Section: New Results

Advances in Parallel Metaheuristics on GPU

Participants : Thé Van Luong, Nouredine Melab, El-Ghazali Talbi.

Nowadays, GPU computing has recently been revealed effective to deal with time-intensive problems. This new emerging technology is believed to be extremely useful to speed up many complex algorithms. One of the major issues for metaheuristics is to rethink existing parallel models and programming paradigms to allow their deployment on GPU accelerators. Generally speaking, the major issues we have to deal with are: the distribution of data processing between CPU and GPU, the thread synchronization, the optimization of data transfer between the different memories, the memory capacity constraints, etc. The contribution of our work is to deal with such issues for the redesign of parallel models of metaheuristics to allow solving of large scale optimization problems on GPU architectures. Our objective is to rethink the existing parallel models and to enable their deployment on GPUs.

Thereby, the new results involve a new generic guideline for building efficient parallel metaheuristics on GPU (e.g. tabu search, iterated local search, island model for evolutionary algorithms, pareto local search or multi-start algorithms). Our challenge is to come out with the GPU-based design of the whole hierarchy of parallel models. In this purpose, very efficient approaches are proposed for CPU-GPU data transfer optimization, thread control, mapping of solutions to GPU threads or memory management. These approaches have been exhaustively experimented using eleven optimization problems and six GPU configurations. Compared to a CPU-based execution, experiments report up to 80-fold acceleration for large combinatorial problems and up to 2000-fold speed-up for a continuous problem. The different works related to our work have been accepted in a dozen of publications, including the IEEE Transactions on Computers journal.