Super-sized Multiplies: How Do FPGAs Fare in Extended Digit Multipliers?

Extended digit multiplication can be an effective benchmark for comparing contemporary CPUs to other architectures and devices. The Great Internet Mersenne Prime Search (GIMPS), a distributed computing effort to find large prime numbers, has produced highly optimized code for multiplying large, multimillion digit numbers on Pentium processors. This paper presents a hardware large integer multiplier implemented in a Xilinx FPGA. The design, which utilizes an all-integer Fast Fourier Transform, is compared to a Pentium processor running the optimized GIMPS code. The results show that while a speed-up is achieved by the FPGA, the cost of the hardware likely outweighs any performance benefit.