Manipulating Visualization, Not Codes

摘要: Manual program parallelization and optimization may be necessary to reach a decent portion of the target architecture's peak performance when automatic tools fail at choosing best strategy. While broad range languages libraries provide convenient ways express parallelism, difficult, time consuming error-prone parallelism identification extraction task is mostly left under programmer's responsibility. To address this issue, we introduce visualization-based approach ease expression that leverages polyhedral compilation technologies. Our interactive tool, Clint, maps direct manipulation visual representation transformations with real-time semantics preservation feedback. We conducted two user studies showing Clint's visualization can accurately understood by both experts non-expert programmers, extracted better from than source code in many cases.