UM0304

Complex 16-bit radix-4 FFT

17/19

4.7

Fixed-point error benchmarking

Table 8 shows fixed-point error analysis of FFT function (ARM966E-S) with respect to

floating-point and fixed-point C function, and amplitude, for a 16-point FFT.

Table 8.

The FFT function fixed-point error benchmarking

Real Part

Imaginary Part

Amplitude

ARM966E-S

Floating point Fixed-point C

ARM966E-S

Floating point Fixed-point C

-5552

-5549.4375

-5550

-5552

-5549.4375

-5550

7848.089844

21318

21320.33984

21319

-5841

-5840.29541

-5840

22105.78906

6297

6299.319336

6298

-1503

-1501.481567

-1502

6475.791016

4100

4101.071289

4100

-402

-401.039703

-401

4120.633301

3146

3146.563477

3146

208

208.557663

208

3153.467529

2573

2574.675781

2574

632

631.753662

631

2651.050293

2165

2166.108887

2165

967

967.559753

967

2372.382568

1836

1836.466064

1836

1262

1262.519897

1262

2228.578857

1543

1544.313599

1544

1543

1544.311401

1544

2183.987549

1261

1262.530396

1262

1836

1836.464233

1836

2228.583496

966

967.565186

967

2164

2166.102539

2165

2372.37915

631

631.76416

631

2574

2574.674316

2574

2651.051514

208

208.583069

208

3146

3146.570068

3146

3153.47583

-402

-401.032806

-401

4101

4101.062988

4100

4120.624512

-1501

-1501.468384

-1502

6298

6299.319824

6298

6475.788574

-5840

-5840.282227

-5840

21320

21320.34766

21319

22105.79297