CAREER: Future Values: Reshaping the Future of Instruction Level Parallelism

职业：未来价值：重塑指令级并行的未来

• 批准号: 0347592
• 负责人: Soner Onder
• 金额: --
• 依托单位: Michigan Technological University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-02-15 至 2011-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0347592&HistoricalAwards=false
• 关键词: CAREER; Future; Values; Reshaping; Instruction

CAREER: Future Values: Reshaping the Future of Instruction Level ParallelismAbstractThis research aims at providing an effective solution to the fetch problem of instruction level parallel processors. With the concept of "future values", one can represent traditional programs in a new control-flow form such that on any control flow path the data-flow aspect of the computation is either traditional (i.e., definition of a value precedes its consumers), or reversed (i.e., consumers of a value precedes its definition). The representation hence allows ordering of instructions even when the data dependencies would prohibit doing so in a control-flow graph. As a result, it is possible to seamlessly exploit control independence, use predication to eliminate any branch instruction, and design processors, which can reuse the internally generated dataflow graph as another source of instructions. This approach has the potential to significantly impact the design of processors and optimizing compilers, and positively affect every field of science and commerce that relies on processor performance. The project will also result in course materials and software tool sets that will be made publicly available. Finally, the project will contribute greatly to the development of the recently initiated Computer Science Ph.D. program at Michigan Technological University.

摘要本研究旨在为指令级并行处理器的读取问题提供一个有效的解决方案。有了“未来值”的概念，人们可以用一种新的控制流形式来表示传统程序，这样在任何控制流路径上，计算的数据流方面要么是传统的（即，值的定义先于其消费者），要么是反向的（即，值的消费者先于其定义）。因此，即使数据依赖关系在控制流图中禁止这样做，这种表示也允许对指令进行排序。因此，可以无缝地利用控制独立性，使用预测来消除任何分支指令，并设计可以重用内部生成的数据流图作为另一个指令源的处理器。这种方法有可能对处理器的设计和优化编译器产生重大影响，并对依赖处理器性能的科学和商业领域产生积极影响。该项目还将产生课程材料和软件工具集，这些将向公众开放。最后，该项目将极大地促进密歇根理工大学最近启动的计算机科学博士项目的发展。