In a world first, Exegy set up the low latency Exegy Ticker Plant to use FPGA, GPU and CPU technology side by side to solve real world financial risk calculations extremely efficiently. Calculations that might normally take an hour can now be done in a little more than 10 seconds.

"We believe we're the first to use these diverse platforms simultaneously to solve this important problem," said Jeff Wells, Vice President of Product Management at Exegy. "Whereas financial institutions formerly had to depend on the all purpose CPU for most computational functions, now there are hybrid solutions that integrate diverse platforms to crunch enormous amounts of data very rapidly. Massive parallel processing can be done economically in one computer chassis."

Exegy and Washington University St Louis submitted a paper to SuperComputing 2008 describing the benchmark studies. The paper has been welcomed by expert reviewers. One reviewer commented "I believe that this model represents what the future holds in terms of utilizing the most efficient silicon".

The massively parallel architecture of NVIDIA Tesla GPUs enables the processing of thousands of floating point calculations simultaneously, making it an extremely powerful platform for computational finance algorithms such as Monte Carlo. Andy Keane, general manager of the Tesla business at NVIDIA said: "At the core of the Tesla GPU is our CUDA parallel architecture. This architecture is accessible through an industry standard C language programming environment that allows developers and researchers to tap into the GPU more quickly and easily than any other solution shipping today. The maturity of this architecture and the tools that we provide have enabled a wide audience to leverage it with tremendous success. Companies like Exegy are proof of this and we're delighted to see the results they are getting with Monte Carlo simulations."

"Our customers, including a number of international banks who are trading in extremely busy stock, commodities and futures markets, are able to see prices faster and before others, enabling them to make smarter trading decisions," says Wells. "Naturally they also need to be able to assess risk faster than ever before and this is how to do it."

One of the co-authors of the report, Dr Roger Chamberlain, said "the most impressive results showed a 127x speedup relative to an 8 processor implementation or a 310x performance improvement over a single 2.2 GHz CPU."

John Barr, Research Director of Financial Markets at The 451 Group and Chair of the British Computer Society's Parallel Processing Specialist Group, said "The stalled CPU clock in commodity processors opens the door to acceleration technologies such as FPGA and GPU, ideally in an appliance model for ease of use. And, as highlighted in our recent Financial Messaging report, the performance gains mean that a single Exegy appliance can replace a handful or more servers - with significant power, cooling and space savings."

Kevin McPartland from Tabb Group pointed out that "the computing power needed to achieve such accuracy and speed is immense; therefore hardware acceleration can provide a major edge towards solving this tough problem."

McPartland added, "As portfolios become more complex and data volumes grow, the calculations required to understand a portfolio's risk profile cannot be carried out quickly enough to be effective using legacy technology. When properly architected, the use of new hardware accelerated solutions in conjunction with sophisticated software models provide investors with the fastest and most accurate information, ultimately increasing real-time market transparency."

This latest announcement from Exegy comes on the heels of the company's unveiling of MarketDataPeaks.com , the first public website that provides a minute-by-minute account of the aggregated volume of market data messages across major North American exchanges. The site, sponsored by Exegy, Xasax and the Financial Information Forum, uses the Exegy Ticker Plant to process and update the aggregated market data.