GPUSS - GPU Stochastic Simulation for statistical data analysis


This page is aimed as a resource for those interested in using GPUs and other many-core devices for statistical data analysis and statistical computation.

GPUs are attractive for certain types of scientific computation as they offer the potential speed-up of multi-processor devices with the added advantages of being low cost, low maintenance, energy efficient, dedicated local devices that are easy to program.

NVIDIA is a manufacturer of graphics cards which can be used for scientific computing using their CUDA environment.

Below we provide links to papers in statistics and related fields which make use of GPU technology.


Background on GPU Computation


Tutorials and CUDA Information


Overview Papers and Books


Links to application papers and methods


Coded Examples


Acknowledgements

Research for this page has been supported by the Oxford-Man Institute of Quantitative Finance.


Contact

Please email us with suggestions on how to improve the site, or with relevant links and papers to add.

lee@stats.ox.ac.uk or c.holmes@stats.ox.ac.uk