完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.author | Chua, Huey-Ting | |
dc.contributor.author | Pean, Der-Lin | |
dc.contributor.author | Chen, Cheng | |
dc.date.accessioned | 2009-06-02T07:20:45Z | |
dc.date.accessioned | 2020-05-29T06:18:24Z | - |
dc.date.available | 2009-06-02T07:20:45Z | |
dc.date.available | 2020-05-29T06:18:24Z | - |
dc.date.issued | 2006-10-30T01:18:51Z | |
dc.date.submitted | 1999-12-20 | |
dc.identifier.uri | http://dspace.fcu.edu.tw/handle/2377/2804 | - |
dc.description.abstract | In general, synchronization mechanism can be used to preserve dependence constraints of any nested loops, and it can be combined with loop scheduling scheme to form a uniform framework. Meanwhile, correct execution order and balance workload distribution will be achieved. In this paper, we propose a new scheduling scheme called M-hopping method to schedule non-uniform dependence doubly nested loop on multiprocessor systems. To initialize a set of hopping information, our approach is based on the concept of minimum dependence distance. During runtime, hopping information will be used to adjust number of parallelizable iterations. According to our experimental results, if loops carry sufficient parallelism, our proposed method will reliably exploit parallelism, and outperform most of the existing non-uniform dependence loop scheduling schemes by 20.29 in average. | |
dc.description.sponsorship | 淡江大學, 台北縣 | |
dc.format.extent | 8p. | |
dc.format.extent | 921854 bytes | |
dc.format.mimetype | application/pdf | |
dc.language.iso | zh_TW | |
dc.relation.ispartofseries | 1999 NCS會議 | |
dc.subject | Scheduling | |
dc.subject | Parallelizing Compiler | |
dc.subject | Loop | |
dc.subject | Multiprocessor | |
dc.subject | Synchronization | |
dc.subject.other | Microarchitecture | |
dc.title | Non-uniform Dependence Loops | |
分類: | 1999年 NCS 全國計算機會議 |
文件中的檔案:
檔案 | 描述 | 大小 | 格式 | |
---|---|---|---|---|
ce07ncs001999000066.pdf | 900.25 kB | Adobe PDF | 檢視/開啟 |
在 DSpace 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。