k60-keil

K60-Keil版本（下载安装MDK4.23）

          /* Read y[srcBLen - 3] sample */
          c0 = *(py--);

          /* Read x[5] sample */
          x1 = *(px++);

          /* Perform the multiply-accumulates */
          /* acc0 +=  x[2] * y[srcBLen - 3] */
          acc0 += (q63_t) x2 *c0;
          /* acc1 +=  x[3] * y[srcBLen - 2] */
          acc1 += (q63_t) x3 *c0;
          /* acc2 +=  x[4] * y[srcBLen - 2] */
          acc2 += (q63_t) x0 *c0;
          /* acc3 +=  x[5] * y[srcBLen - 2] */
          acc3 += (q63_t) x1 *c0;

          /* Read y[srcBLen - 4] sample */
          c0 = *(py--);

          /* Read x[6] sample */
          x2 = *(px++);

          /* Perform the multiply-accumulates */
          /* acc0 +=  x[3] * y[srcBLen - 4] */
          acc0 += (q63_t) x3 *c0;
          /* acc1 +=  x[4] * y[srcBLen - 4] */
          acc1 += (q63_t) x0 *c0;
          /* acc2 +=  x[5] * y[srcBLen - 4] */
          acc2 += (q63_t) x1 *c0;
          /* acc3 +=  x[6] * y[srcBLen - 4] */
          acc3 += (q63_t) x2 *c0;

        } while(--k);

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

        while(k > 0u)
        {
          /* Read y[srcBLen - 5] sample */
          c0 = *(py--);

          /* Read x[7] sample */
          x3 = *(px++);

          /* Perform the multiply-accumulates */
          /* acc0 +=  x[4] * y[srcBLen - 5] */
          acc0 += (q63_t) x0 *c0;
          /* acc1 +=  x[5] * y[srcBLen - 5] */
          acc1 += (q63_t) x1 *c0;
          /* acc2 +=  x[6] * y[srcBLen - 5] */
          acc2 += (q63_t) x2 *c0;
          /* acc3 +=  x[7] * y[srcBLen - 5] */
          acc3 += (q63_t) x3 *c0;

          /* Reuse the present samples for the next MAC */
          x0 = x1;
          x1 = x2;
          x2 = x3;

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

        /* Store the result in the accumulator in the destination buffer. */
        *pOut++ = (q31_t) (acc0 >> 31);
        *pOut++ = (q31_t) (acc1 >> 31);
        *pOut++ = (q31_t) (acc2 >> 31);
        *pOut++ = (q31_t) (acc3 >> 31);

        /* Update the inputA and inputB pointers for next MAC calculation */
        px = pIn1 + (count * 4u);
        py = pSrc2;

        /* Increment the pointer pIn1 index, count by 1 */
        count++;

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

      /* If the blockSize2 is not a multiple of 4, compute any remaining output samples here.   
       ** No loop unrolling is used. */
      blkCnt = (uint32_t) blockSize2 % 0x4u;

      while(blkCnt > 0u)
      {
        /* Accumulator is made zero for every iteration */
        sum = 0;

        /* 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. */
        while(k > 0u)
        {
          /* Perform the multiply-accumulates */
          sum += (q63_t) * px++ * (*py--);
          sum += (q63_t) * px++ * (*py--);
          sum += (q63_t) * px++ * (*py--);
          sum += (q63_t) * px++ * (*py--);

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

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

        while(k > 0u)
        {
          /* Perform the multiply-accumulate */
          sum += (q63_t) * px++ * (*py--);

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

        /* Store the result in the accumulator in the destination buffer. */
        *pOut++ = (q31_t) (sum >> 31);

        /* Update the inputA and inputB pointers for next MAC calculation */
        px = pIn1 + count;
        py = pSrc2;

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

        /* Decrement the loop counter */
        blkCnt--;
      }
    }
    else
    {
      /* If the srcBLen is not a multiple of 4,   
       * the blockSize2 loop cannot be unrolled by 4 */
      blkCnt = (uint32_t) blockSize2;

      while(blkCnt > 0u)
      {
        /* Accumulator is made zero for every iteration */
        sum = 0;

        /* srcBLen number of MACS should be performed */
        k = srcBLen;

        while(k > 0u)
        {
          /* Perform the multiply-accumulate */
          sum += (q63_t) * px++ * (*py--);

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

        /* Store the result in the accumulator in the destination buffer. */
        *pOut++ = (q31_t) (sum >> 31);

        /* Update the inputA and inputB pointers for next MAC calculation */
        px = pIn1 + count;
        py = pSrc2;

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

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


    /* --------------------------   
     * Initializations of stage3   
     * -------------------------*/

    /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]   
     * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]   
     * ....   
     * sum +=  x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]   
     * sum +=  x[srcALen-1] * y[srcBLen-1]   
     */

    /* In this stage the MAC operations are decreased by 1 for every iteration.   
       The blockSize3 variable holds the number of MAC operations performed */
    count = srcBLen - 1u;

    /* Working pointer of inputA */
    pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
    px = pSrc1;

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

    /* -------------------   
     * Stage3 process   
     * ------------------*/

    while(blockSize3 > 0)
    {
      /* Accumulator is made zero for every iteration */
      sum = 0;

      /* 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)
      {
        sum += (q63_t) * px++ * (*py--);
        sum += (q63_t) * px++ * (*py--);
        sum += (q63_t) * px++ * (*py--);
        sum += (q63_t) * px++ * (*py--);

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

      /* If the blockSize3 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-accumulate */
        sum += (q63_t) * px++ * (*py--);

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

      /* Store the result in the accumulator in the destination buffer. */
      *pOut++ = (q31_t) (sum >> 31);

      /* Update the inputA and inputB pointers for next MAC calculation */
      px = ++pSrc1;
      py = pSrc2;

      /* Decrement the MAC count */
      count--;

      /* Decrement the loop counter */
      blockSize3--;

    }

    /* set status as ARM_MATH_SUCCESS */
    status = ARM_MATH_SUCCESS;
  }

  /* Return to application */
  return (status);

#else

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

  q31_t *pIn1 = pSrcA;                           /* inputA pointer */
  q31_t *pIn2 = pSrcB;                           /* inputB pointer */
  q63_t sum;                                     /* Accumulator */
  uint32_t i, j;                                 /* 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_ARGUMENT_ERROR */
    status = ARM_MATH_ARGUMENT_ERROR;
  }
  else
  {
    /* Loop to calculate convolution for output length number of values */
    for (i = firstIndex; i <= (firstIndex + numPoints - 1); i++)
    {
      /* Initialize sum with zero to carry on MAC operations */
      sum = 0;

      /* Loop to perform MAC operations according to convolution equation */
      for (j = 0; j <= i; j++)
      {
        /* Check the array limitations */
        if(((i - j) < srcBLen) && (j < srcALen))
        {
          /* z[i] += x[i-j] * y[j] */
          sum += ((q63_t) pIn1[j] * (pIn2[i - j]));
        }
      }

      /* Store the output in the destination buffer */
      pDst[i] = (q31_t) (sum >> 31u);
    }
    /* set status as ARM_SUCCESS as there are no argument errors */
    status = ARM_MATH_SUCCESS;
  }
  return (status);

#endif /*    #ifndef ARM_MATH_CM0      */

}

/**   
 * @} end of PartialConv group   
 */