k60-keil

K60-Keil版本（下载安装MDK4.23）

        /* (yb - yd) */
        t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u];

        /* (xb - xd) */
        t2 = pSrc[2u * i1] - pSrc[2u * i3];

        /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */
        pSrc[2u * i1] = (r1 * co2) + (s1 * si2);

        /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */
        pSrc[(2u * i1) + 1u] = (s1 * co2) - (r1 * si2);

        /* (xa - xc) + (yb - yd) */
        r1 = r2 + t1;

        /* (xa - xc) - (yb - yd) */
        r2 = r2 - t1;

        /* (ya - yc) -  (xb - xd) */
        s1 = s2 - t2;

        /* (ya - yc) +  (xb - xd) */
        s2 = s2 + t2;

        /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */
        pSrc[2u * i2] = (r1 * co1) + (s1 * si1);

        /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */
        pSrc[(2u * i2) + 1u] = (s1 * co1) - (r1 * si1);

        /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */
        pSrc[2u * i3] = (r2 * co3) + (s2 * si3);

        /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */
        pSrc[(2u * i3) + 1u] = (s2 * co3) - (r2 * si3);
      }
    }
    twidCoefModifier <<= 2u;
  }

  /*  Initializations of last stage */
  n1 = n2;
  n2 >>= 2u;

  /*  Calculations of last stage */
  for (i0 = 0u; i0 <= (fftLen - n1); i0 += n1)
  {
    /*  index calculation for the input as, */
    /*  pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */
    i1 = i0 + n2;
    i2 = i1 + n2;
    i3 = i2 + n2;

    /*  Butterfly implementation */

    /* xa + xb */
    r1 = pSrc[2u * i0] + pSrc[2u * i2];

    /* xa - xb */
    r2 = pSrc[2u * i0] - pSrc[2u * i2];

    /* ya + yc */
    s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u];

    /* ya - yc */
    s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u];

    /* xc + xd */
    t1 = pSrc[2u * i1] + pSrc[2u * i3];

    /* xa' = xa + xb + xc + xd */
    pSrc[2u * i0] = r1 + t1;

    /* (xa + xb) - (xc + xd) */
    r1 = r1 - t1;

    /* yb + yd */
    t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u];

    /* ya' = ya + yb + yc + yd */
    pSrc[(2u * i0) + 1u] = s1 + t2;

    /* (ya + yc) - (yb + yd) */
    s1 = s1 - t2;

    /* (yb-yd) */
    t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u];

    /* (xb-xd) */
    t2 = pSrc[2u * i1] - pSrc[2u * i3];

    /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */
    pSrc[2u * i1] = r1;

    /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */
    pSrc[(2u * i1) + 1u] = s1;

    /* (xa+yb-xc-yd) */
    r1 = r2 + t1;

    /* (xa-yb-xc+yd) */
    r2 = r2 - t1;

    /* (ya-xb-yc+xd) */
    s1 = s2 - t2;

    /* (ya+xb-yc-xd) */
    s2 = s2 + t2;

    /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */
    pSrc[2u * i2] = r1;

    /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */
    pSrc[(2u * i2) + 1u] = s1;

    /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */
    pSrc[2u * i3] = r2;

    /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */
    pSrc[(2u * i3) + 1u] = s2;
  }


#else

  /* Run the below code for Cortex-M0 */

  /*  Initializations for the fft calculation */
  n2 = fftLen;
  n1 = n2;
  for (k = fftLen; k > 1u; k >>= 2u)
  {
    /*  Initializations for the fft calculation */
    n1 = n2;
    n2 >>= 2u;
    ia1 = 0u;

    /*  FFT Calculation */
    for (j = 0u; j <= (n2 - 1u); j++)
    {
      /*  index calculation for the coefficients */
      ia2 = ia1 + ia1;
      ia3 = ia2 + ia1;
      co1 = pCoef[ia1 * 2u];
      si1 = pCoef[(ia1 * 2u) + 1u];
      co2 = pCoef[ia2 * 2u];
      si2 = pCoef[(ia2 * 2u) + 1u];
      co3 = pCoef[ia3 * 2u];
      si3 = pCoef[(ia3 * 2u) + 1u];

      /*  Twiddle coefficients index modifier */
      ia1 = ia1 + twidCoefModifier;

      for (i0 = j; i0 < fftLen; i0 += n1)
      {
        /*  index calculation for the input as, */
        /*  pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */
        i1 = i0 + n2;
        i2 = i1 + n2;
        i3 = i2 + n2;

        /* xa + xc */
        r1 = pSrc[(2u * i0)] + pSrc[(2u * i2)];

        /* xa - xc */
        r2 = pSrc[(2u * i0)] - pSrc[(2u * i2)];

        /* ya + yc */
        s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u];

        /* ya - yc */
        s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u];

        /* xb + xd */
        t1 = pSrc[2u * i1] + pSrc[2u * i3];

        /* xa' = xa + xb + xc + xd */
        pSrc[2u * i0] = r1 + t1;

        /* xa + xc -(xb + xd) */
        r1 = r1 - t1;

        /* yb + yd */
        t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u];

        /* ya' = ya + yb + yc + yd */
        pSrc[(2u * i0) + 1u] = s1 + t2;

        /* (ya + yc) - (yb + yd) */
        s1 = s1 - t2;

        /* (yb - yd) */
        t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u];

        /* (xb - xd) */
        t2 = pSrc[2u * i1] - pSrc[2u * i3];

        /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */
        pSrc[2u * i1] = (r1 * co2) + (s1 * si2);

        /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */
        pSrc[(2u * i1) + 1u] = (s1 * co2) - (r1 * si2);

        /* (xa - xc) + (yb - yd) */
        r1 = r2 + t1;

        /* (xa - xc) - (yb - yd) */
        r2 = r2 - t1;

        /* (ya - yc) -  (xb - xd) */
        s1 = s2 - t2;

        /* (ya - yc) +  (xb - xd) */
        s2 = s2 + t2;

        /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */
        pSrc[2u * i2] = (r1 * co1) + (s1 * si1);

        /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */
        pSrc[(2u * i2) + 1u] = (s1 * co1) - (r1 * si1);

        /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */
        pSrc[2u * i3] = (r2 * co3) + (s2 * si3);

        /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */
        pSrc[(2u * i3) + 1u] = (s2 * co3) - (r2 * si3);
      }
    }
    twidCoefModifier <<= 2u;
  }

#endif /* #ifndef ARM_MATH_CM0 */

}

/*   
 * @brief  Core function for the floating-point CIFFT butterfly process.  
 * @param[in, out] *pSrc            points to the in-place buffer of floating-point data type.  
 * @param[in]      fftLen           length of the FFT.  
 * @param[in]      *pCoef           points to twiddle coefficient buffer.  
 * @param[in]      twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.  
 * @param[in]      onebyfftLen      value of 1/fftLen.  
 * @return none.  
 */

void arm_radix4_butterfly_inverse_f32(
  float32_t * pSrc,
  uint16_t fftLen,
  float32_t * pCoef,
  uint16_t twidCoefModifier,
  float32_t onebyfftLen)
{
  float32_t co1, co2, co3, si1, si2, si3;
  float32_t t1, t2, r1, r2, s1, s2;
  uint32_t ia1, ia2, ia3;
  uint32_t i0, i1, i2, i3;
  uint32_t n1, n2, j, k;

#ifndef ARM_MATH_CM0

  /* Run the below code for Cortex-M4 and Cortex-M3 */

  /*  Initializations for the first stage */
  n2 = fftLen;
  n1 = n2;

  /* n2 = fftLen/4 */
  n2 >>= 2u;
  i0 = 0u;
  ia1 = 0u;

  j = n2;

  /*  Calculation of first stage */
  do
  {
    /*  index calculation for the input as, */
    /*  pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */
    i1 = i0 + n2;
    i2 = i1 + n2;
    i3 = i2 + n2;

    /*  Butterfly implementation */
    /* xa + xc */
    r1 = pSrc[(2u * i0)] + pSrc[(2u * i2)];

    /* xa - xc */
    r2 = pSrc[2u * i0] - pSrc[2u * i2];

    /* ya + yc */
    s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u];

    /* ya - yc */
    s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u];

    /* xb + xd */
    t1 = pSrc[2u * i1] + pSrc[2u * i3];

    /* xa' = xa + xb + xc + xd */
    pSrc[2u * i0] = r1 + t1;

    /* (xa + xc) - (xb + xd) */
    r1 = r1 - t1;

    /* yb + yd */
    t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u];

    /* ya' = ya + yb + yc + yd */
    pSrc[(2u * i0) + 1u] = s1 + t2;

    /* (ya + yc) - (yb + yd) */
    s1 = s1 - t2;

    /* yb - yd */
    t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u];

    /* xb - xd */
    t2 = pSrc[2u * i1] - pSrc[2u * i3];

    /*  index calculation for the coefficients */
    ia2 = ia1 + ia1;
    co2 = pCoef[ia2 * 2u];
    si2 = pCoef[(ia2 * 2u) + 1u];

    /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */
    pSrc[2u * i1] = (r1 * co2) - (s1 * si2);

    /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */
    pSrc[(2u * i1) + 1u] = (s1 * co2) + (r1 * si2);

    /* (xa - xc) - (yb - yd) */
    r1 = r2 - t1;

    /* (xa - xc) + (yb - yd) */
    r2 = r2 + t1;

    /* (ya - yc) + (xb - xd) */
    s1 = s2 + t2;

    /* (ya - yc) - (xb - xd) */
    s2 = s2 - t2;

    co1 = pCoef[ia1 * 2u];
    si1 = pCoef[(ia1 * 2u) + 1u];

    /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */
    pSrc[2u * i2] = (r1 * co1) - (s1 * si1);

    /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */
    pSrc[(2u * i2) + 1u] = (s1 * co1) + (r1 * si1);

    /*  index calculation for the coefficients */
    ia3 = ia2 + ia1;
    co3 = pCoef[ia3 * 2u];
    si3 = pCoef[(ia3 * 2u) + 1u];

    /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */
    pSrc[2u * i3] = (r2 * co3) - (s2 * si3);

    /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */
    pSrc[(2u * i3) + 1u] = (s2 * co3) + (r2 * si3);

    /*  Twiddle coefficients index modifier */
    ia1 = ia1 + twidCoefModifier;

    /*  Updating input index */
    i0 = i0 + 1u;

  }
  while(--j);

  twidCoefModifier <<= 2u;

  /*  Calculation of second stage to excluding last stage */
  for (k = fftLen / 4; k > 4u; k >>= 2u)
  {
    /*  Initializations for the first stage */
    n1 = n2;
    n2 >>= 2u;
    ia1 = 0u;

    /*  Calculation of first stage */
    for (j = 0u; j <= (n2 - 1u); j++)
    {
      /*  index calculation for the coefficients */
      ia2 = ia1 + ia1;
      ia3 = ia2 + ia1;
      co1 = pCoef[ia1 * 2u];
      si1 = pCoef[(ia1 * 2u) + 1u];
      co2 = pCoef[ia2 * 2u];
      si2 = pCoef[(ia2 * 2u) + 1u];
      co3 = pCoef[ia3 * 2u];
      si3 = pCoef[(ia3 * 2u) + 1u];

      /*  Twiddle coefficients index modifier */
      ia1 = ia1 + twidCoefModifier;

      for (i0 = j; i0 < fftLen; i0 += n1)
      {
        /*  index calculation for the input as, */
        /*  pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */
        i1 = i0 + n2;
        i2 = i1 + n2;
        i3 = i2 + n2;

        /* xa + xc */
        r1 = pSrc[(2u * i0)] + pSrc[(2u * i2)];

        /* xa - xc */
        r2 = pSrc[(2u * i0)] - pSrc[(2u * i2)];

        /* ya + yc */