ADVANCE Fellows Award: Caches As Filters Benchmark Characterization and Demonstration

CNS-0340813Dee魏克莱研究所：弗吉尼亚大学题目：缓存作为过滤器基准测试特征和演示此高级研究员奖为一个项目提供支持，该项目使用缓存即过滤器框架来表征基准测试套件，特别是出于缓存设计的目的，并演示了该框架在设计小规模缓存系统时的使用。缓存即过滤器框架解决了使用基准方法进行内存层次结构设计的困难。该框架的演示可以鼓励体系结构社区使用包含更多分析的设计方法，从而加深对设计问题的理解。该项目的智力优势在于有机会为系统的缓存设计方法提供理论基础。这些基准测试的特征提供了对现实世界程序中发现的存储器系统行为类型的有用洞察，从而有助于当前对计算机存储器管理领域的理解。更广泛的影响包括开发可供计算机体系结构研究小组使用的新工具。这可能包括开发新的方法，研究人员将框架问题和交流，以及向学生传达他们的想法。该项目包括在支持性环境中的研究生和本科生。我们坚定地致力于通过研究生课程、网站和出版物进行传播。

CNS-0340813Dee WeikleInstitution: University of VirginiaTitle: Caches as Filters Benchmark Characterization and DemonstrationThis ADVANCE Fellows award provides support to for a project to use the caches-as-filters framework to characterize benchmark suites, specifically for the purpose of cache design, and to demonstrate the use of the framework on the design of a small-scale caching system. The caches-as-filters framework addresses difficulties in using a benchmark methodology for memory hierarchy design. The demonstration of this framework could encourage the architecture community to use design methodologies that include more analysis resulting in a deeper understanding of design issue.The intellectual merit of this project lies in the opportunity to provide a theoretic basis for a systematic approach to cache design. The characterizations of the benchmarks provide useful insight into the kinds of memory systems behavior found in real-world program and thus may contribute to the current understanding in the field of computer memory management. The broader impacts include development of new tools that can be used by computer architecture research groups. This may include development of new ways in which researchers will frame problems and communicate about them as well as convey their ideas to students. The project includes both graduate and undergraduate students within a supportive environment. There is a strong commitment to dissemination, through a graduate course, web site and publications.