Data Structures for Data-Parallel Architectures

数据并行架构的数据结构

• 批准号: 0541448
• 负责人: John Owens
• 金额: $ 20万
• 依托单位: University of California-Davis
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-03-15 至 2011-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0541448&HistoricalAwards=false
• 关键词: Data; Structures; Parallel; Architectures

The project investigates the development of flexible, generic and high-performance data structures for commodity data-parallel architectures. The research will determine appropriate abstractions of data structures suitable for these architectures, determine efficient implementations for the data structures, and develop and build a framework for tuning them for different architectures. The research will address the issues in providing high-performance data structures for GPUs that can be tuned for different parallel architectures. The successful completion of this project will have significant impact in exploiting the growing trend towards multi-core and multi threaded parallel processing in commodity processors, by facilitating the development of parallel software in these environments.

该项目研究了商品数据并行架构的灵活、通用和高性能数据结构的开发。该研究将确定适合这些架构的数据结构的适当抽象，确定数据结构的有效实现，并开发和构建一个框架来针对不同的架构调整它们。 该研究将解决为 GPU 提供高性能数据结构的问题，这些数据结构可以针对不同的并行架构进行调整。该项目的成功完成将对利用商用处理器中多核和多线程并行处理的增长趋势产生重大影响，促进这些环境中并行软件的开发。