Section: New Results

OpenMP Loop Scheduling

Participants : Pedro Penna [Master, PUC Minas, UFSC] , Marcio Castro [Professor, UFSC] , Henrique Cota de Freitas [Professor, PUC Minas] , Francois Broquedis, Jean François Méhaut.

In High Performance Computing, the application’s workload must be well balanced among the threads to achieve better performance. In this work, we propose a methodology that enables the design and exploration of new loop scheduling strategies. In this methodology, a simulator is used to evaluate the most relevant existing scheduling strategies, and a genetic algorithm is employed to explore the solution space of the problem itself. The proposed methodology allowed us to design a new loop scheduling strategy, which showed to be up to 32.3x better than the existing policies in terms of load balance.