Volume 41 Issue 1
May  2017
Article Contents

HUANG Yan, GU Zhi-min. Performance Analysis and Optimization of Prefetching Thread on CMPs[J]. Journal of University of Electronic Science and Technology of China, 2012, 41(1): 85-91. doi: 10.3969/j.issn.1001-0548.2012.01.017
Citation: HUANG Yan, GU Zhi-min. Performance Analysis and Optimization of Prefetching Thread on CMPs[J]. Journal of University of Electronic Science and Technology of China, 2012, 41(1): 85-91. doi: 10.3969/j.issn.1001-0548.2012.01.017

Performance Analysis and Optimization of Prefetching Thread on CMPs

doi: 10.3969/j.issn.1001-0548.2012.01.017
  • Received Date: 2010-04-28
  • Rev Recd Date: 2011-02-19
  • Publish Date: 2012-02-15
  • Memory latency has become a critical bottleneck in achieving high performance on modern processors. Prefetching thread based on multiprocessor (CMP) is a well known approach to reducing memory latency and has been explored in different applications. In this paper, we analyze the performance of prefetching thread for memory intensive applications. The analysis and experimental result show that computation/access latency ratio (CALR) of hotspots has an important affect on prefethcing performance. When the memory access ratio between main thread and prefetching thread is close to (1-CALR)/(1+CALR), prefetching thread gains better performance. The thread prefetching performance of several benchmarks from Olden and SPEC2006 benchmark suite is tested, and the experimental results reflect the impact of different memory access ratio between Prefetching thread and main thread.
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(3496) PDF downloads(60) Cited by()

Related
Proportional views

Performance Analysis and Optimization of Prefetching Thread on CMPs

doi: 10.3969/j.issn.1001-0548.2012.01.017

Abstract: Memory latency has become a critical bottleneck in achieving high performance on modern processors. Prefetching thread based on multiprocessor (CMP) is a well known approach to reducing memory latency and has been explored in different applications. In this paper, we analyze the performance of prefetching thread for memory intensive applications. The analysis and experimental result show that computation/access latency ratio (CALR) of hotspots has an important affect on prefethcing performance. When the memory access ratio between main thread and prefetching thread is close to (1-CALR)/(1+CALR), prefetching thread gains better performance. The thread prefetching performance of several benchmarks from Olden and SPEC2006 benchmark suite is tested, and the experimental results reflect the impact of different memory access ratio between Prefetching thread and main thread.

HUANG Yan, GU Zhi-min. Performance Analysis and Optimization of Prefetching Thread on CMPs[J]. Journal of University of Electronic Science and Technology of China, 2012, 41(1): 85-91. doi: 10.3969/j.issn.1001-0548.2012.01.017
Citation: HUANG Yan, GU Zhi-min. Performance Analysis and Optimization of Prefetching Thread on CMPs[J]. Journal of University of Electronic Science and Technology of China, 2012, 41(1): 85-91. doi: 10.3969/j.issn.1001-0548.2012.01.017

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return