基于不连续伽辽金方法的统一GPU并行框架

2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI) Pub Date : 2021-12-04 DOI:10.1109/APS/URSI47566.2021.9703771

Shucheng Huang, Li Xu, Bingqi Liu, Zhonghai Yang, Bin Li

{"title":"基于不连续伽辽金方法的统一GPU并行框架","authors":"Shucheng Huang, Li Xu, Bingqi Liu, Zhonghai Yang, Bin Li","doi":"10.1109/APS/URSI47566.2021.9703771","DOIUrl":null,"url":null,"abstract":"In this paper, we designed a unified parallel DG framework. Based on the characteristics of DG discretization, we built a preprocessing part in the framework. Then designed different kernel functions to match different numerical problems, such as Maxwell's equation and Euler's equation. A speed-up of 160 has been achieved over an equivalent CPU code. Compared with CPU code, the computing efficiency has been greatly improved.","PeriodicalId":6801,"journal":{"name":"2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI)","volume":"27 1","pages":"1891-1892"},"PeriodicalIF":0.0000,"publicationDate":"2021-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Unified GPU Parallel Framework Based on Discontinuous Galerkin Method\",\"authors\":\"Shucheng Huang, Li Xu, Bingqi Liu, Zhonghai Yang, Bin Li\",\"doi\":\"10.1109/APS/URSI47566.2021.9703771\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In this paper, we designed a unified parallel DG framework. Based on the characteristics of DG discretization, we built a preprocessing part in the framework. Then designed different kernel functions to match different numerical problems, such as Maxwell's equation and Euler's equation. A speed-up of 160 has been achieved over an equivalent CPU code. Compared with CPU code, the computing efficiency has been greatly improved.\",\"PeriodicalId\":6801,\"journal\":{\"name\":\"2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI)\",\"volume\":\"27 1\",\"pages\":\"1891-1892\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2021-12-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/APS/URSI47566.2021.9703771\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/APS/URSI47566.2021.9703771","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

在本文中，我们设计了一个统一的并行DG框架。根据离散化的特点，在该框架中建立了预处理部分。然后设计不同的核函数来匹配不同的数值问题，如麦克斯韦方程和欧拉方程。在相同的CPU代码上实现了160的加速。与CPU代码相比，计算效率大大提高。

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

Unified GPU Parallel Framework Based on Discontinuous Galerkin Method

In this paper, we designed a unified parallel DG framework. Based on the characteristics of DG discretization, we built a preprocessing part in the framework. Then designed different kernel functions to match different numerical problems, such as Maxwell's equation and Euler's equation. A speed-up of 160 has been achieved over an equivalent CPU code. Compared with CPU code, the computing efficiency has been greatly improved.

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

2021 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting (APS/URSI)

自引率

0.00%

发文量