25 December 2011 11:37:59 AM
FFT_SERIAL
C++ version
Demonstrate an implementation of the Fast Fourier Transform
of a complex data vector.
Accuracy check:
FFT ( FFT ( X(1:N) ) ) == N * X(1:N)
N NITS Error Time Time/Call MFLOPS
2 10000 7.85908e-17 0.001346 6.73e-08 148.588
4 10000 1.20984e-16 0.003509 1.7545e-07 227.985
8 10000 6.8208e-17 0.007014 3.507e-07 342.173
16 10000 1.43867e-16 0.018269 9.1345e-07 350.32
32 1000 1.24137e-16 0.004512 2.256e-06 354.61
64 1000 1.86016e-16 0.009766 4.883e-06 393.201
128 1000 1.89545e-16 0.020337 1.01685e-05 440.576
256 1000 2.0702e-16 0.049377 2.46885e-05 414.768
512 100 1.88568e-16 0.010283 5.1415e-05 448.118
1024 100 2.22654e-16 0.023599 0.000117995 433.917
2048 100 2.50688e-16 0.049644 0.00024822 453.791
4096 100 2.58428e-16 0.112087 0.000560435 438.517
8192 10 2.57793e-16 0.023735 0.00118675 448.688
16384 10 2.79687e-16 0.052804 0.0026402 434.391
32768 10 2.95669e-16 0.109421 0.00547105 449.201
65536 10 2.87474e-16 0.239357 0.0119678 438.08
131072 1 3.17514e-16 0.049511 0.0247555 450.046
262144 1 3.20899e-16 0.10875 0.054375 433.894
524288 1 3.28006e-16 0.223635 0.111818 445.434
1048576 1 3.29961e-16 0.482866 0.241433 434.313
FFT_SERIAL:
Normal end of execution.
25 December 2011 11:38:02 AM