Shuai-bing Lu, J. Pang, Yawen Pan, Jie Tan, Tao Dai
{"title":"基于QEMU的mpi友好型高效静态二进制转换器","authors":"Shuai-bing Lu, J. Pang, Yawen Pan, Jie Tan, Tao Dai","doi":"10.1109/CSMA.2015.14","DOIUrl":null,"url":null,"abstract":"While high performance computing (HPC) is flourishing these years, the lack of HPC applications is increasingly serious. Conventional binary translation focused on desktop applications and embedded software, which could not be scaled up to HPC. This paper proposed a novel static binary translator MPI-QEMU aiming at MPI programs, the most commonly used on HPC platforms. Firstly, the efficient dynamic binary translator QEMU was altered to a static binary translator to translate the source binary to target platforms. Secondly, by analysing the symbol table and global offset table, the library function name and address were extracted and jacketed. Finally, the generated codes were linked with corresponding native library functions including MPI to generate the target executive binary. The experiments on NPB and IMB benchmarks show that MPI-QEMU translates the benchmarks successfully and gains dramatically speedup.","PeriodicalId":205396,"journal":{"name":"2015 International Conference on Computer Science and Mechanical Automation (CSMA)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2015-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A MPI-Friendly Efficient Static Binary Translator Based on QEMU\",\"authors\":\"Shuai-bing Lu, J. Pang, Yawen Pan, Jie Tan, Tao Dai\",\"doi\":\"10.1109/CSMA.2015.14\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"While high performance computing (HPC) is flourishing these years, the lack of HPC applications is increasingly serious. Conventional binary translation focused on desktop applications and embedded software, which could not be scaled up to HPC. This paper proposed a novel static binary translator MPI-QEMU aiming at MPI programs, the most commonly used on HPC platforms. Firstly, the efficient dynamic binary translator QEMU was altered to a static binary translator to translate the source binary to target platforms. Secondly, by analysing the symbol table and global offset table, the library function name and address were extracted and jacketed. Finally, the generated codes were linked with corresponding native library functions including MPI to generate the target executive binary. The experiments on NPB and IMB benchmarks show that MPI-QEMU translates the benchmarks successfully and gains dramatically speedup.\",\"PeriodicalId\":205396,\"journal\":{\"name\":\"2015 International Conference on Computer Science and Mechanical Automation (CSMA)\",\"volume\":\"24 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2015-10-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2015 International Conference on Computer Science and Mechanical Automation (CSMA)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/CSMA.2015.14\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2015 International Conference on Computer Science and Mechanical Automation (CSMA)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CSMA.2015.14","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
A MPI-Friendly Efficient Static Binary Translator Based on QEMU
While high performance computing (HPC) is flourishing these years, the lack of HPC applications is increasingly serious. Conventional binary translation focused on desktop applications and embedded software, which could not be scaled up to HPC. This paper proposed a novel static binary translator MPI-QEMU aiming at MPI programs, the most commonly used on HPC platforms. Firstly, the efficient dynamic binary translator QEMU was altered to a static binary translator to translate the source binary to target platforms. Secondly, by analysing the symbol table and global offset table, the library function name and address were extracted and jacketed. Finally, the generated codes were linked with corresponding native library functions including MPI to generate the target executive binary. The experiments on NPB and IMB benchmarks show that MPI-QEMU translates the benchmarks successfully and gains dramatically speedup.