CIGPU-2012 Australia, 10-15 June 2012

cigpu 2008 logo CIGPU 2008

WCCI-2008 Special Session
Computational Intelligence on Consumer Games and Graphics Hardware CIGPU-2008.

Location Hong Kong Convention and Exhibition Centre, 5 June 2008
WCCI 2008 Program

Presentations and Discussions

Chair: W. B. Langdon
Co-Chair: Simon Harding

W. B. Langdon A Fast High Quality Pseudo Random Number Generator for Graphics Processing Units doi slides pic
Garnett Wilson Linear Genetic Programming GPGPU on Microsoft's Xbox 360 doi slides pic
Wai-Man Pang Generating Massive High-quality Random Numbers Using GPU doi slides pic
Raghavendra D. Prabhu SOMGPU: An unsupervised pattern classifier on Graphical Processing Unit doi slides pic
Discussion: Practical Aspects of Running Evolution on GPU
Simon Harding
Tien-Tsin Wong Shader Programming vs CUDA slides
Audience

Discussion

Suggested topics:

  • Strengths and weaknesses of current GPGPU hardware+software tools.
    Ease of use
    debuggging tools
    performance monitoring tools
  • Common pitfals to avoid
  • Good tools. Esp debugging and performance.
  • Languages
  • What else should computer scientist/enginneers keen on artificial intelligence but new to GPGPU know?
  • Documents? Books? WWW pages?
  • etc etc
Simon Harding Evolution of Image Filters on Graphics Processor Units Using Cartesian Genetic Programming doi slides pic
Robert Luke Speedup of Fuzzy Logic through Stream Processing on Graphics Processing Units doi pic
W. B. Langdon Evolving Genechip Correlation Predictors on Parallel Graphics Hardware doi slides pic
Shinji Fukui GPU Based Extraction of Moving Objects without Shadows under Intensity Changes doi
Discussion: Future of CIGPU
W. B. Langdon
Simon Harding
Audience

Discussion
Social Event
Man Leung Wong Buffet

(Photographs by Simon Harding)
CIGPU 2008 and other papers

Report in SIGEvolution

SIGEvolution, 3(1):19-21, 2008

History

Submissions were invited in the following areas

W. B. Langdon, wlangdon@ Departments of Biological and Mathematical Sciences, University of Essex essex.ac.uk
Simon Harding, simon.harding@ Computer Science Department Memorial University of Newfoundland cs.mun.ca
Man Leung Wong, mlwong@ Department of Computing & Decision Sciences, Lingnan University, Hong Kong ln.edu.hk

Authors of papers on research or applications in any area of computational intelligence (be it neural networks, fuzzy system, evolutionary, genetic, etc.) running on graphics hardware or other mass market electronic hardware have been invited to submit their work to CIGPU-2008. CIGPU will be held as part of the IEEE world congress on computational intelligence in Hong Kong 1-6 June 2008.

Due to its speed, price and availability, there is increasing interest in using mass consumer market commodity hardware for engineering and scientific applications. To date most of this interest has concentrated upon graphics hardware, particularly GPUs. However there is increasing interest in using games consoles such as Microsoft's Xbox, Playstation and the Cell processor, for research and applications. Indeed nVidia have recently announced Tesla as a scientificated version of their newest GPUs. In future personal computer physics engines, which are intended to provide realistic real time simulation of multi-body physics for sophisticated games, may also be adapted to serve science (rather than simulated it).
    Papers submitted for special sessions will be peer-reviewed with the same criteria used for the contributed papers.

It is clear that, whilst there is increasing interest, at present computational intelligence is under represented. As with other GPGPU applications, the drivers are: locality, convenience, cost and concentration of computer power. Indeed the principle manufactures (nVidia and ATI) claim faster than Moore's Law increase in performance. Suggesting that GPU floating point performance will continue to double every twelve months, rather than the 18-24 months observed for electronic circuits in general [Moo65] and personal computer CPUs in particular. Indeed the apparent failure of PC CPUs to keep up with Moore's law in the last few years makes GPU computing even more attractive. Even today's top of the range GPU greatly exceed the floating point performance of their hosts' CPU. This speed comes at a price.

Aims and Objectives

GPUs provide a very restricted type of parallel processing. PC cards like 8800 GTX or RV670 aught to give terraflop at your desk but how to actually get hold of it for non-trivial applications? One of the principle benefits of the CIGPU special session will be to disseminate knowledge of tools and techniques that are required to get the best from these widely available but difficult to fully exploit consumer hardware.

Dr. W.B. Langdon, Dr. Man Leung Wong and Dr. Simon Harding are internationally recognised experts at this.

The use of commodity graphics hardware for scientific computing (often referred to as General Purpose computing on Graphics Processing Units, GPGPU) is rapidly becoming established. Yet Computational Intelligence of all types has been slow to adopt GPGPUs. Drs. Wong, Harding and Langdon are at the forefront of use of GPUs in evolutionary computing. The special session will encourage take up of this technology in Fuzzy, Connectionist and other computational intelligence arenas.

At present there are relatively few (general purpose) GP GPU publications especially in computational intelligence, soft computing or genetic algorithms. However, previous experience with GPU areas, shows once started they take off very rapidly. The CI GPU area is about to explode and WCCI will be a very good launch site. Drs. Wong and Harding are very much the founders of the application of modern graphics hardware to evolutionary computing.

The Team

Dr. Wong has worked in computation intelligence in Hong Kong since he gained his doctorate. Recently he made several outstanding contributions to CI on GPUs. This year Dr. Harding presented the first use of GPUs in genetic programming at EuroGP (in Spain), followed by the first demonstration of artificial developmental systems on GPUs at HPCS'07 (Canada). Dr. Langdon recently demonstrated the simultaneous execution of many different programs on a high end nVidia graphics card. He has been a co-organiser of eight international conferences and workshops.

Significance of CIGPU Special Session

Interest in the emergent area of CIGPU is about to explode. Since CIGPU is a special session at WCCI-2008 papers presented at CIGPU will become the foundation stones which others will look to as the start of a new arena.
    As with other WCCI papers, papers accepted by CIGPU will be disseminated by the IEEE and become part of its electronic archive.

Other GPU sites

A few links to other pages interested in geneneral purpose computing on graphics hardware

Acknowledgement

We would like to thank


W.B.Langdon 27 Oct 2007 (last update 2 Nov 2013)