Abstract | ||
---|---|---|
Graphics processing units (GPUs) in recent years have evolved to become powerful, programmable vector processing units. Furthermore, the maximum processing power of current generation GPUs is roughly four times that of current generation CPUs (central processing units), and that power is doubling approximately every nine months, about twice the rate of Moore's law. This research examines the GPU's advantage at performing convolutionbased image processing tasks compared to the CPU. Straight-forward 2D convolutions show up to a 130:1 speedup on the GPU over the CPU, with an average speedup in our tests of 59:1. Over convolutions performed with the highly optimized FFTW routines on the CPU, the GPU showed an average speedup of 18:1 for filter kernel sizes from 3x3 to 29x29. |
Year | DOI | Venue |
---|---|---|
2005 | 10.1007/11428848_32 | International Conference on Computational Science (2) |
Keywords | Field | DocType |
average speedup,fftw routine,current generation gpus,current generation cpus,convolutionbased image processing task,maximum processing power,recent year,programmable vector processing unit,central processing unit,filter kernel size,image processing | Graphics,Central processing unit,Computer science,Parallel computing,Image processing,Fast Fourier transform,Kernel adaptive filter,Digital image processing,Vector processor,Speedup | Conference |
Volume | ISSN | ISBN |
3515 | 0302-9743 | 3-540-26043-9 |
Citations | PageRank | References |
4 | 0.71 | 4 |
Authors | ||
5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Bryson R. Payne | 1 | 23 | 5.93 |
Saeid O. Belkasim | 2 | 4 | 0.71 |
G. Scott Owen | 3 | 4 | 0.71 |
Michael Weeks | 4 | 130 | 16.29 |
Ying Zhu | 5 | 27 | 8.27 |