共享输入同时预积累的局部相邻关系

arXiv - CS - Mathematical Software Pub Date : 2024-05-13 DOI:arxiv-2405.07819

Johannes Blühdorn, Nicolas R. Gauger

{"title":"共享输入同时预积累的局部相邻关系","authors":"Johannes Blühdorn, Nicolas R. Gauger","doi":"arxiv-2405.07819","DOIUrl":null,"url":null,"abstract":"In shared-memory parallel automatic differentiation, shared inputs among\nsimultaneous thread-local preaccumulations lead to data races if Jacobians are\naccumulated with a single, shared vector of adjoint variables. In this work, we\ndiscuss the benefits and tradeoffs of re-enabling such preaccumulations by a\ntransition to suitable local adjoint variables. In particular, we assess the\nperformance of mapped local adjoints in discrete adjoint computations in the\nmultiphysics simulation suite SU2.","PeriodicalId":501256,"journal":{"name":"arXiv - CS - Mathematical Software","volume":"12 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Local Adjoints for Simultaneous Preaccumulations with Shared Inputs\",\"authors\":\"Johannes Blühdorn, Nicolas R. Gauger\",\"doi\":\"arxiv-2405.07819\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In shared-memory parallel automatic differentiation, shared inputs among\\nsimultaneous thread-local preaccumulations lead to data races if Jacobians are\\naccumulated with a single, shared vector of adjoint variables. In this work, we\\ndiscuss the benefits and tradeoffs of re-enabling such preaccumulations by a\\ntransition to suitable local adjoint variables. In particular, we assess the\\nperformance of mapped local adjoints in discrete adjoint computations in the\\nmultiphysics simulation suite SU2.\",\"PeriodicalId\":501256,\"journal\":{\"name\":\"arXiv - CS - Mathematical Software\",\"volume\":\"12 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-05-13\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Mathematical Software\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2405.07819\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Mathematical Software","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2405.07819","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

在共享内存并行自动微分中，如果雅各布数是用单个共享的辅助变量向量累积的，那么同时进行的线程本地预累积之间的共享输入就会导致数据竞赛。在这项工作中，我们讨论了通过转换为合适的本地辅助变量来重新启用这种预累积的好处和权衡。特别是，我们在多物理场仿真套件 SU2 中评估了离散邻接计算中映射局部邻接的性能。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Local Adjoints for Simultaneous Preaccumulations with Shared Inputs

In shared-memory parallel automatic differentiation, shared inputs among simultaneous thread-local preaccumulations lead to data races if Jacobians are accumulated with a single, shared vector of adjoint variables. In this work, we discuss the benefits and tradeoffs of re-enabling such preaccumulations by a transition to suitable local adjoint variables. In particular, we assess the performance of mapped local adjoints in discrete adjoint computations in the multiphysics simulation suite SU2.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

arXiv - CS - Mathematical Software

自引率

0.00%

发文量