SciComp Accelerates Market Leading Derivative Pricing Software with NVIDIA CUDA

SciComp, an Austin, Texas-based company, has a high-tech derivatives software solution to shorten the development time and accelerate the performance of Monte Carlo pricing models. The company has enhanced SciFinance, its flagship product, to deliver accurate NVIDIA CUDA-enabled derivatives pricing models that run up to 100 times faster than serial code. More significantly, this speed up can be achieved without any additional work or hand programming which, in a market where a slight delay or inaccuracy can end up costing millions, is a critical advance.

The key to this speedup is reliance on the graphics processing unit (GPU) for the calculations. A GPU is a many-core (up to 240 cores in the latest models) parallel processor that can run parallel applications many times faster than a computer's CPU. This massive parallel computational power is unlocked by NVIDIA CUDA architecture, a programming environment based on the industry-standard C language that enables developers to write software to solve complex computational problems in a fraction of the time.

"The old aphorism 'time is money' has never been more true than in the pressurized world of OTC derivatives trading, where the constant flow of new contracts demands the ability to produce complex mathematical pricing models rapidly," said Curt Randall, executive vice president of SciComp, in a press release. "This process used to take days if not weeks of error prone hand coding, but with SciFinance, model developers make a few changes to a model specification of a half page or less and then generate accurate C or C++ source code pricing models in minutes," he said.

"SciFinance's new ability to generate GPU-enabled code without additional programming is game-changing technology for our customers," added Randall. "The code takes full advantage of the GPU's parallel architecture, delivering an immediate 20-100X execution speed increase. Pricing models that used to run in minutes now complete in seconds, allowing financial institutions to test alternatives models, increase scenario analysis, and better understand their potential risk exposure. And best of all, they need not become experts in parallel coding concepts. SciFinance takes care of that," he said.