k60-keil

K60-Keil版本（下载安装MDK4.23）

/* ----------------------------------------------------------------------   
* Copyright (C) 2010 ARM Limited. All rights reserved.   
*   
* $Date:        15. July 2011  
* $Revision: 	V1.0.10  
*   
* Project: 	    CMSIS DSP Library   
* Title:	    arm_cfft_radix4_q15.c   
*   
* Description:	This file has function definition of Radix-4 FFT & IFFT function and   
*				In-place bit reversal using bit reversal table   
*   
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
*  
* Version 1.0.10 2011/7/15 
*    Big Endian support added and Merged M0 and M3/M4 Source code.  
*   
* Version 1.0.3 2010/11/29  
*    Re-organized the CMSIS folders and updated documentation.   
*    
* Version 1.0.2 2010/11/11   
*    Documentation updated.    
*   
* Version 1.0.1 2010/10/05    
*    Production release and review comments incorporated.   
*   
* Version 1.0.0 2010/09/20    
*    Production release and review comments incorporated.   
*   
* Version 0.0.5  2010/04/26    
* 	 incorporated review comments and updated with latest CMSIS layer   
*   
* Version 0.0.3  2010/03/10    
*    Initial version   
* -------------------------------------------------------------------- */

#include "arm_math.h"

/**   
 * @ingroup groupTransforms   
 */

/**   
 * @addtogroup CFFT_CIFFT   
 * @{   
 */


/**   
 * @details   
 * @brief Processing function for the Q15 CFFT/CIFFT.  
 * @param[in]      *S    points to an instance of the Q15 CFFT/CIFFT structure.  
 * @param[in, out] *pSrc points to the complex data buffer. Processing occurs in-place.  
 * @return none.  
 *    
 * \par Input and output formats:   
 * \par   
 * Internally input is downscaled by 2 for every stage to avoid saturations inside CFFT/CIFFT process.  
 * Hence the output format is different for different FFT sizes.   
 * The input and output formats for different FFT sizes and number of bits to upscale are mentioned in the tables below for CFFT and CIFFT:  
 * \par  
 * \image html CFFTQ15.gif "Input and Output Formats for Q15 CFFT"   
 * \image html CIFFTQ15.gif "Input and Output Formats for Q15 CIFFT"   
 */

void arm_cfft_radix4_q15(
  const arm_cfft_radix4_instance_q15 * S,
  q15_t * pSrc)
{
  if(S->ifftFlag == 1u)
  {
    /*  Complex IFFT radix-4  */
    arm_radix4_butterfly_inverse_q15(pSrc, S->fftLen, S->pTwiddle,
                                     S->twidCoefModifier);
  }
  else
  {
    /*  Complex FFT radix-4  */
    arm_radix4_butterfly_q15(pSrc, S->fftLen, S->pTwiddle,
                             S->twidCoefModifier);
  }

  if(S->bitReverseFlag == 1u)
  {
    /*  Bit Reversal */
    arm_bitreversal_q15(pSrc, S->fftLen, S->bitRevFactor, S->pBitRevTable);
  }

}

/**   
 * @} end of CFFT_CIFFT group   
 */

/*   
* Radix-4 FFT algorithm used is :   
*   
* Input real and imaginary data:   
* x(n) = xa + j * ya   
* x(n+N/4 ) = xb + j * yb   
* x(n+N/2 ) = xc + j * yc   
* x(n+3N 4) = xd + j * yd   
*   
*   
* Output real and imaginary data:   
* x(4r) = xa'+ j * ya'   
* x(4r+1) = xb'+ j * yb'   
* x(4r+2) = xc'+ j * yc'   
* x(4r+3) = xd'+ j * yd'   
*   
*   
* Twiddle factors for radix-4 FFT:   
* Wn = co1 + j * (- si1)   
* W2n = co2 + j * (- si2)   
* W3n = co3 + j * (- si3)   
   
* The real and imaginary output values for the radix-4 butterfly are   
* xa' = xa + xb + xc + xd   
* ya' = ya + yb + yc + yd   
* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1)   
* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1)   
* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2)   
* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2)   
* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3)   
* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3)   
*   
*/

/**   
 * @brief  Core function for the Q15 CFFT butterfly process.  
 * @param[in, out] *pSrc16          points to the in-place buffer of Q15 data type.  
 * @param[in]      fftLen           length of the FFT.  
 * @param[in]      *pCoef16         points to twiddle coefficient buffer.  
 * @param[in]      twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.  
 * @return none.  
 */

void arm_radix4_butterfly_q15(
  q15_t * pSrc16,
  uint32_t fftLen,
  q15_t * pCoef16,
  uint32_t twidCoefModifier)
{

#ifndef ARM_MATH_CM0

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

  q31_t R, S, T, U;
  q31_t C1, C2, C3, out1, out2;
  q31_t *pSrc, *pCoeff;
  uint32_t n1, n2, ic, i0, i1, i2, i3, j, k;
  q15_t in;

  /* Total process is divided into three stages */

  /* process first stage, middle stages, & last stage */

  /*  pointer initializations for SIMD calculations */
  pSrc = (q31_t *) pSrc16;
  pCoeff = (q31_t *) pCoef16;

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

  /* n2 = fftLen/4 */
  n2 >>= 2u;

  /* Index for twiddle coefficient */
  ic = 0u;

  /* Index for input read and output write */
  i0 = 0u;
  j = n2;

  /* Input is in 1.15(q15) format */

  /*  start of first stage process */
  do
  {
    /*  Butterfly implementation */

    /*  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;

    /*  Reading i0, i0+fftLen/2 inputs */
    /* Read ya (real), xa(imag) input */
    T = pSrc[i0];
    in = ((int16_t) (T & 0xFFFF)) >> 2;
    T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF);
    /* Read yc (real), xc(imag) input */
    S = pSrc[i2];
    in = ((int16_t) (S & 0xFFFF)) >> 2;
    S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF);
    /* R = packed((ya + yc), (xa + xc) ) */
    R = __QADD16(T, S);
    /* S = packed((ya - yc), (xa - xc) ) */
    S = __QSUB16(T, S);

    /*  Reading i0+fftLen/4 , i0+3fftLen/4 inputs */
    /* Read yb (real), xb(imag) input */
    T = pSrc[i1];
    in = ((int16_t) (T & 0xFFFF)) >> 2;
    T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF);
    /* Read yd (real), xd(imag) input */
    U = pSrc[i3];
    in = ((int16_t) (U & 0xFFFF)) >> 2;
    U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF);
    /* T = packed((yb + yd), (xb + xd) ) */
    T = __QADD16(T, U);

    /*  writing the butterfly processed i0 sample */
    /* xa' = xa + xb + xc + xd */
    /* ya' = ya + yb + yc + yd */
    pSrc[i0] = __SHADD16(R, T);

    /* R = packed((ya + yc) - (yb + yd), (xa + xc)- (xb + xd)) */
    R = __QSUB16(R, T);

    /* co2 & si2 are read from SIMD Coefficient pointer */
    C2 = pCoeff[2u * ic];


#ifndef ARM_MATH_BIG_ENDIAN

    /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */
    out1 = __SMUAD(C2, R) >> 16u;
    /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */
    out2 = __SMUSDX(C2, R);

#else

    /* xc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */
    out1 = __SMUSDX(R, C2) >> 16u;
    /* yc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */
    out2 = __SMUAD(C2, R);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

    /*  Reading i0+fftLen/4 */
    /* T = packed(yb, xb) */
    T = pSrc[i1];
    in = ((int16_t) (T & 0xFFFF)) >> 2;
    T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF);

    /* writing the butterfly processed i0 + fftLen/4 sample */
    /* writing output(xc', yc') in little endian format */
    pSrc[i1] = (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);

    /*  Butterfly calculations */
    /* U = packed(yd, xd) */
    U = pSrc[i3];
    in = ((int16_t) (U & 0xFFFF)) >> 2;
    U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF);
    /* T = packed(yb-yd, xb-xd) */
    T = __QSUB16(T, U);


#ifndef ARM_MATH_BIG_ENDIAN

    /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */
    R = __QASX(S, T);
    /* S = packed((ya-yc) - (xb- xd),  (xa-xc) + (yb-yd)) */
    S = __QSAX(S, T);

#else

    /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */
    R = __QSAX(S, T);
    /* S = packed((ya-yc) - (xb- xd),  (xa-xc) + (yb-yd)) */
    S = __QASX(S, T);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

    /* co1 & si1 are read from SIMD Coefficient pointer */
    C1 = pCoeff[ic];
    /*  Butterfly process for the i0+fftLen/2 sample */

#ifndef ARM_MATH_BIG_ENDIAN

    /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */
    out1 = __SMUAD(C1, S) >> 16u;
    /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */
    out2 = __SMUSDX(C1, S);

#else

    /* xb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */
    out1 = __SMUSDX(S, C1) >> 16u;
    /* yb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */
    out2 = __SMUAD(C1, S);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

    /* writing output(xb', yb') in little endian format */
    pSrc[i2] = ((out2) & 0xFFFF0000) | ((out1) & 0x0000FFFF);


    /* co3 & si3 are read from SIMD Coefficient pointer */
    C3 = pCoeff[3u * ic];
    /*  Butterfly process for the i0+3fftLen/4 sample */

#ifndef ARM_MATH_BIG_ENDIAN

    /* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */
    out1 = __SMUAD(C3, R) >> 16u;
    /* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */
    out2 = __SMUSDX(C3, R);

#else

    /* xd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */
    out1 = __SMUSDX(R, C3) >> 16u;
    /* yd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */
    out2 = __SMUAD(C3, R);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

    /* writing output(xd', yd') in little endian format */
    pSrc[i3] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);

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

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

  } while(--j);
  /* data is in 4.11(q11) format */

  /* end of first stage process */


  /* start of middle stage process */

  /*  Twiddle coefficients index modifier */
  twidCoefModifier <<= 2u;

  /*  Calculation of Middle stage */
  for (k = fftLen / 4u; k > 4u; k >>= 2u)
  {
    /*  Initializations for the middle stage */
    n1 = n2;
    n2 >>= 2u;
    ic = 0u;

    for (j = 0u; j <= (n2 - 1u); j++)
    {
      /*  index calculation for the coefficients */
      C1 = pCoeff[ic];
      C2 = pCoeff[2u * ic];
      C3 = pCoeff[3u * ic];

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

      /*  Butterfly implementation */
      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;

        /*  Reading i0, i0+fftLen/2 inputs */
        /* Read ya (real), xa(imag) input */
        T = pSrc[i0];

        /* Read yc (real), xc(imag) input */
        S = pSrc[i2];

        /* R = packed( (ya + yc), (xa + xc)) */
        R = __QADD16(T, S);

        /* S = packed((ya - yc), (xa - xc)) */
        S = __QSUB16(T, S);

        /*  Reading i0+fftLen/4 , i0+3fftLen/4 inputs */
        /* Read yb (real), xb(imag) input */
        T = pSrc[i1];

        /* Read yd (real), xd(imag) input */
        U = pSrc[i3];


        /* T = packed( (yb + yd), (xb + xd)) */
        T = __QADD16(T, U);


        /*  writing the butterfly processed i0 sample */

        /* xa' = xa + xb + xc + xd */
        /* ya' = ya + yb + yc + yd */
        out1 = __SHADD16(R, T);
        in = ((int16_t) (out1 & 0xFFFF)) >> 1;
        out1 = ((out1 >> 1) & 0xFFFF0000) | (in & 0xFFFF);
        pSrc[i0] = out1;

        /* R = packed( (ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */
        R = __SHSUB16(R, T);


#ifndef ARM_MATH_BIG_ENDIAN

        /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */
        out1 = __SMUAD(C2, R) >> 16u;

        /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */
        out2 = __SMUSDX(C2, R);

#else

        /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */
        out1 = __SMUSDX(R, C2) >> 16u;

        /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */
        out2 = __SMUAD(C2, R);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /*  Reading i0+3fftLen/4 */
        /* Read yb (real), xb(imag) input */
        T = pSrc[i1];

        /*  writing the butterfly processed i0 + fftLen/4 sample */
        /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */
        /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */
        pSrc[i1] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);

        /*  Butterfly calculations */

        /* Read yd (real), xd(imag) input */
        U = pSrc[i3];

        /* T = packed(yb-yd, xb-xd) */
        T = __QSUB16(T, U);


#ifndef ARM_MATH_BIG_ENDIAN

        /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */
        R = __SHASX(S, T);

        /* S = packed((ya-yc) - (xb- xd),  (xa-xc) + (yb-yd)) */
        S = __SHSAX(S, T);


        /*  Butterfly process for the i0+fftLen/2 sample */
        out1 = __SMUAD(C1, S) >> 16u;
        out2 = __SMUSDX(C1, S);

#else

        /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */
        R = __SHSAX(S, T);

        /* S = packed((ya-yc) - (xb- xd),  (xa-xc) + (yb-yd)) */
        S = __SHASX(S, T);


        /*  Butterfly process for the i0+fftLen/2 sample */
        out1 = __SMUSDX(S, C1) >> 16u;
        out2 = __SMUAD(C1, S);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */
        /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */
        pSrc[i2] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF);

        /*  Butterfly process for the i0+3fftLen/4 sample */

#ifndef ARM_MATH_BIG_ENDIAN

        out1 = __SMUAD(C3, R) >> 16u;
        out2 = __SMUSDX(C3, R);

#else

        out1 = __SMUSDX(R, C3) >> 16u;
        out2 = __SMUAD(C3, R);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */