Accelerating sequential computer vision algorithms using commodity parallel hardware
“The Free Lunch Is Over: A Fundamental Turn Toward Concurrency in Software” Sutter (2005)
Door: Jaap van de Loosdrecht
Kernwoorden: parallelization of sequential code, Multi-core CPU/GPU, OpenMP, OpenCL
From 2004 onwards the clock frequency of CPU’s has not increased significantly. The only way to get more performance is to go for parallel programming, which is not an easy way. This presentation is a result of a research master project at LIT in Ireland:
- CPU versus GPU architecture
- Overview standards for parallel programming
- Choice of standard for multi-core CPU programming (= OpenMP)
- Choice of standard for multi-core GPU programming (= OpenCL)
- Benchmark examples
- Future work
- Conclusions
The results of this work can be applied to other domains than Computer Vision.