Mapping of application program tasks onto a suite of heterogeneous machines requires the estimation of execution times of the tasks on these machines. In this paper, an efficient methodology for estimating the execution times of a given program on various machines available in an HP environment is presented. The methodology uses parametric code profiling and parametric analytical benchmarking techniques and incorporates the concept of an architecture-independent computation model to estimate the execution times
Published in:
Heterogeneous Computing Workshop, 1994., Proceedings
Date of Conference: 26 Apr 1994