An FPGA-based parallel processor for Black-Scholes option pricing using finite differences schemes

Financial engineering is a very active research field as a result of the growth of the derivative markets and the complexity of the mathematical models utilized in pricing the numerous financial products. In this paper, we present an FPGA-based parallel processor optimized for solving the Black-Scholes partial derivative equation utilized in option pricing which employs the two most widely used finite difference schemes: Crank-Nicholson and explicit differences. As our measurements demonstrate, the presented architecture is expandable and the speedup triggered is increased almost linearly with the available silicon resources. Although the processor is optimized for this specific application, it is highly programmable and thus it can significantly accelerate all applications that use finite differences computations. Performance measurements show that our FPGA prototype triggers a 5× speedup when compared with a 2GHz dual-core Intel CPU (Core2Duo). Moreover, for the explicit scheme, our FPGA processor provides an 8× speedup over the same Intel processor.