arm_fir_lattice_f32.c

Go to the documentation of this file.

/* ----------------------------------------------------------------------    
* Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
*    
* $Date:        19. March 2015
* $Revision:    V.1.4.5
*    
* Project:      CMSIS DSP Library    
* Title:        arm_fir_lattice_f32.c    
*    
* Description:  Processing function for the floating-point FIR Lattice filter.    
*    
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
*  
* Redistribution and use in source and binary forms, with or without 
* modification, are permitted provided that the following conditions
* are met:
*   - Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*   - Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in
*     the documentation and/or other materials provided with the 
*     distribution.
*   - Neither the name of ARM LIMITED nor the names of its contributors
*     may be used to endorse or promote products derived from this
*     software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.   
* -------------------------------------------------------------------- */

#include "arm_math.h"

void arm_fir_lattice_f32(
  const arm_fir_lattice_instance_f32 * S,
  float32_t * pSrc,
  float32_t * pDst,
  uint32_t blockSize)
{
  float32_t *pState;                             /* State pointer */
  float32_t *pCoeffs = S->pCoeffs;               /* Coefficient pointer */
  float32_t *px;                                 /* temporary state pointer */
  float32_t *pk;                                 /* temporary coefficient pointer */


#ifndef ARM_MATH_CM0_FAMILY

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

  float32_t fcurr1, fnext1, gcurr1, gnext1;      /* temporary variables for first sample in loop unrolling */
  float32_t fcurr2, fnext2, gnext2;              /* temporary variables for second sample in loop unrolling */
  float32_t fcurr3, fnext3, gnext3;              /* temporary variables for third sample in loop unrolling */
  float32_t fcurr4, fnext4, gnext4;              /* temporary variables for fourth sample in loop unrolling */
  uint32_t numStages = S->numStages;             /* Number of stages in the filter */
  uint32_t blkCnt, stageCnt;                     /* temporary variables for counts */

  gcurr1 = 0.0f;
  pState = &S->pState[0];

  blkCnt = blockSize >> 2;

  /* 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. */
  while(blkCnt > 0u)
  {

    /* Read two samples from input buffer */
    /* f0(n) = x(n) */
    fcurr1 = *pSrc++;
    fcurr2 = *pSrc++;

    /* Initialize coeff pointer */
    pk = (pCoeffs);

    /* Initialize state pointer */
    px = pState;

    /* Read g0(n-1) from state */
    gcurr1 = *px;

    /* Process first sample for first tap */
    /* f1(n) = f0(n) +  K1 * g0(n-1) */
    fnext1 = fcurr1 + ((*pk) * gcurr1);
    /* g1(n) = f0(n) * K1  +  g0(n-1) */
    gnext1 = (fcurr1 * (*pk)) + gcurr1;

    /* Process second sample for first tap */
    /* for sample 2 processing */
    fnext2 = fcurr2 + ((*pk) * fcurr1);
    gnext2 = (fcurr2 * (*pk)) + fcurr1;

    /* Read next two samples from input buffer */
    /* f0(n+2) = x(n+2) */
    fcurr3 = *pSrc++;
    fcurr4 = *pSrc++;

    /* Copy only last input samples into the state buffer    
       which will be used for next four samples processing */
    *px++ = fcurr4;

    /* Process third sample for first tap */
    fnext3 = fcurr3 + ((*pk) * fcurr2);
    gnext3 = (fcurr3 * (*pk)) + fcurr2;

    /* Process fourth sample for first tap */
    fnext4 = fcurr4 + ((*pk) * fcurr3);
    gnext4 = (fcurr4 * (*pk++)) + fcurr3;

    /* Update of f values for next coefficient set processing */
    fcurr1 = fnext1;
    fcurr2 = fnext2;
    fcurr3 = fnext3;
    fcurr4 = fnext4;

    /* Loop unrolling.  Process 4 taps at a time . */
    stageCnt = (numStages - 1u) >> 2u;

    /* Loop over the number of taps.  Unroll by a factor of 4.    
     ** Repeat until we've computed numStages-3 coefficients. */

    /* Process 2nd, 3rd, 4th and 5th taps ... here */
    while(stageCnt > 0u)
    {
      /* Read g1(n-1), g3(n-1) .... from state */
      gcurr1 = *px;

      /* save g1(n) in state buffer */
      *px++ = gnext4;

      /* Process first sample for 2nd, 6th .. tap */
      /* Sample processing for K2, K6.... */
      /* f2(n) = f1(n) +  K2 * g1(n-1) */
      fnext1 = fcurr1 + ((*pk) * gcurr1);
      /* Process second sample for 2nd, 6th .. tap */
      /* for sample 2 processing */
      fnext2 = fcurr2 + ((*pk) * gnext1);
      /* Process third sample for 2nd, 6th .. tap */
      fnext3 = fcurr3 + ((*pk) * gnext2);
      /* Process fourth sample for 2nd, 6th .. tap */
      fnext4 = fcurr4 + ((*pk) * gnext3);

      /* g2(n) = f1(n) * K2  +  g1(n-1) */
      /* Calculation of state values for next stage */
      gnext4 = (fcurr4 * (*pk)) + gnext3;
      gnext3 = (fcurr3 * (*pk)) + gnext2;
      gnext2 = (fcurr2 * (*pk)) + gnext1;
      gnext1 = (fcurr1 * (*pk++)) + gcurr1;


      /* Read g2(n-1), g4(n-1) .... from state */
      gcurr1 = *px;

      /* save g2(n) in state buffer */
      *px++ = gnext4;

      /* Sample processing for K3, K7.... */
      /* Process first sample for 3rd, 7th .. tap */
      /* f3(n) = f2(n) +  K3 * g2(n-1) */
      fcurr1 = fnext1 + ((*pk) * gcurr1);
      /* Process second sample for 3rd, 7th .. tap */
      fcurr2 = fnext2 + ((*pk) * gnext1);
      /* Process third sample for 3rd, 7th .. tap */
      fcurr3 = fnext3 + ((*pk) * gnext2);
      /* Process fourth sample for 3rd, 7th .. tap */
      fcurr4 = fnext4 + ((*pk) * gnext3);

      /* Calculation of state values for next stage */
      /* g3(n) = f2(n) * K3  +  g2(n-1) */
      gnext4 = (fnext4 * (*pk)) + gnext3;
      gnext3 = (fnext3 * (*pk)) + gnext2;
      gnext2 = (fnext2 * (*pk)) + gnext1;
      gnext1 = (fnext1 * (*pk++)) + gcurr1;


      /* Read g1(n-1), g3(n-1) .... from state */
      gcurr1 = *px;

      /* save g3(n) in state buffer */
      *px++ = gnext4;

      /* Sample processing for K4, K8.... */
      /* Process first sample for 4th, 8th .. tap */
      /* f4(n) = f3(n) +  K4 * g3(n-1) */
      fnext1 = fcurr1 + ((*pk) * gcurr1);
      /* Process second sample for 4th, 8th .. tap */
      /* for sample 2 processing */
      fnext2 = fcurr2 + ((*pk) * gnext1);
      /* Process third sample for 4th, 8th .. tap */
      fnext3 = fcurr3 + ((*pk) * gnext2);
      /* Process fourth sample for 4th, 8th .. tap */
      fnext4 = fcurr4 + ((*pk) * gnext3);

      /* g4(n) = f3(n) * K4  +  g3(n-1) */
      /* Calculation of state values for next stage */
      gnext4 = (fcurr4 * (*pk)) + gnext3;
      gnext3 = (fcurr3 * (*pk)) + gnext2;
      gnext2 = (fcurr2 * (*pk)) + gnext1;
      gnext1 = (fcurr1 * (*pk++)) + gcurr1;

      /* Read g2(n-1), g4(n-1) .... from state */
      gcurr1 = *px;

      /* save g4(n) in state buffer */
      *px++ = gnext4;

      /* Sample processing for K5, K9.... */
      /* Process first sample for 5th, 9th .. tap */
      /* f5(n) = f4(n) +  K5 * g4(n-1) */
      fcurr1 = fnext1 + ((*pk) * gcurr1);
      /* Process second sample for 5th, 9th .. tap */
      fcurr2 = fnext2 + ((*pk) * gnext1);
      /* Process third sample for 5th, 9th .. tap */
      fcurr3 = fnext3 + ((*pk) * gnext2);
      /* Process fourth sample for 5th, 9th .. tap */
      fcurr4 = fnext4 + ((*pk) * gnext3);

      /* Calculation of state values for next stage */
      /* g5(n) = f4(n) * K5  +  g4(n-1) */
      gnext4 = (fnext4 * (*pk)) + gnext3;
      gnext3 = (fnext3 * (*pk)) + gnext2;
      gnext2 = (fnext2 * (*pk)) + gnext1;
      gnext1 = (fnext1 * (*pk++)) + gcurr1;

      stageCnt--;
    }

    /* If the (filter length -1) is not a multiple of 4, compute the remaining filter taps */
    stageCnt = (numStages - 1u) % 0x4u;

    while(stageCnt > 0u)
    {
      gcurr1 = *px;

      /* save g value in state buffer */
      *px++ = gnext4;

      /* Process four samples for last three taps here */
      fnext1 = fcurr1 + ((*pk) * gcurr1);
      fnext2 = fcurr2 + ((*pk) * gnext1);
      fnext3 = fcurr3 + ((*pk) * gnext2);
      fnext4 = fcurr4 + ((*pk) * gnext3);

      /* g1(n) = f0(n) * K1  +  g0(n-1) */
      gnext4 = (fcurr4 * (*pk)) + gnext3;
      gnext3 = (fcurr3 * (*pk)) + gnext2;
      gnext2 = (fcurr2 * (*pk)) + gnext1;
      gnext1 = (fcurr1 * (*pk++)) + gcurr1;

      /* Update of f values for next coefficient set processing */
      fcurr1 = fnext1;
      fcurr2 = fnext2;
      fcurr3 = fnext3;
      fcurr4 = fnext4;

      stageCnt--;

    }

    /* The results in the 4 accumulators, store in the destination buffer. */
    /* y(n) = fN(n) */
    *pDst++ = fcurr1;
    *pDst++ = fcurr2;
    *pDst++ = fcurr3;
    *pDst++ = fcurr4;

    blkCnt--;
  }

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

  while(blkCnt > 0u)
  {
    /* f0(n) = x(n) */
    fcurr1 = *pSrc++;

    /* Initialize coeff pointer */
    pk = (pCoeffs);

    /* Initialize state pointer */
    px = pState;

    /* read g2(n) from state buffer */
    gcurr1 = *px;

    /* for sample 1 processing */
    /* f1(n) = f0(n) +  K1 * g0(n-1) */
    fnext1 = fcurr1 + ((*pk) * gcurr1);
    /* g1(n) = f0(n) * K1  +  g0(n-1) */
    gnext1 = (fcurr1 * (*pk++)) + gcurr1;

    /* save g1(n) in state buffer */
    *px++ = fcurr1;

    /* f1(n) is saved in fcurr1    
       for next stage processing */
    fcurr1 = fnext1;

    stageCnt = (numStages - 1u);

    /* stage loop */
    while(stageCnt > 0u)
    {
      /* read g2(n) from state buffer */
      gcurr1 = *px;

      /* save g1(n) in state buffer */
      *px++ = gnext1;

      /* Sample processing for K2, K3.... */
      /* f2(n) = f1(n) +  K2 * g1(n-1) */
      fnext1 = fcurr1 + ((*pk) * gcurr1);
      /* g2(n) = f1(n) * K2  +  g1(n-1) */
      gnext1 = (fcurr1 * (*pk++)) + gcurr1;

      /* f1(n) is saved in fcurr1    
         for next stage processing */
      fcurr1 = fnext1;

      stageCnt--;

    }

    /* y(n) = fN(n) */
    *pDst++ = fcurr1;

    blkCnt--;

  }

#else

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

  float32_t fcurr, fnext, gcurr, gnext;          /* temporary variables */
  uint32_t numStages = S->numStages;             /* Length of the filter */
  uint32_t blkCnt, stageCnt;                     /* temporary variables for counts */

  pState = &S->pState[0];

  blkCnt = blockSize;

  while(blkCnt > 0u)
  {
    /* f0(n) = x(n) */
    fcurr = *pSrc++;

    /* Initialize coeff pointer */
    pk = pCoeffs;

    /* Initialize state pointer */
    px = pState;

    /* read g0(n-1) from state buffer */
    gcurr = *px;

    /* for sample 1 processing */
    /* f1(n) = f0(n) +  K1 * g0(n-1) */
    fnext = fcurr + ((*pk) * gcurr);
    /* g1(n) = f0(n) * K1  +  g0(n-1) */
    gnext = (fcurr * (*pk++)) + gcurr;

    /* save f0(n) in state buffer */
    *px++ = fcurr;

    /* f1(n) is saved in fcurr            
       for next stage processing */
    fcurr = fnext;

    stageCnt = (numStages - 1u);

    /* stage loop */
    while(stageCnt > 0u)
    {
      /* read g2(n) from state buffer */
      gcurr = *px;

      /* save g1(n) in state buffer */
      *px++ = gnext;

      /* Sample processing for K2, K3.... */
      /* f2(n) = f1(n) +  K2 * g1(n-1) */
      fnext = fcurr + ((*pk) * gcurr);
      /* g2(n) = f1(n) * K2  +  g1(n-1) */
      gnext = (fcurr * (*pk++)) + gcurr;

      /* f1(n) is saved in fcurr1            
         for next stage processing */
      fcurr = fnext;

      stageCnt--;

    }

    /* y(n) = fN(n) */
    *pDst++ = fcurr;

    blkCnt--;

  }

#endif /*   #ifndef ARM_MATH_CM0_FAMILY */

}