MPI Island-based GPU Implementation of GA Running the Knapsack Benchmark

MPI Island-based GPU Implementation of GA Running the Knapsack Benchmark

Software

This package contains an efficient GPU implementation of the island based Genetic Algorithm. This GA solves the knapsack problem without any restrictions on the problem size (tested up to 100k items). The implementation is written in C++ utilising MPI and CUDA library. It is possible to be run on clusters of workstations or servers equipped with one or more GPUs as well as on a single multi-GPU machine. The performance of the implementation, the quality of produced results as well as the idea of implementation is explained in Jaros, J.: Multi-GPU Island-Based Genetic Algorithm Solving the Knapsack Problem, In: 2012 IEEE World Congress on Computational Intelligence, CA, US, IEEE, 2012, s. 217-224, ISBN 978-1-4673-1508-1

This package contains an efficient GPU implementation of the island based Genetic Algorithm. This GA solves the knapsack problem without any restrictions on the problem size (tested up to 100k items). The implementation is written in C++ utilising MPI and CUDA library. It is possible to be run on clusters of workstations or servers equipped with one or more GPUs as well as on a single multi-GPU machine. The performance of the implementation, the quality of produced results as well as the idea of implementation is explained in Jaros, J.: Multi-GPU Island-Based Genetic Algorithm Solving the Knapsack Problem, In: 2012 IEEE World Congress on Computational Intelligence, CA, US, IEEE, 2012, s. 217-224, ISBN 978-1-4673-1508-1

Genetic Algorithm, Island model, Knapsack, MPI, GPU, CUDA, SSE instructions

Genetic Algorithm, Island model, Knapsack, MPI, GPU, CUDA, SSE instructions

viz níže

K využití výsledku jiným subjektem je vždy nutné nabytí licence

https://www.fit.vut.cz/research/product/253/

Source codes