ST-CRTS: A Coordinated Performance-Directed Attack on Underperformance

点击翻译按钮获取中文摘要

BackgroundModern high-performance systems are chronically underutilized not because their job queuesare short, but because each task uses just a small fraction of the power available on each node.Compiled code simply does not achieve a reasonable fraction of the available performance, despiteyears of work in academia and in research labs that aimed to improve the situation. Some of theperformance gap is due to architecture and design; however, a portion of it is directly attributableto the design of optimizing compilers. Standard optimization techniques fail to discover and fix such problems, in part because they operate at a procedure-level granularity and in part because the remedy requires the compiler to change its approach to optimizing that part of the code to use another order or another transformation scheme. Intellectual Merit In this project, the PI will develop a feedback-driven adaptive compilation harnessthat specifically targets improvements in small regions of underperformance. To make the regionalscheme work, his team will enhance and reformulate two powerful optimizations, rematerialization andreassociation, and re-scope others so that they operate on a regional rather than a global scope.Broader ImpactThis project will directly support graduate education at Rice. The positive researchresults will be taught in a Rice graduate course on the subject; the material will reach a broadaudience because the teaching materials for the course are used at other institutions. The work inthis project will add to the code base for the Rice University research compiler, which is distributed free ofcharge in source-code form to others in both academia and industry. The PI will use a combination oflocal funds and directed research projects to include undergraduates directly in this work.