Data centric techniques for mapping performance data to program variables

Traditional methods of performance analysis offer a code centric view, presenting performance data in terms of blocks of contiguous code (statement, basic block, loop, function). Data centric techniques, combined with hardware counter information, allow various program properties including cache misses and cycle count to be mapped directly to variables. We introduce mechanisms for efficiently collecting data centric performance numbers independent of hardware support. We create extended data centric mappings, which we call variable blame, that relates data centric information to high level data structures. Finally, we show performance data gathered from three parallel programs using our technique.

► Inclusive data centric profiling can be done by mapping samples to data flow.
► Data from this analysis are often unique compared to code centric analysis.
► Can map performance information to high level abstractions within the code.
► Betters program understanding with the potential of performance optimizations.