arm_fir_decimate_q31.c

00001 /* ----------------------------------------------------------------------  
00002 * Copyright (C) 2010 ARM Limited. All rights reserved.  
00003 *  
00004 * $Date:        29. November 2010  
00005 * $Revision:    V1.0.3  
00006 *  
00007 * Project:      CMSIS DSP Library  
00008 * Title:        arm_fir_decimate_q31.c  
00009 *  
00010 * Description:  Q31 FIR Decimator.  
00011 *  
00012 * Target Processor: Cortex-M4/Cortex-M3
00013 *  
00014 * Version 1.0.3 2010/11/29 
00015 *    Re-organized the CMSIS folders and updated documentation.  
00016 *   
00017 * Version 1.0.2 2010/11/11  
00018 *    Documentation updated.   
00019 *  
00020 * Version 1.0.1 2010/10/05   
00021 *    Production release and review comments incorporated.  
00022 *  
00023 * Version 1.0.0 2010/09/20   
00024 *    Production release and review comments incorporated  
00025 *  
00026 * Version 0.0.7  2010/06/10   
00027 *    Misra-C changes done  
00028 * -------------------------------------------------------------------- */ 
00029  
00030 #include "arm_math.h" 
00031  
00032 /**  
00033  * @ingroup groupFilters  
00034  */ 
00035  
00036 /**  
00037  * @addtogroup FIR_decimate  
00038  * @{  
00039  */ 
00040  
00041 /**  
00042  * @brief Processing function for the Q31 FIR decimator.  
00043  * @param[in] *S points to an instance of the Q31 FIR decimator structure.  
00044  * @param[in] *pSrc points to the block of input data.  
00045  * @param[out] *pDst points to the block of output data  
00046  * @param[in] blockSize number of input samples to process per call.  
00047  * @return none  
00048  *  
00049  * <b>Scaling and Overflow Behavior:</b>  
00050  * \par  
00051  * The function is implemented using an internal 64-bit accumulator.  
00052  * The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.  
00053  * Thus, if the accumulator result overflows it wraps around rather than clip.  
00054  * In order to avoid overflows completely the input signal must be scaled down by log2(numTaps) bits (where log2 is read as log to the base 2).  
00055  * After all multiply-accumulates are performed, the 2.62 accumulator is truncated to 1.32 format and then saturated to 1.31 format.  
00056  *  
00057  * \par  
00058  * Refer to the function <code>arm_fir_decimate_fast_q31()</code> for a faster but less precise implementation of this function.  
00059  */ 
00060  
00061 void arm_fir_decimate_q31( 
00062   const arm_fir_decimate_instance_q31 * S, 
00063   q31_t * pSrc, 
00064   q31_t * pDst, 
00065   uint32_t blockSize) 
00066 { 
00067   q31_t *pState = S->pState;                     /* State pointer */ 
00068   q31_t *pCoeffs = S->pCoeffs;                   /* Coefficient pointer */ 
00069   q31_t *pStateCurnt;                            /* Points to the current sample of the state */ 
00070   q31_t x0, c0;                                  /* Temporary variables to hold state and coefficient values */ 
00071   q31_t *px;                                     /* Temporary pointers for state buffer */ 
00072   q31_t *pb;                                     /* Temporary pointers for coefficient buffer */ 
00073   q63_t sum0;                                    /* Accumulator */ 
00074   uint32_t numTaps = S->numTaps;                 /* Number of taps */ 
00075   uint32_t i, tapCnt, blkCnt, outBlockSize = blockSize / S->M;  /* Loop counters */ 
00076  
00077  
00078   /* S->pState buffer contains previous frame (numTaps - 1) samples */ 
00079   /* pStateCurnt points to the location where the new input data should be written */ 
00080   pStateCurnt = S->pState + (numTaps - 1u); 
00081  
00082   /* Total number of output samples to be computed */ 
00083   blkCnt = outBlockSize; 
00084  
00085   while(blkCnt > 0u) 
00086   { 
00087     /* Copy decimation factor number of new input samples into the state buffer */ 
00088     i = S->M; 
00089  
00090     do 
00091     { 
00092       *pStateCurnt++ = *pSrc++; 
00093  
00094     } while(--i); 
00095  
00096     /* Set accumulator to zero */ 
00097     sum0 = 0; 
00098  
00099     /* Initialize state pointer */ 
00100     px = pState; 
00101  
00102     /* Initialize coeff pointer */ 
00103     pb = pCoeffs; 
00104  
00105     /* Loop unrolling.  Process 4 taps at a time. */ 
00106     tapCnt = numTaps >> 2; 
00107  
00108     /* Loop over the number of taps.  Unroll by a factor of 4.  
00109      ** Repeat until we've computed numTaps-4 coefficients. */ 
00110     while(tapCnt > 0u) 
00111     { 
00112       /* Read the b[numTaps-1] coefficient */ 
00113       c0 = *(pb++); 
00114  
00115       /* Read x[n-numTaps-1] sample */ 
00116       x0 = *(px++); 
00117  
00118       /* Perform the multiply-accumulate */ 
00119       sum0 += (q63_t) x0 *c0; 
00120  
00121       /* Read the b[numTaps-2] coefficient */ 
00122       c0 = *(pb++); 
00123  
00124       /* Read x[n-numTaps-2] sample */ 
00125       x0 = *(px++); 
00126  
00127       /* Perform the multiply-accumulate */ 
00128       sum0 += (q63_t) x0 *c0; 
00129  
00130       /* Read the b[numTaps-3] coefficient */ 
00131       c0 = *(pb++); 
00132  
00133       /* Read x[n-numTaps-3] sample */ 
00134       x0 = *(px++); 
00135  
00136       /* Perform the multiply-accumulate */ 
00137       sum0 += (q63_t) x0 *c0; 
00138  
00139       /* Read the b[numTaps-4] coefficient */ 
00140       c0 = *(pb++); 
00141  
00142       /* Read x[n-numTaps-4] sample */ 
00143       x0 = *(px++); 
00144  
00145       /* Perform the multiply-accumulate */ 
00146       sum0 += (q63_t) x0 *c0; 
00147  
00148       /* Decrement the loop counter */ 
00149       tapCnt--; 
00150     } 
00151  
00152     /* If the filter length is not a multiple of 4, compute the remaining filter taps */ 
00153     tapCnt = numTaps % 0x4u; 
00154  
00155     while(tapCnt > 0u) 
00156     { 
00157       /* Read coefficients */ 
00158       c0 = *(pb++); 
00159  
00160       /* Fetch 1 state variable */ 
00161       x0 = *(px++); 
00162  
00163       /* Perform the multiply-accumulate */ 
00164       sum0 += (q63_t) x0 *c0; 
00165  
00166       /* Decrement the loop counter */ 
00167       tapCnt--; 
00168     } 
00169  
00170     /* Advance the state pointer by the decimation factor  
00171      * to process the next group of decimation factor number samples */ 
00172     pState = pState + S->M; 
00173  
00174     /* The result is in the accumulator, store in the destination buffer. */ 
00175     *pDst++ = (q31_t) (sum0 >> 31); 
00176  
00177     /* Decrement the loop counter */ 
00178     blkCnt--; 
00179   } 
00180  
00181   /* Processing is complete.  
00182    ** Now copy the last numTaps - 1 samples to the satrt of the state buffer.  
00183    ** This prepares the state buffer for the next function call. */ 
00184  
00185   /* Points to the start of the state buffer */ 
00186   pStateCurnt = S->pState; 
00187  
00188   i = (numTaps - 1u) >> 2u; 
00189  
00190   /* copy data */ 
00191   while(i > 0u) 
00192   { 
00193     *pStateCurnt++ = *pState++; 
00194     *pStateCurnt++ = *pState++; 
00195     *pStateCurnt++ = *pState++; 
00196     *pStateCurnt++ = *pState++; 
00197  
00198     /* Decrement the loop counter */ 
00199     i--; 
00200   } 
00201  
00202   i = (numTaps - 1u) % 0x04u; 
00203  
00204   /* copy data */ 
00205   while(i > 0u) 
00206   { 
00207     *pStateCurnt++ = *pState++; 
00208  
00209     /* Decrement the loop counter */ 
00210     i--; 
00211   } 
00212 } 
00213  
00214 /**  
00215  * @} end of FIR_decimate group  
00216  */