GENERAL CONCLUSIONS



  • Comprehensive benchmarks for contemporary FFT algorithms

  • Proved the strong influence of machine architecture on algorithm performance

  • Demonstrated the large influence of compilers

  • Shown that execution time closely related to the number of FLOPS

  • Created a framework to perform poly-algorithmic experimentation

  • Code and results available for public from out website at: http://www.cavs.msstate.edu/projects/speech/software/parallel_dsp/index.html