k60-keil

K60-Keil版本（下载安装MDK4.23）

    i--;
  } while(i > 0u);


  /* ---------------------------------------------------------   
   *     Step2: Calculate RFFT for N-point input   
   * ---------------------------------------------------------- */
  /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */
  arm_rfft_f32(S->pRfft, pInlineBuffer, pState);

        /*----------------------------------------------------------------------   
	 *  Step3: Multiply the FFT output with the weights.   
	 *----------------------------------------------------------------------*/
  arm_cmplx_mult_cmplx_f32(pState, weights, pState, S->N);

  /* ----------- Post-processing ---------- */
  /* DCT-IV can be obtained from DCT-II by the equation,   
   *       Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0)   
   *       Hence, Y4(0) = Y2(0)/2  */
  /* Getting only real part from the output and Converting to DCT-IV */

  /* Initializing the loop counter to N >> 2 for loop unrolling by 4 */
  i = ((uint32_t) S->N - 1u) >> 2u;

  /* pbuff initialized to input buffer. */
  pbuff = pInlineBuffer;

  /* pS1 initialized to pState */
  pS1 = pState;

  /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */
  in = *pS1++ * (float32_t) 0.5;
  /* input buffer acts as inplace, so output values are stored in the input itself. */
  *pbuff++ = in;

  /* pState pointer is incremented twice as the real values are located alternatively in the array */
  pS1++;

  /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.   
   ** a second loop below computes the remaining 1 to 3 samples. */
  do
  {
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
    in = *pS1++ - in;
    *pbuff++ = in;
    /* points to the next real value */
    pS1++;

    in = *pS1++ - in;
    *pbuff++ = in;
    pS1++;

    in = *pS1++ - in;
    *pbuff++ = in;
    pS1++;

    in = *pS1++ - in;
    *pbuff++ = in;
    pS1++;

    /* Decrement the loop counter */
    i--;
  } while(i > 0u);

  /* If the blockSize is not a multiple of 4, compute any remaining output samples here.   
   ** No loop unrolling is used. */
  i = ((uint32_t) S->N - 1u) % 0x4u;

  while(i > 0u)
  {
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
    in = *pS1++ - in;
    *pbuff++ = in;
    /* points to the next real value */
    pS1++;

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


        /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/

  /* Initializing the loop counter to N/4 instead of N for loop unrolling */
  i = (uint32_t) S->N >> 2u;

  /* pbuff initialized to the pInlineBuffer(now contains the output values) */
  pbuff = pInlineBuffer;

  /* Processing with loop unrolling 4 times as N is always multiple of 4.  Compute 4 outputs at a time */
  do
  {
    /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */
    in = *pbuff;
    *pbuff++ = in * S->normalize;

    in = *pbuff;
    *pbuff++ = in * S->normalize;

    in = *pbuff;
    *pbuff++ = in * S->normalize;

    in = *pbuff;
    *pbuff++ = in * S->normalize;

    /* Decrement the loop counter */
    i--;
  } while(i > 0u);


#else

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

  /* Initializing the loop counter to N/2 */
  i = (uint32_t) S->Nby2;

  do
  {
    /* Re-ordering of even and odd elements */
    /* pState[i] =  pInlineBuffer[2*i] */
    *pS1++ = *pbuff++;
    /* pState[N-i-1] = pInlineBuffer[2*i+1] */
    *pS2-- = *pbuff++;

    /* Decrement the loop counter */
    i--;
  } while(i > 0u);

  /* pbuff initialized to input buffer */
  pbuff = pInlineBuffer;

  /* pS1 initialized to pState */
  pS1 = pState;

  /* Initializing the loop counter */
  i = (uint32_t) S->N;

  do
  {
    /* Writing the re-ordered output back to inplace input buffer */
    *pbuff++ = *pS1++;

    /* Decrement the loop counter */
    i--;
  } while(i > 0u);


  /* ---------------------------------------------------------   
   *     Step2: Calculate RFFT for N-point input   
   * ---------------------------------------------------------- */
  /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */
  arm_rfft_f32(S->pRfft, pInlineBuffer, pState);

        /*----------------------------------------------------------------------   
	 *  Step3: Multiply the FFT output with the weights.   
	 *----------------------------------------------------------------------*/
  arm_cmplx_mult_cmplx_f32(pState, weights, pState, S->N);

  /* ----------- Post-processing ---------- */
  /* DCT-IV can be obtained from DCT-II by the equation,   
   *       Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0)   
   *       Hence, Y4(0) = Y2(0)/2  */
  /* Getting only real part from the output and Converting to DCT-IV */

  /* pbuff initialized to input buffer. */
  pbuff = pInlineBuffer;

  /* pS1 initialized to pState */
  pS1 = pState;

  /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */
  in = *pS1++ * (float32_t) 0.5;
  /* input buffer acts as inplace, so output values are stored in the input itself. */
  *pbuff++ = in;

  /* pState pointer is incremented twice as the real values are located alternatively in the array */
  pS1++;

  /* Initializing the loop counter */
  i = ((uint32_t) S->N - 1u);

  do
  {
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
    in = *pS1++ - in;
    *pbuff++ = in;
    /* points to the next real value */
    pS1++;


    /* Decrement the loop counter */
    i--;
  } while(i > 0u);


        /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/

  /* Initializing the loop counter */
  i = (uint32_t) S->N;

  /* pbuff initialized to the pInlineBuffer(now contains the output values) */
  pbuff = pInlineBuffer;

  do
  {
    /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */
    in = *pbuff;
    *pbuff++ = in * S->normalize;

    /* Decrement the loop counter */
    i--;
  } while(i > 0u);

#endif /* #ifndef ARM_MATH_CM0 */

}

/**   
   * @} end of DCT4_IDCT4 group   
   */