SPRACN4 August   2019 66AK2G12 , 66AK2H06 , 66AK2H12 , 66AK2H14 , OMAP-L132 , OMAP-L138 , TMS320C6452 , TMS320C6454 , TMS320C6455 , TMS320C6457 , TMS320C6652 , TMS320C6654 , TMS320C6655 , TMS320C6657 , TMS320C6672 , TMS320C6674 , TMS320C6678 , TMS320C6742 , TMS320C6743 , TMS320C6745 , TMS320C6746 , TMS320C6747 , TMS320C6748

 

  1.   Using DSPLIB FFT Implementation for Real Input and Without Data Scaling
    1.     Trademarks
    2. 1 Real Input Introduction
      1. 1.1 Prerequisites
      2. 1.2 Computing a Length N/2 Complex FFT From a Length N Real Input Sequence
      3. 1.3 Returning to a Length N Real Sequence Using a Length N/2 Complex IFFT
      4. 1.4 Benchmark of the Efficient Compute of FFT
    3. 2 Fixed Point FFT With No Data Scaling
      1. 2.1 Suggested Change
      2. 2.2 Example Application
    4. 3 Summary
    5. 4 References

1.4 Benchmark of the Efficient Compute of FFT

Table 1 compares the performance of two methods of calculating the FFT of an N-point real sequence. Complex FFT refers to using an N-point complex FFT in the standard way, while Real FFT refers to using an N/2-point complex FFT as described in this topic. As expected, the Real FFT method yields superior performance.

Table 1. Performance Comparison

N Cycle Count
Complex FFT Real FFT Improvement
128 1134 827 27.07%
256 2094 1709 18.38%
512 4944 3181 35.65%
1024 9680 7055 27.11%
2048 22770 13839 39.22%
4096 45298 31025 31.50%
8192 104724 61745 41.04%

NOTE

  • The data was measured on C6748 DSP with the data in L2 SRAM.
  • L1D and L1P cache was enabled.
  • Compiler CGT 6.x and CGT 7.4.x was used due to need to support legacy COFF binary format.