Applying data copy to improve memory performance of general array computations.pdf

Applying data copy to improve memory performance of general array computations.pdf

  1. 1、本文档共15页,可阅读全部内容。
  2. 2、有哪些信誉好的足球投注网站(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
  3. 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载
  4. 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
查看更多
Applying data copy to improve memory performance of general array computations

Applying Data Copy to Improve Memory Performance of General Array Computations Qing Yi Department of Computer Science, University of Texas at San Antonio ? Abstract. Data copy is an important compiler optimization which dy- namically rearranges the layout of arrays by copying their elements into local buffers. Traditionally, array copy is considered expensive and has been applied only to the working sets of fully blocked computations. This paper presents an algorithm which automatically applies data copy to optimize the performance of general computations independent of block- ing. The algorithm automatically decides where to insert copy operations and which regions of arrays to copy. In addition, when specialized, it is equivalent to a general scalar replacement algorithm on arbitrary array computations. The algorithm is fully implemented and has been applied to optimize several scientific kernels. The results show that the algorithm is highly effective and that data copy can significantly improve the per- formance of scientific computations, both when combined with blocking and when applied alone without blocking. 1 Introduction Most scientific applications operate on large multi-dimensional arrays that can- not fit in the caches of modern computers. Such computations typically include sequences of loop nests, with each loop selectively accessing elements of arrays. When a loop accesses a non-continuous collection of array elements, that is, when the array elements accessed together close in time are far from each other in the memory, the loop demonstrates poor spatial locality and additionally could incur conflict misses in the cache. Data copy is an important compiler optimization that can dynamically re- arrange the layout of arrays. At the beginning of each computation phase, the transformation can choose to copy a subset of array elements into local buffers. All the relevant array accesses within the computation phase can then be changed to instead operate on

文档评论(0)

l215322 + 关注
实名认证
内容提供者

该用户很懒,什么也没介绍

1亿VIP精品文档

相关文档