SHF: Small: Pattern-Aware Design

点击翻译按钮获取中文摘要

Current processors employ aggressive prediction mechanisms to improve performance and reduce power. To design effective optimizations, it is increasingly important to understand and quantify a program's dynamic behavior. However, today, most of these optimizations rely on heuristics and are fairly ad-hoc. After some patterns are observed, a hardware decision is made and the design space of the hardware optimization is explored through simulation to determine the best performing configuration. However, because the design is targeted for observed and/or anticipated patterns, some dynamic behavior is not captured and remains undetected.This project develops a framework for quantitatively analyzing a program's behavior in terms of regularities and patterns to provide insights into the design of next-generation hardware prediction mechanisms. Inspired by DNA discovery tools, the research adopts algorithms used in string processing, compression and bioinformatics, in order to summarize applications' dynamic data reference and branch outcome behaviors. The findings from the methods and tools developed in this project will have a big impact on the design and efficiency of future hardware components. The PI will collaborate with industry partners to further increase the broader impact. The research will engage graduate and undergraduate students and promote the participation of underrepresented groups.