Scalable Tuning of (OpenMP) GPU Applications via Kernel Record and Replay
Konstantinos Parasyris, Giorgis Georgakoudis, Esteban Rangel, Ignacio Laguna, Johannes Doerfert
Abstract
HPC is a heterogeneous world in which host and device code are interleaved throughout the application. Given the significant performance advantage of accelerators, device code execution time is becoming the new bottleneck. Tuning the accelerated parts is consequently highly desirable but often impractical due to the large overall application runtime which includes unrelated host parts.
Topics & Concepts
Computer scienceBottleneckScalabilityHost (biology)Kernel (algebra)Parallel computingCode (set theory)SupercomputerOperating systemComputer architectureEmbedded systemProgramming languageCombinatoricsSet (abstract data type)BiologyEcologyMathematicsParallel Computing and Optimization TechniquesDistributed and Parallel Computing SystemsAdvanced Data Storage Technologies