void fft8_dit_core_p1(int *fr, int *fi) {
int plus1a, plus2a, plus3a, plus4a, plus1b, plus2b;
int mins1a, mins2a, mins3a, mins4a, mins1b, mins2b, mM1a, mM2a;
sum_dif_I(fr[0], fr[1], &plus1a, &mins1a);
sum_dif_I(fr[2], fr[3], &plus2a, &mins2a);
sum_dif_I(fr[4], fr[5], &plus3a, &mins3a);
sum_dif_I(fr[6], fr[7], &plus4a, &mins4a);
sum_dif_I(plus1a, plus2a, &plus1b, &mins1b);
sum_dif_I(plus3a, plus4a, &plus2b, &mins2b);
sum_dif_I(plus1b, plus2b, &fr[0], &fr[4]);
sum_dif_I(mins3a, mins4a, &mM1a, &mM2a);
int prib1a, prib2a, prib3a, prib4a, prib1b, prib2b;
int otnt1a, otnt2a, otnt3a, otnt4a, otnt1b, otnt2b, oT1a, oT2a;
sum_dif_I(fi[0], fi[1], &prib1a, &otnt1a);
sum_dif_I(fi[2], fi[3], &prib2a, &otnt2a);
sum_dif_I(fi[4], fi[5], &prib3a, &otnt3a);
sum_dif_I(fi[6], fi[7], &prib4a, &otnt4a);
sum_dif_I(prib1a, prib2a, &prib1b, &otnt1b);
sum_dif_I(prib3a, prib4a, &prib2b, &otnt2b);
sum_dif_I(prib1b, prib2b, &fi[0], &fi[4]);
sum_dif_I(otnt3a, otnt4a, &oT1a, &oT2a);
mM2a = mult_shf_s16x16(mM2a, 181);
sum_dif_I(mins1a, mM2a, &plus1a, &plus2a);
prib2b = mult_shf_s16x16(oT1a, 181);
sum_dif_I(otnt2a, prib2b, &mins3a, &plus3a);
sum_dif_I(plus1a, mins3a, &fr[7], &fr[1]);
sum_dif_I(mins1b, otnt2b, &fr[6], &fr[2]);
sum_dif_I(plus2a, plus3a, &fr[3], &fr[5]);
oT2a = mult_shf_s16x16(oT2a, 181);
sum_dif_I( otnt1a, oT2a, &plus1a, &plus2a);
plus2b = mult_shf_s16x16(mM1a, 181);
sum_dif_I(-mins2a, plus2b, &plus3a, &mins3a);
sum_dif_I(plus1a, mins3a, &fi[7], &fi[1]);
sum_dif_I(otnt1b,-mins2b, &fi[6], &fi[2]);
sum_dif_I(plus2a, plus3a, &fi[3], &fi[5]);
}
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;
}
}
}