The NVISION ’08 visual computing conference, taking place Aug. 25-27 in San Jose, Calif., will include a CUDA Developer Conference for software developers interested in accelerating their applications using the NVIDIA CUDA software development environment. Sessions are designed for both experienced CUDA programmers as well as those new to the technology, which is based on the C programming language and is becoming a standard in the computing market.
One of the highlights of the CUDA Conference will be a session by William Dorland of the University of Maryland. The program will focus on his work in porting simulations of black hole dynamics, plasma turbulence and n-body dynamics to clusters of NVIDIA GPUs using CUDA technology. Modest-size codes of a few thousand lines were ported to the GPU hardware in less than a day with speedups of 25 to 30 times that achieved with traditional architectures. By moving to clusters of high-performance GPUs, Dorland expects to be able to produce results on an inexpensive, many-teraflop local cluster.
Also planned is a panel discussion on the future of parallel and GPU computing. The chat will be moderated by Dr. David Kirk, chief scientist at NVIDIA, and will feature Dr. Wen-mei Hwu of the University of Illinois at Urbana-Champaign, Satoshi Matsuoka of the Tokyo Institute of Technology and Kathy Yelick of the Lawrence Berkeley National Laboratory.
Other highlights include a session by John Stone of the University of Illinois at Urbana-Champaign on accelerating computational biology programs up to 100 times using CUDA technology; a roundtable discussion on how to teach parallel programming using CUDA technology, led by Dr. Wen-mei Hwu of the University of Illinois at Urbana-Champaign; and a session presented by NVIDIA’s Joseph Stam on next-generation computer vision using CUDA technology.
The CUDA Developer Conference will also include workshops on CUDA development tools and libraries, basic and advanced training on how to develop applications using CUDA technology, and demonstrations and discussions of the performance gains and cost savings achieved. Participants in these sessions include leading industry and academic experts in the fields of molecular dynamics and computational chemistry; computational finance and quantitative risk analysis; geophysical and seismic processing; video, imaging, and computer vision; and astrophysics and astronomy.
Registration for the NVISION 08 conference is now open. Complete information on the CUDA Developer Conference, including how to register, can be found at www.nvision2008.com/Professionals/computing-developer.cfm. For more information on NVIDIA CUDA technology, see www.nvidia.com/cuda.
