ANALYZING PARALLEL PROGRAM PERFORMANCE USING NORMALIZED PERFORMANCE INDEXES AND TRACE TRANSFORMATION TECHNIQUES

In this paper we describe how a performance tuning tool-set, AIMS, gui des the user towards developing efficient and scalable production-leve l parallel programs by locating performance improvement opportunities and determining optimization benefits. AIMS's Xisk helps identify pote ntial optimizations by computing various pre-defined normalized perfor mance indices from program traces. Inspection of these index point to specific optimizations that may benefit program performance. After ide ntifying and characterizing performance problems, AIMS's MK can provid e quantitative estimates of performance benefits to help the user avoi d arduous optimizations that may not lead to expected performance impr ovements by. MK also helps identify potential pitfalls or benefits of changing any of various system parameters. Based on MK's performance p rojection, an informed decision regarding the most beneficial program optimizations or upgrades in execution environments can be chosen.