void fft_2(Param n, Param p, Param N2, Param N1, char* ix, CplxSp* i0, CplxSp* i1, CplxSp* o0, CplxSp* o1){
#if VERBOSE
printf("Execute fft_2\n");
#endif
const int id = (*ix)*n;
const int m = (id % (N2*(1<<(p))));
const unsigned short incr = N1 / (1 << (p+1));
CplxSp const* restrict pd_in0 = i0;
CplxSp const* restrict pd_in1 = i1;
CplxSp const* restrict pd_twi = twi64k;
CplxSp * restrict pd_out0 = o0;
CplxSp * restrict pd_out1 = o1;
#pragma MUST_ITERATE(8, ,8)
for(int k=0; k<n; k++){
unsigned short r = (m+k)*incr;
__float2_t v_in0 = _amem8_f2_const(&pd_in0[k]);
__float2_t v_in1 = _amem8_f2_const(&pd_in1[k]);
/* Get the corresponding twiddle factor */
__float2_t v_twi = _amem8_f2_const(&pd_twi[r]);
/* Execute:
* out0 = in0 + in1*twi
* out1 = in0 - in1*twi
* with out0, out1, in0, in1 and twi complex floating numbers
* [even]: real part and [odd]: imag part
*/
__float2_t v_in1Twi = _complex_mpysp(v_in1, v_twi);
_amem8_f2(&pd_out0[k]) = _daddsp(v_in0, v_in1Twi);
_amem8_f2(&pd_out1[k]) = _dsubsp(v_in0, v_in1Twi);
}
}
void intrinsicC_filters(float *inputComplex, int N_E, float *output )
{
int filter_size, i, j;
__float2_t x[MAX_FILTER_SIZE];
__float2_t y,z;
__float2_t *p_in;
__float2_t sum;
float total_sum;
/*
* First filter, coefficients are (1/2 -1/2) (-1/2 1/2)
*/
p_in = (__float2_t *) inputComplex;
filter_size = 2;
x[0] = _ftof2(0.5, -0.5);
x[1] = _ftof2(-0.5, 0.5);
//#pragma MUST_ITERATE (128)
for (i=0; i < N_E; i = i + 2*filter_size)
{
sum =_ftof2(0.0, 0.0);
for (j=0; j < filter_size; j++)
{
y = _mem8_f2(p_in++);
z = complexMultiply(y,x[j]);
sum = _daddsp(sum, z);
}
}
total_sum = _hif2(sum) * _hif2(sum) + _lof2(sum) * _lof2(sum);
*output++ = total_sum;
/*
* End of filter 1
*/
/*
* Second filter, coefficients are (1/4 -1/4) (1/2 -1/2) (-1/4 1/4) (-1/2 1/2)
*/
p_in = (__float2_t *)inputComplex;
filter_size = 4;
x[0] = _ftof2(0.25, -0.25);
x[1] = _ftof2(0.5, -0.5);
x[2] = _ftof2(-0.25, 0.25);
x[3] = _ftof2(-0.5, 0.5);
//#pragma MUST_ITERATE (64)
for (i=0; i < N_E; i = i + 2*filter_size)
{
sum =_ftof2(0.0, 0.0);
for (j=0; j < filter_size; j++)
{
y = _mem8_f2(p_in++);
z = complexMultiply(y,x[j]);
sum = _daddsp(sum, z);
}
}
total_sum = _hif2(sum) * _hif2(sum) + _lof2(sum) * _lof2(sum);
*output++ = total_sum;
/*
* End of filter 2
*/
/*
* Third filter, coefficients are (1/4 -1/4) (1/2 -1/2) (-1/4 1/4) (-1/2 1/2)
*/
p_in = (__float2_t *) inputComplex;
filter_size = 8;
x[0] = _ftof2(0.5, -0.5);
x[1] = _ftof2(1.0, -1.0);
x[2] = _ftof2(0.5, -0.5);
x[3] = _ftof2(0.0, 0.0);
x[4] = _ftof2(-0.5, 0.5);
x[5] = _ftof2(-1.0, 1.0);
x[6] = _ftof2(-0.5, 0.5);
x[7] = _ftof2(0.0, 0.0);
//#pragma MUST_ITERATE (32)
for (i=0; i < N_E; i = i + 2*filter_size)
{
sum =_ftof2(0.0, 0.0);
for (j=0; j < filter_size; j++)
{
y = _mem8_f2(p_in++);
z = complexMultiply(y,x[j]);
sum = _daddsp(sum, z);
}
}
total_sum = _hif2(sum) * _hif2(sum) + _lof2(sum) * _lof2(sum);
*output++ = total_sum;
p_in = (__float2_t *) inputComplex ;
filter_size = 16 ;
x[0] = _ftof2(0.25, -0.25) ;
x[1] = _ftof2(0.5, -0.5) ;
x[2] = _ftof2(0.75, -0.75) ;
x[3] = _ftof2(1.0, -1.0) ;
x[4] = _ftof2(0.75, -0.75) ;
x[5] = _ftof2(0.5, -0.5) ;
x[6] = _ftof2(0.25, -0.25) ;
x[7] = _ftof2(0.0, 0.0) ;
x[8] = _ftof2(-0.25, 0.25) ;
x[9] = _ftof2(-0.5, 0.5) ;
x[10] = _ftof2(-0.75, 0.75) ;
x[11] = _ftof2(-1.0, 1.0) ;
x[12] = _ftof2(-0.75, 0.75) ;
x[13] = _ftof2(-0.5, 0.5) ;
x[14] = _ftof2(-0.25, 0.25) ;
x[15] = _ftof2(0.0, 0.0) ;
//#pragma MUST_ITERATE (16)
for (i=0; i < N_E; i = i + 2*filter_size)
{
sum =_ftof2(0.0, 0.0) ;
for (j=0; j < filter_size; j++)
{
y = _mem8_f2(p_in++);
z = complexMultiply(y,x[j]);
sum = _daddsp(sum, z);
}
}
total_sum = _hif2(sum) * _hif2(sum) + _lof2(sum) * _lof2(sum);
*output++ = total_sum;
/*
* End of filter 3
*/
}
