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_conv_partial_f32.c   
*   
* Description:	Partial convolution of floating-point sequences.   
*   
* 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 0.0.7  2010/06/10    
*    Misra-C changes done   
*   
* -------------------------------------------------------------------------- */

#include "arm_math.h"

/**   
 * @ingroup groupFilters   
 */

/**   
 * @defgroup PartialConv Partial Convolution   
 *   
 * Partial Convolution is equivalent to Convolution except that a subset of the output samples is generated.   
 * Each function has two additional arguments.   
 * <code>firstIndex</code> specifies the starting index of the subset of output samples.   
 * <code>numPoints</code> is the number of output samples to compute.   
 * The function computes the output in the range   
 * <code>[firstIndex, ..., firstIndex+numPoints-1]</code>.   
 * The output array <code>pDst</code> contains <code>numPoints</code> values.   
 *   
 * The allowable range of output indices is [0 srcALen+srcBLen-2].   
 * If the requested subset does not fall in this range then the functions return ARM_MATH_ARGUMENT_ERROR.   
 * Otherwise the functions return ARM_MATH_SUCCESS.   
 * \note Refer arm_conv_f32() for details on fixed point behavior.  
 */

/**   
 * @addtogroup PartialConv   
 * @{   
 */

/**   
 * @brief Partial convolution of floating-point sequences.   
 * @param[in]       *pSrcA points to the first input sequence.   
 * @param[in]       srcALen length of the first input sequence.   
 * @param[in]       *pSrcB points to the second input sequence.   
 * @param[in]       srcBLen length of the second input sequence.   
 * @param[out]      *pDst points to the location where the output result is written.   
 * @param[in]       firstIndex is the first output sample to start with.   
 * @param[in]       numPoints is the number of output points to be computed.   
 * @return  Returns either ARM_MATH_SUCCESS if the function completed correctly or ARM_MATH_ARGUMENT_ERROR if the requested subset is not in the range [0 srcALen+srcBLen-2].   
 */

arm_status arm_conv_partial_f32(
  float32_t * pSrcA,
  uint32_t srcALen,
  float32_t * pSrcB,
  uint32_t srcBLen,
  float32_t * pDst,
  uint32_t firstIndex,
  uint32_t numPoints)
{


#ifndef ARM_MATH_CM0

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

  float32_t *pIn1 = pSrcA;                       /* inputA pointer */
  float32_t *pIn2 = pSrcB;                       /* inputB pointer */
  float32_t *pOut = pDst;                        /* output pointer */
  float32_t *px;                                 /* Intermediate inputA pointer */
  float32_t *py;                                 /* Intermediate inputB pointer */
  float32_t *pSrc1, *pSrc2;                      /* Intermediate pointers */
  float32_t sum, acc0, acc1, acc2, acc3;         /* Accumulator */
  float32_t x0, x1, x2, x3, c0;                  /* Temporary variables to hold state and coefficient values */
  uint32_t j, k, count = 0u, blkCnt, check;
  int32_t blockSize1, blockSize2, blockSize3;    /* loop counters */
  arm_status status;                             /* status of Partial convolution */


  /* Check for range of output samples to be calculated */
  if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
  {
    /* Set status as ARM_MATH_ARGUMENT_ERROR */
    status = ARM_MATH_ARGUMENT_ERROR;
  }
  else
  {

    /* The algorithm implementation is based on the lengths of the inputs. */
    /* srcB is always made to slide across srcA. */
    /* So srcBLen is always considered as shorter or equal to srcALen */
    if(srcALen >= srcBLen)
    {
      /* Initialization of inputA pointer */
      pIn1 = pSrcA;

      /* Initialization of inputB pointer */
      pIn2 = pSrcB;
    }
    else
    {
      /* Initialization of inputA pointer */
      pIn1 = pSrcB;

      /* Initialization of inputB pointer */
      pIn2 = pSrcA;

      /* srcBLen is always considered as shorter or equal to srcALen */
      j = srcBLen;
      srcBLen = srcALen;
      srcALen = j;
    }

    /* Conditions to check which loopCounter holds   
     * the first and last indices of the output samples to be calculated. */
    check = firstIndex + numPoints;
    blockSize3 = (int32_t) check - (int32_t) srcALen;
    blockSize3 = (blockSize3 > 0) ? blockSize3 : 0;
    blockSize1 = ((int32_t) srcBLen - 1) - (int32_t) firstIndex;
    blockSize1 = (blockSize1 > 0) ? ((check > (srcBLen - 1u)) ? blockSize1 :
                                     (int32_t) numPoints) : 0;
    blockSize2 = ((int32_t) check - blockSize3) -
      (blockSize1 + (int32_t) firstIndex);
    blockSize2 = (blockSize2 > 0) ? blockSize2 : 0;

    /* conv(x,y) at n = x[n] * y[0] + x[n-1] * y[1] + x[n-2] * y[2] + ...+ x[n-N+1] * y[N -1] */
    /* The function is internally   
     * divided into three stages according to the number of multiplications that has to be   
     * taken place between inputA samples and inputB samples. In the first stage of the   
     * algorithm, the multiplications increase by one for every iteration.   
     * In the second stage of the algorithm, srcBLen number of multiplications are done.   
     * In the third stage of the algorithm, the multiplications decrease by one   
     * for every iteration. */

    /* Set the output pointer to point to the firstIndex   
     * of the output sample to be calculated. */
    pOut = pDst + firstIndex;

    /* --------------------------   
     * Initializations of stage1   
     * -------------------------*/

    /* sum = x[0] * y[0]   
     * sum = x[0] * y[1] + x[1] * y[0]   
     * ....   
     * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]   
     */

    /* In this stage the MAC operations are increased by 1 for every iteration.   
       The count variable holds the number of MAC operations performed.   
       Since the partial convolution starts from from firstIndex   
       Number of Macs to be performed is firstIndex + 1 */
    count = 1u + firstIndex;

    /* Working pointer of inputA */
    px = pIn1;

    /* Working pointer of inputB */
    pSrc1 = pIn2 + firstIndex;
    py = pSrc1;

    /* ------------------------   
     * Stage1 process   
     * ----------------------*/

    /* The first stage starts here */
    while(blockSize1 > 0)
    {
      /* Accumulator is made zero for every iteration */
      sum = 0.0f;

      /* Apply loop unrolling and compute 4 MACs simultaneously. */
      k = count >> 2u;

      /* First part of the processing with loop unrolling.  Compute 4 MACs at a time.   
       ** a second loop below computes MACs for the remaining 1 to 3 samples. */
      while(k > 0u)
      {
        /* x[0] * y[srcBLen - 1] */
        sum += *px++ * *py--;

        /* x[1] * y[srcBLen - 2] */
        sum += *px++ * *py--;

        /* x[2] * y[srcBLen - 3] */
        sum += *px++ * *py--;

        /* x[3] * y[srcBLen - 4] */
        sum += *px++ * *py--;

        /* Decrement the loop counter */
        k--;
      }

      /* If the count is not a multiple of 4, compute any remaining MACs here.   
       ** No loop unrolling is used. */
      k = count % 0x4u;

      while(k > 0u)
      {
        /* Perform the multiply-accumulates */
        sum += *px++ * *py--;

        /* Decrement the loop counter */
        k--;
      }

      /* Store the result in the accumulator in the destination buffer. */
      *pOut++ = sum;

      /* Update the inputA and inputB pointers for next MAC calculation */
      py = ++pSrc1;
      px = pIn1;

      /* Increment the MAC count */
      count++;

      /* Decrement the loop counter */
      blockSize1--;
    }

    /* --------------------------   
     * Initializations of stage2   
     * ------------------------*/

    /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]   
     * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]   
     * ....   
     * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]   
     */

    /* Working pointer of inputA */
    px = pIn1;

    /* Working pointer of inputB */
    pSrc2 = pIn2 + (srcBLen - 1u);
    py = pSrc2;

    /* count is index by which the pointer pIn1 to be incremented */
    count = 1u;

    /* -------------------   
     * Stage2 process   
     * ------------------*/

    /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.   
     * So, to loop unroll over blockSize2,   
     * srcBLen should be greater than or equal to 4 */
    if(srcBLen >= 4u)
    {
      /* Loop unroll over blockSize2, by 4 */
      blkCnt = ((uint32_t) blockSize2 >> 2u);

      while(blkCnt > 0u)
      {
        /* Set all accumulators to zero */
        acc0 = 0.0f;
        acc1 = 0.0f;
        acc2 = 0.0f;
        acc3 = 0.0f;

        /* read x[0], x[1], x[2] samples */
        x0 = *(px++);
        x1 = *(px++);
        x2 = *(px++);

        /* Apply loop unrolling and compute 4 MACs simultaneously. */
        k = srcBLen >> 2u;

        /* First part of the processing with loop unrolling.  Compute 4 MACs at a time.   
         ** a second loop below computes MACs for the remaining 1 to 3 samples. */
        do
        {
          /* Read y[srcBLen - 1] sample */
          c0 = *(py--);

          /* Read x[3] sample */
          x3 = *(px++);

          /* Perform the multiply-accumulate */
          /* acc0 +=  x[0] * y[srcBLen - 1] */
          acc0 += x0 * c0;

          /* acc1 +=  x[1] * y[srcBLen - 1] */
          acc1 += x1 * c0;

          /* acc2 +=  x[2] * y[srcBLen - 1] */
          acc2 += x2 * c0;

          /* acc3 +=  x[3] * y[srcBLen - 1] */
          acc3 += x3 * c0;

          /* Read y[srcBLen - 2] sample */
          c0 = *(py--);

          /* Read x[4] sample */
          x0 = *(px++);

          /* Perform the multiply-accumulate */