{"title":"可扩展延迟容忍架构(SCALT)及其评估","authors":"N. Shimizu, D. Mitake","doi":"10.1109/APASIC.1999.824068","DOIUrl":null,"url":null,"abstract":"The deviation of the memory latency is hard to be predicted for in software, especially on the SMP or NUMA systems. As a hardware correspondent method, the multi-thread processor has been devised. However, it is difficult to improve the processor performance with a single program. We have proposed SCALT that uses a buffer in a software context. For the deviation of a latency problem, we have proposed a instruction to check the data arrival existence in a buffer. This paper describes the SCALT, which uses a buffer check instruction, and its performance evaluation results, obtained analyzing the SMP system through event-driven simulation.","PeriodicalId":346808,"journal":{"name":"AP-ASIC'99. First IEEE Asia Pacific Conference on ASICs (Cat. No.99EX360)","volume":"188 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1999-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Scalable latency tolerant architecture (SCALT) and its evaluation\",\"authors\":\"N. Shimizu, D. Mitake\",\"doi\":\"10.1109/APASIC.1999.824068\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The deviation of the memory latency is hard to be predicted for in software, especially on the SMP or NUMA systems. As a hardware correspondent method, the multi-thread processor has been devised. However, it is difficult to improve the processor performance with a single program. We have proposed SCALT that uses a buffer in a software context. For the deviation of a latency problem, we have proposed a instruction to check the data arrival existence in a buffer. This paper describes the SCALT, which uses a buffer check instruction, and its performance evaluation results, obtained analyzing the SMP system through event-driven simulation.\",\"PeriodicalId\":346808,\"journal\":{\"name\":\"AP-ASIC'99. First IEEE Asia Pacific Conference on ASICs (Cat. No.99EX360)\",\"volume\":\"188 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1999-08-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"AP-ASIC'99. First IEEE Asia Pacific Conference on ASICs (Cat. No.99EX360)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/APASIC.1999.824068\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"AP-ASIC'99. First IEEE Asia Pacific Conference on ASICs (Cat. No.99EX360)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/APASIC.1999.824068","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Scalable latency tolerant architecture (SCALT) and its evaluation
The deviation of the memory latency is hard to be predicted for in software, especially on the SMP or NUMA systems. As a hardware correspondent method, the multi-thread processor has been devised. However, it is difficult to improve the processor performance with a single program. We have proposed SCALT that uses a buffer in a software context. For the deviation of a latency problem, we have proposed a instruction to check the data arrival existence in a buffer. This paper describes the SCALT, which uses a buffer check instruction, and its performance evaluation results, obtained analyzing the SMP system through event-driven simulation.