by Andrew Humber

Last night saw the opening gala of SC09. Attendees streamed through the NVIDIA booth in the hundreds to get a sneak peak at the new Tesla 20-series products. Running n-body, simulating the motion of 20,000 particles and their gravitational interaction, the new Tesla C2050 showed more than a 6X improvement in double precision performance over a Tesla C1060 (previous generation). And this is still a pre-production product– expect this number to get bigger as the Tesla products are readied to hit the market in Q2 of next year.

One of the key themes of SC09 is 3D Internet and NVIDIA’s booth was THE place to see it in the flesh. NVIDIA RealityServer combines NVIDIA Tesla GPUs and 3D web services software to deliver interactive, photorealistic images over the internet to any platform, even an iPhone – an incredibly powerful tool for product designers, architects and consumers. I fear that if online shopping gets this real, then my wife may never leave the house again!
Poser 3
NVIDIA also gave a group of computational researchers, who are using CUDA-based GPUs in their work, an opportunity to display their research posters for press and other show attendees.  Every day we see more and more examples of the amazing work being done on the CUDA architecture – CUDA Zone now has over 650 postings from researchers in the form of codes, papers and applications. The impact that GPU Computing is having in this space is incredible and a lot of the excitement behind it was felt at the NVIDIA booth this evening. We’d like to thank all of the posters that participated – keep up the awesome work!

So the first 3 hours of the show were busy…really busy. Today looks set to be a pretty exciting day as well. The on-booth theater kicks off with presentations from prominent HPC figures including Prof. Satoshi Matsuoka from the Tokyo Institute of Technology talking about the path to exascale computing and Prof. Ross Walker, from UC San Diego working on the AMBER molecular dynamics software. We’re also unveiling details of a slew of new updates to all of the tools, languages and APIs that are supported by the CUDA architecture as well as having NVIDIA engineers presenting and on hand to discuss these topics as well.