The Jacobi Method in Reconfigurable Hardware

—Linear equations provide useful tools for understanding the behavior of a wide variety of phenomena— from science and engineering to social sciences. A number of techniques have arisen to find the solution of these systems; examples are Jacobi, Gauss-Seidel, Successive Over Relaxation, and Multigrid. In this paper, we present an accelerated version of the Jacobi algorithm by implementing it on reconfigurable hardware devices- Field Programmable Gate Arrays such as Virtex II Pro, Altera Stratix and Spartan3L. The design presented is implemented using Handel-C, a hardware compiler. The implementation results obtained are compared with a software version results written in C++ and running on a general purpose processor. Final results illustrate that Jacobi on a reconfigurable hardware can outperform a software version of the same algorithm.

• Type: Conference
• Year: 2007
• Where: WCE
• Authors: Safaa J. Kasbah, Issam W. Damaj
• Keywords: Reconfigurable Hardware, Reconfigurable Hardware Devices, Successive Over Relaxation, WCE 2007, WCE 2008