The research analyzed the improvement in the efficiency of systems using Multi-Threaded Bacterial Iterated Greedy (MBIG) techniques, when the number of parallel processing threads is scaled.
The different variants of MBIG are recently proposed approaches for combining Iterated Greedy techniques, as state-of-the-art methods, with bacterial evolutionary algorithms based on a hybrid technique involving the Multi-Threaded Iterated Greedy heuristic and a Genetic Algorithm based memetic technique in order to efficiently solve the Permutation Flow Shop Problem on parallel computing architectures.
In the research the MBIG variants have been executed involving various number of processing threads in order to examine the scalability of the approaches. The simulation runs have been carried out on instances from the well-known Taillard’s benchmark problem set.
The scalability is then evaluated by comparing the results to each other and to the results given by the virtually parallelized implementation of the techniques.
A teljes kutatási beszámoló letölthető innen (PDF)
- Balázs Krisztián, doktorjelölt, BME VIK Távközlési és Médiainformatikai Tanszék, balazs@tmit.bme.hu
- Pusztai Pál, egyetemi adjunktus, Széchenyi István Egyetem, Matematika és Számítástudomány Tanszék, pusztai@sze.hu
- Horváth Zoltán, főiskolai tanár, Széchenyi István Egyetem, Matematika és Számítástudomány Tanszék, horvathz@sze.hu
- Kóczy T. László, egyetemi tanár, BME VIK Távközlési és Médiainformatikai Tanszék, koczy@tmit.bme.hu
2013. szeptember 8.