up with Parallel Hardware? Trends in Automatic Parallelization","authors":"M. Guo","doi":"10.1109/ICITA.2005.170","DOIUrl":null,"url":null,"abstract":"Summary form only given. Supercomputers have to be proved powerful for various fields including the development of advanced technologies such as large-scale scientific and engineering computing, new material manufacture, nuclear fusion simulation, and automotive design. On October 20, 2004, NEC Corporation announced the availability of their new supercomputer 'SX-8' the world's most powerful vector supercomputer with a peak processing performance of 65TFLOPS. In last few years, the hardware of supercomputers has undergone rapid development - from Earth simulator to SX-8, only goes through 3 years - the peak performance of SX-8 exceeds 1.8 times over Earth simulator. However, the parallel software - especially parallel programming tools - is still underdevelopment. We still use MPI, high performance Fortran and OpenMP mostly for our parallel programming tasks. In fact, these languages and libraries are difficult to use for most of scientific and engineering users. In this talk, we outlook some of the existing parallel language and automatic parallelization tools and also, we address the potential technologies of automatic parallelization. Finally, we summarize how parallel software to make effort to catch up with the development of supercomputer hardware.","PeriodicalId":388442,"journal":{"name":"International Conference in IT & Application","volume":"56 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2005-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Conference in IT & Application","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICITA.2005.170","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 0

Abstract

Summary form only given. Supercomputers have to be proved powerful for various fields including the development of advanced technologies such as large-scale scientific and engineering computing, new material manufacture, nuclear fusion simulation, and automotive design. On October 20, 2004, NEC Corporation announced the availability of their new supercomputer 'SX-8' the world's most powerful vector supercomputer with a peak processing performance of 65TFLOPS. In last few years, the hardware of supercomputers has undergone rapid development - from Earth simulator to SX-8, only goes through 3 years - the peak performance of SX-8 exceeds 1.8 times over Earth simulator. However, the parallel software - especially parallel programming tools - is still underdevelopment. We still use MPI, high performance Fortran and OpenMP mostly for our parallel programming tasks. In fact, these languages and libraries are difficult to use for most of scientific and engineering users. In this talk, we outlook some of the existing parallel language and automatic parallelization tools and also, we address the potential technologies of automatic parallelization. Finally, we summarize how parallel software to make effort to catch up with the development of supercomputer hardware.

查看原文本刊更多论文

主题演讲4:并行软件能赶上并行硬件吗?自动并行化的趋势

只提供摘要形式。超级计算机必须在各个领域得到证明，包括大规模科学和工程计算、新材料制造、核聚变模拟和汽车设计等先进技术的发展。2004年10月20日，NEC公司宣布他们的新超级计算机“SX-8”的可用性，这是世界上最强大的矢量超级计算机，峰值处理性能为65TFLOPS。近年来，超级计算机硬件发展迅速，从地球模拟器到SX-8，仅用了3年时间，SX-8的峰值性能超过地球模拟器1.8倍。然而，并行软件——尤其是并行编程工具——仍处于开发阶段。我们仍然使用MPI，高性能Fortran和OpenMP，主要用于并行编程任务。事实上，对于大多数科学和工程用户来说，这些语言和库很难使用。在这次演讲中，我们展望了一些现有的并行语言和自动并行化工具，同时我们也讨论了自动并行化的潜在技术。最后，我们总结了并行软件如何努力赶上超级计算机硬件的发展。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

International Conference in IT & Application

自引率

0.00%

发文量