arm_fir_f32.c

00001 /* ----------------------------------------------------------------------  
00002 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.  
00003 *  
00004 * $Date:        17. January 2013
00005 * $Revision:    V1.4.1
00006 *  
00007 * Project:      CMSIS DSP Library  
00008 * Title:        arm_fir_f32.c  
00009 *  
00010 * Description:  Floating-point FIR filter processing function.  
00011 *  
00012 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
00013 *  
00014 * Redistribution and use in source and binary forms, with or without 
00015 * modification, are permitted provided that the following conditions
00016 * are met:
00017 *   - Redistributions of source code must retain the above copyright
00018 *     notice, this list of conditions and the following disclaimer.
00019 *   - Redistributions in binary form must reproduce the above copyright
00020 *     notice, this list of conditions and the following disclaimer in
00021 *     the documentation and/or other materials provided with the 
00022 *     distribution.
00023 *   - Neither the name of ARM LIMITED nor the names of its contributors
00024 *     may be used to endorse or promote products derived from this
00025 *     software without specific prior written permission.
00026 *
00027 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
00028 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
00029 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
00030 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 
00031 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
00032 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
00033 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
00034 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
00035 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00036 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
00037 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
00038 * POSSIBILITY OF SUCH DAMAGE. 
00039 * -------------------------------------------------------------------- */
00040 
00041 #include "arm_math.h"
00042 
00043 /**  
00044 * @ingroup groupFilters  
00045 */
00046 
00047 /**  
00048 * @defgroup FIR Finite Impulse Response (FIR) Filters  
00049 *  
00050 * This set of functions implements Finite Impulse Response (FIR) filters  
00051 * for Q7, Q15, Q31, and floating-point data types.  Fast versions of Q15 and Q31 are also provided.  
00052 * The functions operate on blocks of input and output data and each call to the function processes  
00053 * <code>blockSize</code> samples through the filter.  <code>pSrc</code> and  
00054 * <code>pDst</code> points to input and output arrays containing <code>blockSize</code> values.  
00055 *  
00056 * \par Algorithm:  
00057 * The FIR filter algorithm is based upon a sequence of multiply-accumulate (MAC) operations.  
00058 * Each filter coefficient <code>b[n]</code> is multiplied by a state variable which equals a previous input sample <code>x[n]</code>.  
00059 * <pre>  
00060 *    y[n] = b[0] * x[n] + b[1] * x[n-1] + b[2] * x[n-2] + ...+ b[numTaps-1] * x[n-numTaps+1]  
00061 * </pre>  
00062 * \par  
00063 * \image html FIR.gif "Finite Impulse Response filter"  
00064 * \par  
00065 * <code>pCoeffs</code> points to a coefficient array of size <code>numTaps</code>.  
00066 * Coefficients are stored in time reversed order.  
00067 * \par  
00068 * <pre>  
00069 *    {b[numTaps-1], b[numTaps-2], b[N-2], ..., b[1], b[0]}  
00070 * </pre>  
00071 * \par  
00072 * <code>pState</code> points to a state array of size <code>numTaps + blockSize - 1</code>.  
00073 * Samples in the state buffer are stored in the following order.  
00074 * \par  
00075 * <pre>  
00076 *    {x[n-numTaps+1], x[n-numTaps], x[n-numTaps-1], x[n-numTaps-2]....x[0], x[1], ..., x[blockSize-1]}  
00077 * </pre>  
00078 * \par  
00079 * Note that the length of the state buffer exceeds the length of the coefficient array by <code>blockSize-1</code>.  
00080 * The increased state buffer length allows circular addressing, which is traditionally used in the FIR filters,  
00081 * to be avoided and yields a significant speed improvement.  
00082 * The state variables are updated after each block of data is processed; the coefficients are untouched.  
00083 * \par Instance Structure  
00084 * The coefficients and state variables for a filter are stored together in an instance data structure.  
00085 * A separate instance structure must be defined for each filter.  
00086 * Coefficient arrays may be shared among several instances while state variable arrays cannot be shared.  
00087 * There are separate instance structure declarations for each of the 4 supported data types.  
00088 *  
00089 * \par Initialization Functions  
00090 * There is also an associated initialization function for each data type.  
00091 * The initialization function performs the following operations:  
00092 * - Sets the values of the internal structure fields.  
00093 * - Zeros out the values in the state buffer.  
00094 * To do this manually without calling the init function, assign the follow subfields of the instance structure:
00095 * numTaps, pCoeffs, pState. Also set all of the values in pState to zero. 
00096 *  
00097 * \par  
00098 * Use of the initialization function is optional.  
00099 * However, if the initialization function is used, then the instance structure cannot be placed into a const data section.  
00100 * To place an instance structure into a const data section, the instance structure must be manually initialized.  
00101 * Set the values in the state buffer to zeros before static initialization.  
00102 * The code below statically initializes each of the 4 different data type filter instance structures  
00103 * <pre>  
00104 *arm_fir_instance_f32 S = {numTaps, pState, pCoeffs};  
00105 *arm_fir_instance_q31 S = {numTaps, pState, pCoeffs};  
00106 *arm_fir_instance_q15 S = {numTaps, pState, pCoeffs};  
00107 *arm_fir_instance_q7 S =  {numTaps, pState, pCoeffs};  
00108 * </pre>  
00109 *  
00110 * where <code>numTaps</code> is the number of filter coefficients in the filter; <code>pState</code> is the address of the state buffer;  
00111 * <code>pCoeffs</code> is the address of the coefficient buffer.  
00112 *  
00113 * \par Fixed-Point Behavior  
00114 * Care must be taken when using the fixed-point versions of the FIR filter functions.  
00115 * In particular, the overflow and saturation behavior of the accumulator used in each function must be considered.  
00116 * Refer to the function specific documentation below for usage guidelines.  
00117 */
00118 
00119 /**  
00120 * @addtogroup FIR  
00121 * @{  
00122 */
00123 
00124 /**  
00125 *  
00126 * @param[in]  *S points to an instance of the floating-point FIR filter structure.  
00127 * @param[in]  *pSrc points to the block of input data.  
00128 * @param[out] *pDst points to the block of output data.  
00129 * @param[in]  blockSize number of samples to process per call.  
00130 * @return     none.  
00131 *  
00132 */
00133 
00134 #ifndef ARM_MATH_CM0_FAMILY
00135 
00136 /* Run the below code for Cortex-M4 and Cortex-M3 */
00137 
00138 void arm_fir_f32(
00139 const arm_fir_instance_f32 * S,
00140 float32_t * pSrc,
00141 float32_t * pDst,
00142 uint32_t blockSize)
00143 {
00144    float32_t *pState = S->pState;                 /* State pointer */
00145    float32_t *pCoeffs = S->pCoeffs;               /* Coefficient pointer */
00146    float32_t *pStateCurnt;                        /* Points to the current sample of the state */
00147    float32_t *px, *pb;                            /* Temporary pointers for state and coefficient buffers */
00148    float32_t acc0, acc1, acc2, acc3, acc4, acc5, acc6, acc7;     /* Accumulators */
00149    float32_t x0, x1, x2, x3, x4, x5, x6, x7, c0;  /* Temporary variables to hold state and coefficient values */
00150    uint32_t numTaps = S->numTaps;                 /* Number of filter coefficients in the filter */
00151    uint32_t i, tapCnt, blkCnt;                    /* Loop counters */
00152    float32_t p0,p1,p2,p3,p4,p5,p6,p7;             /* Temporary product values */
00153 
00154    /* S->pState points to state array which contains previous frame (numTaps - 1) samples */
00155    /* pStateCurnt points to the location where the new input data should be written */
00156    pStateCurnt = &(S->pState[(numTaps - 1u)]);
00157 
00158    /* Apply loop unrolling and compute 8 output values simultaneously.  
00159     * The variables acc0 ... acc7 hold output values that are being computed:  
00160     *  
00161     *    acc0 =  b[numTaps-1] * x[n-numTaps-1] + b[numTaps-2] * x[n-numTaps-2] + b[numTaps-3] * x[n-numTaps-3] +...+ b[0] * x[0]  
00162     *    acc1 =  b[numTaps-1] * x[n-numTaps] +   b[numTaps-2] * x[n-numTaps-1] + b[numTaps-3] * x[n-numTaps-2] +...+ b[0] * x[1]  
00163     *    acc2 =  b[numTaps-1] * x[n-numTaps+1] + b[numTaps-2] * x[n-numTaps] +   b[numTaps-3] * x[n-numTaps-1] +...+ b[0] * x[2]  
00164     *    acc3 =  b[numTaps-1] * x[n-numTaps+2] + b[numTaps-2] * x[n-numTaps+1] + b[numTaps-3] * x[n-numTaps]   +...+ b[0] * x[3]  
00165     */
00166    blkCnt = blockSize >> 3;
00167 
00168    /* First part of the processing with loop unrolling.  Compute 8 outputs at a time.  
00169    ** a second loop below computes the remaining 1 to 7 samples. */
00170    while(blkCnt > 0u)
00171    {
00172       /* Copy four new input samples into the state buffer */
00173       *pStateCurnt++ = *pSrc++;
00174       *pStateCurnt++ = *pSrc++;
00175       *pStateCurnt++ = *pSrc++;
00176       *pStateCurnt++ = *pSrc++;
00177 
00178       /* Set all accumulators to zero */
00179       acc0 = 0.0f;
00180       acc1 = 0.0f;
00181       acc2 = 0.0f;
00182       acc3 = 0.0f;
00183       acc4 = 0.0f;
00184       acc5 = 0.0f;
00185       acc6 = 0.0f;
00186       acc7 = 0.0f;      
00187 
00188       /* Initialize state pointer */
00189       px = pState;
00190 
00191       /* Initialize coeff pointer */
00192       pb = (pCoeffs);       
00193    
00194       /* This is separated from the others to avoid 
00195        * a call to __aeabi_memmove which would be slower
00196        */
00197       *pStateCurnt++ = *pSrc++;
00198       *pStateCurnt++ = *pSrc++;
00199       *pStateCurnt++ = *pSrc++;
00200       *pStateCurnt++ = *pSrc++;
00201 
00202       /* Read the first seven samples from the state buffer:  x[n-numTaps], x[n-numTaps-1], x[n-numTaps-2] */
00203       x0 = *px++;
00204       x1 = *px++;
00205       x2 = *px++;
00206       x3 = *px++;
00207       x4 = *px++;
00208       x5 = *px++;
00209       x6 = *px++;
00210 
00211       /* Loop unrolling.  Process 8 taps at a time. */
00212       tapCnt = numTaps >> 3u;
00213       
00214       /* Loop over the number of taps.  Unroll by a factor of 8.  
00215        ** Repeat until we've computed numTaps-8 coefficients. */
00216       while(tapCnt > 0u)
00217       {
00218          /* Read the b[numTaps-1] coefficient */
00219          c0 = *(pb++);
00220 
00221          /* Read x[n-numTaps-3] sample */
00222          x7 = *(px++);
00223 
00224          /* acc0 +=  b[numTaps-1] * x[n-numTaps] */
00225          p0 = x0 * c0;
00226 
00227          /* acc1 +=  b[numTaps-1] * x[n-numTaps-1] */
00228          p1 = x1 * c0;
00229 
00230          /* acc2 +=  b[numTaps-1] * x[n-numTaps-2] */
00231          p2 = x2 * c0;
00232 
00233          /* acc3 +=  b[numTaps-1] * x[n-numTaps-3] */
00234          p3 = x3 * c0;
00235 
00236          /* acc4 +=  b[numTaps-1] * x[n-numTaps-4] */
00237          p4 = x4 * c0;
00238 
00239          /* acc1 +=  b[numTaps-1] * x[n-numTaps-5] */
00240          p5 = x5 * c0;
00241 
00242          /* acc2 +=  b[numTaps-1] * x[n-numTaps-6] */
00243          p6 = x6 * c0;
00244 
00245          /* acc3 +=  b[numTaps-1] * x[n-numTaps-7] */
00246          p7 = x7 * c0;
00247          
00248          /* Read the b[numTaps-2] coefficient */
00249          c0 = *(pb++);
00250 
00251          /* Read x[n-numTaps-4] sample */
00252          x0 = *(px++);
00253          
00254          acc0 += p0;
00255          acc1 += p1;
00256          acc2 += p2;
00257          acc3 += p3;
00258          acc4 += p4;
00259          acc5 += p5;
00260          acc6 += p6;
00261          acc7 += p7;
00262 
00263 
00264          /* Perform the multiply-accumulate */
00265          p0 = x1 * c0;
00266          p1 = x2 * c0;   
00267          p2 = x3 * c0;   
00268          p3 = x4 * c0;   
00269          p4 = x5 * c0;   
00270          p5 = x6 * c0;   
00271          p6 = x7 * c0;   
00272          p7 = x0 * c0;   
00273          
00274          /* Read the b[numTaps-3] coefficient */
00275          c0 = *(pb++);
00276 
00277          /* Read x[n-numTaps-5] sample */
00278          x1 = *(px++);
00279          
00280          acc0 += p0;
00281          acc1 += p1;
00282          acc2 += p2;
00283          acc3 += p3;
00284          acc4 += p4;
00285          acc5 += p5;
00286          acc6 += p6;
00287          acc7 += p7;
00288 
00289          /* Perform the multiply-accumulates */      
00290          p0 = x2 * c0;
00291          p1 = x3 * c0;   
00292          p2 = x4 * c0;   
00293          p3 = x5 * c0;   
00294          p4 = x6 * c0;   
00295          p5 = x7 * c0;   
00296          p6 = x0 * c0;   
00297          p7 = x1 * c0;   
00298 
00299          /* Read the b[numTaps-4] coefficient */
00300          c0 = *(pb++);
00301 
00302          /* Read x[n-numTaps-6] sample */
00303          x2 = *(px++);
00304          
00305          acc0 += p0;
00306          acc1 += p1;
00307          acc2 += p2;
00308          acc3 += p3;
00309          acc4 += p4;
00310          acc5 += p5;
00311          acc6 += p6;
00312          acc7 += p7;
00313 
00314          /* Perform the multiply-accumulates */      
00315          p0 = x3 * c0;
00316          p1 = x4 * c0;   
00317          p2 = x5 * c0;   
00318          p3 = x6 * c0;   
00319          p4 = x7 * c0;   
00320          p5 = x0 * c0;   
00321          p6 = x1 * c0;   
00322          p7 = x2 * c0;   
00323 
00324          /* Read the b[numTaps-4] coefficient */
00325          c0 = *(pb++);
00326 
00327          /* Read x[n-numTaps-6] sample */
00328          x3 = *(px++);
00329          
00330          acc0 += p0;
00331          acc1 += p1;
00332          acc2 += p2;
00333          acc3 += p3;
00334          acc4 += p4;
00335          acc5 += p5;
00336          acc6 += p6;
00337          acc7 += p7;
00338 
00339          /* Perform the multiply-accumulates */      
00340          p0 = x4 * c0;
00341          p1 = x5 * c0;   
00342          p2 = x6 * c0;   
00343          p3 = x7 * c0;   
00344          p4 = x0 * c0;   
00345          p5 = x1 * c0;   
00346          p6 = x2 * c0;   
00347          p7 = x3 * c0;   
00348 
00349          /* Read the b[numTaps-4] coefficient */
00350          c0 = *(pb++);
00351 
00352          /* Read x[n-numTaps-6] sample */
00353          x4 = *(px++);
00354          
00355          acc0 += p0;
00356          acc1 += p1;
00357          acc2 += p2;
00358          acc3 += p3;
00359          acc4 += p4;
00360          acc5 += p5;
00361          acc6 += p6;
00362          acc7 += p7;
00363 
00364          /* Perform the multiply-accumulates */      
00365          p0 = x5 * c0;
00366          p1 = x6 * c0;   
00367          p2 = x7 * c0;   
00368          p3 = x0 * c0;   
00369          p4 = x1 * c0;   
00370          p5 = x2 * c0;   
00371          p6 = x3 * c0;   
00372          p7 = x4 * c0;   
00373 
00374          /* Read the b[numTaps-4] coefficient */
00375          c0 = *(pb++);
00376 
00377          /* Read x[n-numTaps-6] sample */
00378          x5 = *(px++);
00379          
00380          acc0 += p0;
00381          acc1 += p1;
00382          acc2 += p2;
00383          acc3 += p3;
00384          acc4 += p4;
00385          acc5 += p5;
00386          acc6 += p6;
00387          acc7 += p7;
00388 
00389          /* Perform the multiply-accumulates */      
00390          p0 = x6 * c0;
00391          p1 = x7 * c0;   
00392          p2 = x0 * c0;   
00393          p3 = x1 * c0;   
00394          p4 = x2 * c0;   
00395          p5 = x3 * c0;   
00396          p6 = x4 * c0;   
00397          p7 = x5 * c0;   
00398 
00399          /* Read the b[numTaps-4] coefficient */
00400          c0 = *(pb++);
00401 
00402          /* Read x[n-numTaps-6] sample */
00403          x6 = *(px++);
00404          
00405          acc0 += p0;
00406          acc1 += p1;
00407          acc2 += p2;
00408          acc3 += p3;
00409          acc4 += p4;
00410          acc5 += p5;
00411          acc6 += p6;
00412          acc7 += p7;
00413 
00414          /* Perform the multiply-accumulates */      
00415          p0 = x7 * c0;
00416          p1 = x0 * c0;   
00417          p2 = x1 * c0;   
00418          p3 = x2 * c0;   
00419          p4 = x3 * c0;   
00420          p5 = x4 * c0;   
00421          p6 = x5 * c0;   
00422          p7 = x6 * c0;   
00423 
00424          tapCnt--;
00425          
00426          acc0 += p0;
00427          acc1 += p1;
00428          acc2 += p2;
00429          acc3 += p3;
00430          acc4 += p4;
00431          acc5 += p5;
00432          acc6 += p6;
00433          acc7 += p7;
00434       }
00435 
00436       /* If the filter length is not a multiple of 8, compute the remaining filter taps */
00437       tapCnt = numTaps % 0x8u;
00438 
00439       while(tapCnt > 0u)
00440       {
00441          /* Read coefficients */
00442          c0 = *(pb++);
00443 
00444          /* Fetch 1 state variable */
00445          x7 = *(px++);
00446 
00447          /* Perform the multiply-accumulates */      
00448          p0 = x0 * c0;
00449          p1 = x1 * c0;   
00450          p2 = x2 * c0;   
00451          p3 = x3 * c0;   
00452          p4 = x4 * c0;   
00453          p5 = x5 * c0;   
00454          p6 = x6 * c0;   
00455          p7 = x7 * c0;   
00456 
00457          /* Reuse the present sample states for next sample */
00458          x0 = x1;
00459          x1 = x2;
00460          x2 = x3;
00461          x3 = x4;
00462          x4 = x5;
00463          x5 = x6;
00464          x6 = x7;
00465          
00466          acc0 += p0;
00467          acc1 += p1;
00468          acc2 += p2;
00469          acc3 += p3;
00470          acc4 += p4;
00471          acc5 += p5;
00472          acc6 += p6;
00473          acc7 += p7;
00474 
00475          /* Decrement the loop counter */
00476          tapCnt--;
00477       }
00478 
00479       /* Advance the state pointer by 8 to process the next group of 8 samples */
00480       pState = pState + 8;
00481 
00482       /* The results in the 8 accumulators, store in the destination buffer. */
00483       *pDst++ = acc0;
00484       *pDst++ = acc1;
00485       *pDst++ = acc2;
00486       *pDst++ = acc3;
00487       *pDst++ = acc4;
00488       *pDst++ = acc5;
00489       *pDst++ = acc6;
00490       *pDst++ = acc7;
00491 
00492       blkCnt--;
00493    }
00494 
00495    /* If the blockSize is not a multiple of 8, compute any remaining output samples here.  
00496    ** No loop unrolling is used. */
00497    blkCnt = blockSize % 0x8u;
00498 
00499    while(blkCnt > 0u)
00500    {
00501       /* Copy one sample at a time into state buffer */
00502       *pStateCurnt++ = *pSrc++;
00503 
00504       /* Set the accumulator to zero */
00505       acc0 = 0.0f;
00506 
00507       /* Initialize state pointer */
00508       px = pState;
00509 
00510       /* Initialize Coefficient pointer */
00511       pb = (pCoeffs);
00512 
00513       i = numTaps;
00514 
00515       /* Perform the multiply-accumulates */
00516       do
00517       {
00518          acc0 += *px++ * *pb++;
00519          i--;
00520 
00521       } while(i > 0u);
00522 
00523       /* The result is store in the destination buffer. */
00524       *pDst++ = acc0;
00525 
00526       /* Advance state pointer by 1 for the next sample */
00527       pState = pState + 1;
00528 
00529       blkCnt--;
00530    }
00531 
00532    /* Processing is complete.  
00533    ** Now copy the last numTaps - 1 samples to the start of the state buffer.  
00534    ** This prepares the state buffer for the next function call. */
00535 
00536    /* Points to the start of the state buffer */
00537    pStateCurnt = S->pState;
00538 
00539    tapCnt = (numTaps - 1u) >> 2u;
00540 
00541    /* copy data */
00542    while(tapCnt > 0u)
00543    {
00544       *pStateCurnt++ = *pState++;
00545       *pStateCurnt++ = *pState++;
00546       *pStateCurnt++ = *pState++;
00547       *pStateCurnt++ = *pState++;
00548 
00549       /* Decrement the loop counter */
00550       tapCnt--;
00551    }
00552 
00553    /* Calculate remaining number of copies */
00554    tapCnt = (numTaps - 1u) % 0x4u;
00555 
00556    /* Copy the remaining q31_t data */
00557    while(tapCnt > 0u)
00558    {
00559       *pStateCurnt++ = *pState++;
00560 
00561       /* Decrement the loop counter */
00562       tapCnt--;
00563    }
00564 }
00565 
00566 #else
00567 
00568 void arm_fir_f32(
00569 const arm_fir_instance_f32 * S,
00570 float32_t * pSrc,
00571 float32_t * pDst,
00572 uint32_t blockSize)
00573 {
00574    float32_t *pState = S->pState;                 /* State pointer */
00575    float32_t *pCoeffs = S->pCoeffs;               /* Coefficient pointer */
00576    float32_t *pStateCurnt;                        /* Points to the current sample of the state */
00577    float32_t *px, *pb;                            /* Temporary pointers for state and coefficient buffers */
00578    uint32_t numTaps = S->numTaps;                 /* Number of filter coefficients in the filter */
00579    uint32_t i, tapCnt, blkCnt;                    /* Loop counters */
00580 
00581    /* Run the below code for Cortex-M0 */
00582 
00583    float32_t acc;
00584 
00585    /* S->pState points to state array which contains previous frame (numTaps - 1) samples */
00586    /* pStateCurnt points to the location where the new input data should be written */
00587    pStateCurnt = &(S->pState[(numTaps - 1u)]);
00588 
00589    /* Initialize blkCnt with blockSize */
00590    blkCnt = blockSize;
00591 
00592    while(blkCnt > 0u)
00593    {
00594       /* Copy one sample at a time into state buffer */
00595       *pStateCurnt++ = *pSrc++;
00596 
00597       /* Set the accumulator to zero */
00598       acc = 0.0f;
00599 
00600       /* Initialize state pointer */
00601       px = pState;
00602 
00603       /* Initialize Coefficient pointer */
00604       pb = pCoeffs;
00605 
00606       i = numTaps;
00607 
00608       /* Perform the multiply-accumulates */
00609       do
00610       {
00611          /* acc =  b[numTaps-1] * x[n-numTaps-1] + b[numTaps-2] * x[n-numTaps-2] + b[numTaps-3] * x[n-numTaps-3] +...+ b[0] * x[0] */
00612          acc += *px++ * *pb++;
00613          i--;
00614 
00615       } while(i > 0u);
00616 
00617       /* The result is store in the destination buffer. */
00618       *pDst++ = acc;
00619 
00620       /* Advance state pointer by 1 for the next sample */
00621       pState = pState + 1;
00622 
00623       blkCnt--;
00624    }
00625 
00626    /* Processing is complete.         
00627    ** Now copy the last numTaps - 1 samples to the starting of the state buffer.       
00628    ** This prepares the state buffer for the next function call. */
00629 
00630    /* Points to the start of the state buffer */
00631    pStateCurnt = S->pState;
00632 
00633    /* Copy numTaps number of values */
00634    tapCnt = numTaps - 1u;
00635 
00636    /* Copy data */
00637    while(tapCnt > 0u)
00638    {
00639       *pStateCurnt++ = *pState++;
00640 
00641       /* Decrement the loop counter */
00642       tapCnt--;
00643    }
00644 
00645 }
00646 
00647 #endif /*   #ifndef ARM_MATH_CM0_FAMILY */
00648 
00649 /**  
00650 * @} end of FIR group  
00651 */