Bachelor and Master Theses

Title: Re-engineering sequential software to introduce parallelization
Level: Advanced
Description: Today, many systems have at their disposal multi-core processing units, but run sequential software; this leads to an under-utilization of the available hardware. Therefore, in order to increase the utilization of the available hardware the software should be modified so to introduce parallel computation, thus increasing its performance. However, due to the complex nature of parallel software in comparison to sequential software, there is a risk that the effort required to modify the software becomes high. While refactoring software comes with its own challenges, this study focuses on the process of introducing parallelization into existing sequential software. Therefore, this study tackles the following research challenges:
- Location of parallelization potential
- Utilization of both multi-core CPU and GPU when introducing parallelization

A limiting factor of the potential speedup through parallelization is that not all computations cannot run in parallel due to dependencies on previous computations. This introduces the first challenge, which is to locate parallel potential within the existing software. With the access to both multi-core CPUs and GPUs, it is important to maximize the utilization of the hardware in order to allow the software to reach its full potential. Due to the differences of the two platforms, on the hardware level, they are suitable for different problems and choosing the most suitable one can therefore yield more optimal performance. Choosing which processing unit type is most suitable for a specific parallel software piece is the second research challenge of this study. The contribution of the study is a generic parallelization method for re-engineering existing sequential software. The method is meant to provide a structured and well-defined way to port legacy software from sequential to parallel platforms.
Company: ABB Robotics, kontaktperson: Johnny Holmberg
Prel. end date: 2017-09-26
Presentation date: 2017-09-22
Student: Andreas Granholm
IDT supervisor: Federico Ciccozzi, +46 21 151762
Examinator: Antonio Cicchetti
Antonio Cicchetti, +46-21-151762

Rapport och bilagor


Senaste uppdatering



2017-09-26, 14:32

  • Mälardalen University |
  • Box 883 |
  • 721 23 Västerås/Eskilstuna |
  • 021-101300, 016-153600 |
  • webmaster |
  • Latest update: 2018.05.24