ALGORITHMIC PERFORMANCE



  • FHT is the best algorithm for real data

  • SRFFT fastest for complex data

  • Algorithms with multiple calls to real functions lose for complex signals (FHT and QFT)

  • SRFFT has the least number of computations

  • QFT is the worst w.r.t. memory usage

  • Literature neglects computations involved in decomposition into even and odd components

  • Recursive implementations perform the worst

  • DITF performs only marginally better than Radix-2

  • Literature neglects complexity in finding transition factors