void fft_radix4_I( int *fr, int *fi, int ldn)
{
const int n = (1UL<<ldn);
int ldm = 0, rdx = 2;
for (int i0 = 0; i0 < n; i0 += 4)
{
int xr,yr,ur,vr, xi,yi,ui,vi;
int i1 = i0 + 1;
int i2 = i1 + 1;
int i3 = i2 + 1;
sum_dif_I(fr[i0], fr[i1], xr, ur);
sum_dif_I(fr[i2], fr[i3], yr, vi);
sum_dif_I(fi[i0], fi[i1], xi, ui);
sum_dif_I(fi[i3], fi[i2], yi, vr);
sum_dif_I(ui, vi, fi[i1], fi[i3]);
sum_dif_I(xi, yi, fi[i0], fi[i2]);
sum_dif_I(ur, vr, fr[i1], fr[i3]);
sum_dif_I(xr, yr, fr[i0], fr[i2]);
}
for (ldm = 2 * rdx; ldm <= ldn; ldm += rdx)
{
int m = (1UL<<ldm);
int m4 = (m>>rdx);
int phI0 = NWAVE / m;
int phI = 0;
for (int j = 0; j < m4; j++)
{
int c,s,c2,s2,c3,s3;
s = Sinewave[ phI];
s2 = Sinewave[ 2*phI];
s3 = Sinewave[ 3*phI];
c = Sinewave[ phI + NWAVE/4];
c2 = Sinewave[ 2*phI + NWAVE/4];
c3 = Sinewave[ 3*phI + NWAVE/4];
for (int r = 0; r < n; r += m)
{
int i0 = j + r;
int i1 = i0 + m4;
int i2 = i1 + m4;
int i3 = i2 + m4;
int xr,yr,ur,vr, xi,yi,ui,vi;
mult_shf_I( c2, s2, fr[i1], fi[i1], xr, xi);
mult_shf_I( c, s, fr[i2], fi[i2], yr, vr);
mult_shf_I( c3, s3, fr[i3], fi[i3], vi, yi);
int t = yi - vr;
yi += vr;
vr = t;
ur = fr[i0] - xr;
xr += fr[i0];
sum_dif_I(ur, vr, fr[i1], fr[i3]);
t = yr - vi;
yr += vi;
vi = t;
ui = fi[i0] - xi;
xi += fi[i0];
sum_dif_I(ui, vi, fi[i1], fi[i3]);
sum_dif_I(xr, yr, fr[i0], fr[i2]);
sum_dif_I(xi, yi, fi[i0], fi[i2]);
}
phI += phI0;
}
}
}
void fft_radix4_I(int *fr, int *fi, int ldn) {
const int n = (1UL<<ldn);
int ldm = 0, rdx = 2;
ldm = (ldn&1);
if ( ldm!=0 ) {
for (int i0=0; i0<n; i0+=8) {
fft8_dit_core_p1(fr+i0, fi+i0);
}
} else {
for (int i0 = 0; i0 < n; i0 += 4) {
int xr,yr,ur,vr, xi,yi,ui,vi;
int i1 = i0 + 1;
int i2 = i1 + 1;
int i3 = i2 + 1;
sum_dif_I(fr[i0], fr[i1], &xr, &ur);
sum_dif_I(fr[i2], fr[i3], &yr, &vi);
sum_dif_I(fi[i0], fi[i1], &xi, &ui);
sum_dif_I(fi[i3], fi[i2], &yi, &vr);
sum_dif_I(ui, vi, &fi[i1], &fi[i3]);
sum_dif_I(xi, yi, &fi[i0], &fi[i2]);
sum_dif_I(ur, vr, &fr[i1], &fr[i3]);
sum_dif_I(xr, yr, &fr[i0], &fr[i2]);
}
}
for (ldm += 2 * rdx; ldm <= ldn; ldm += rdx) {
int m = (1UL<<ldm);
int m4 = (m>>rdx);
int phI0 = NWAVE / m;
int phI = 0;
for (int j = 0; j < m4; j++) {
int c,s,c2,s2,c3,s3;
s = pgm_read_word(&Sinewave[ phI]);
s2 = pgm_read_word(&Sinewave[ 2*phI]);
s3 = pgm_read_word(&Sinewave[ 3*phI]);
c = pgm_read_word(&Sinewave[ phI + NWAVE/4]);
c2 = pgm_read_word(&Sinewave[ 2*phI + NWAVE/4]);
c3 = pgm_read_word(&Sinewave[ 3*phI + NWAVE/4]);
for (int r = 0; r < n; r += m) {
int i0 = j + r;
int i1 = i0 + m4;
int i2 = i1 + m4;
int i3 = i2 + m4;
int xr,yr,ur,vr,xi,yi,ui,vi;
mult_shf_I( c2, s2, fr[i1], fi[i1], &xr, &xi);
mult_shf_I( c, s, fr[i2], fi[i2], &yr, &vr);
mult_shf_I( c3, s3, fr[i3], fi[i3], &vi, &yi);
int t = yi - vr;
yi += vr;
vr = t;
ur = fr[i0] - xr;
xr += fr[i0];
sum_dif_I(ur, vr, &fr[i1], &fr[i3]);
t = yr - vi;
yr += vi;
vi = t;
ui = fi[i0] - xi;
xi += fi[i0];
sum_dif_I(ui, vi, &fi[i1], &fi[i3]);
sum_dif_I(xr, yr, &fr[i0], &fr[i2]);
sum_dif_I(xi, yi, &fi[i0], &fi[i2]);
}
phI += phI0;
}
}
}
