MOTIVATION

Need for a detailed comparison of algorithms on general purpose computers

Most previous benchmarks consider very few criteria

Need to resolve performance differences between many algorithms developed over the years

Correlation between implementation and theoretical performance of algorithms

Most previous benchmarks on special machines - DSPs and Supercomputers with highly optimized code

Comprehensive collection of state-of-the-art FFT algorithms under a common framework

Work funded by DARPA through U. S. Air Force's Rome Laboratories