From video encoding to oil and gas exploration and from medical imaging to scientific research, thousands of CUDA developers in the high performance computing (HPC) community are leveraging a revolutionary GPU computing platform that was announced just one year ago. With over 60,000 downloads of the C-compiler to date, the combination of CUDA™ and Tesla® technologies have been the foundation for industry changing applications, making it the most rapidly adopted GPU computing technology platform in the HPC community.
“GPU Computing is coming at a time when we are running out of gas for time-critical forecasts on conventional clusters,” says National Center of Atmospheric Research’s John Michalakes, lead software developer for the Weather Research & Forecasting Model “We aim to cut the time for a forecast by at least a factor of two as we incorporate NVIDIA’s GPU computing technology into more of WRF. I expect the affect of accelerators in weather and climate modeling will be transformative.”
This year at the International Supercomputing Conference, NVIDIA Corporation, the leader in GPU technologies, has strengthened the CUDA and Tesla technology platforms with the introduction of its second-generation platform, the new Tesla 10 series computing solutions. Binary compatible and supporting the industry standard language of C, the new products enable developers to solve their computational challenges in a common and familiar development environment that scales effortlessly from one generation to the next with no re-coding required.
“The path to great discovery is filled with many challenges and for today's scientists, engineers and researchers, in fields such as drug research, seismic exploration or medical science, one of the biggest challenges is computation,” said Jen-Hsun Huang, president and CEO of NVIDIA. “The GPU is taking high performance computing in a fundamentally new direction. Now, hundreds of processor cores are able to work together to give scientists and engineers massive jumps in performance, while dramatically reducing the footprint in their datacenter.”
The new Tesla product family includes the Tesla S1070 1U computing system and the Tesla C1060 computing processor and delivers:
* Double the performance: up to 4 Teraflops per 1U system
* Double precision: IEEE 754 arithmetic support
* Double the memory: with 16 Gigabytes of memory per 1U system
* Up to 3x the power efficiency: for a more efficient computing environment
“As a world leader of derivatives, BNPParibas is constantly investing in research for more efficient financial calculations. To this end, we are currently conducting research on the use of GPUs to speed up options processing and early access to the new NVIDIA Tesla products has been highly beneficial. These experiments, done with ANEO consulting firm, delivered very promising results with impressive performance speedup and exceptional accuracy. NVIDIA Tesla computing products, which offers IEEE 754 double precision compliance, is also a very important step forward for us,” declares Stephane Tyc, BNPParibas Global Head of Equities & Derivates Quantitative Research.
When combined with the award-winning CUDA C-language development software for parallel computing, the new Tesla products extend the reach of GPUs to any computationally intensive applications requiring double precision accuracy. To date, over 70 million CUDA enabled GPUs have been sold into the market and over 60,000 downloads of the C-compiler have been recorded through the community Web site, CUDA Zone, which is located at http://www.nvidia.com/cuda . As a result, developers across a wide variety of fields including financial analysis, astrophysics and seismic imaging are leveraging NVIDIA’s CUDA development tools. These developers can now simply parallelize their software and exploit the GPU’s parallel computing architecture to automatically distribute computing work to hundreds of processor cores.
“CUDA has allowed us to tap into the processing power of the GPU with tremendous ease, saving us time and money,” says Jim Hardwick, Senior Software Engineer at Techniscan. “A single host system and two Tesla D870s are considerably cheaper than the 16-core cluster. This is significant not only for production and sales, but also reduces research and development costs in engineering.”
“Using only 12 GPUs, Volera is capable of analyzing the entire U.S. options market, in real time, with latencies of less than 10 milliseconds,” says Gerald A. Hanweck, Jr., founder and principal partner of Hanweck Associates. “That sort of result would usually require at least 60 traditional 1U servers. By using NVIDIA GPU technology, our clients are able to see better results while saving maintenance costs, with lower power consumption, and a smaller real-estate footprint.”
The Tesla S1070 1U computing system and Tesla C1060 computing processor board will be available for purchase for $7999 and $1699 respectively. These products are sampling today and will ship in August 2008.
For more information and for the free download of CUDA,please visit: http://www.nvidia.com/cuda and for more information on the Tesla 10 Series computing products, please visit http://www.nvidia.com/tesla
NVIDIA (Nasdaq: NVDA) is the world leader in visual computing technologies and the inventor of the GPU, a high-performance processor which generates breathtaking, interactive graphics on workstations, personal computers, game consoles, and mobile devices. NVIDIA serves the entertainment and consumer market with its GeForce® products, the professional design and visualization market with its Quadro® products, and the high-performance computing market with its Tesla™ products. NVIDIA is headquartered in Santa Clara, California, and has offices throughout Asia, Europe, and the Americas. NVIDIA's inaugural NVISION 08 conference will be held August 25-27, 2008 in San Jose, California. For more information, visit http://www.nvidia.com and http://www.nvision2008.com