Function Based Benchmarks to Abstract Parallel Hardware and Predict Efficient Code Partitioning

To increase the performance of a program, developers have to parallelize their code due to trends in modern hardware development. Since the parallelization of source code is paired with additional programming effort, it is desirable to know if a parallelization would result in an advantage in performance before implementing it. This paper examines the use of benchmarks for estimating the performance gain looking at the parallelization of Population Based Algorithms (PBAs) like Genetic Algorithms (GAs) and Particle Swarm Optimization Algorithms (PSOs) to be implemented on multi- and many-cores. These benchmarks are named function based benchmarks due to their dependence on the PBAs' functions. Furthermore, the software-hardware mapping with the most performance gain is suggested.