arm_fir_decimate_q15.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_q15.c  
00009 *  
00010 * Description:  Q15 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 Q15 FIR decimator.  
00043  * @param[in] *S points to an instance of the Q15 FIR decimator structure.  
00044  * @param[in] *pSrc points to the block of input data.  
00045  * @param[out] *pDst points to the location where the output result is written.  
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 a 64-bit internal accumulator.  
00052  * Both coefficients and state variables are represented in 1.15 format and multiplications yield a 2.30 result.  
00053  * The 2.30 intermediate results are accumulated in a 64-bit accumulator in 34.30 format.  
00054  * There is no risk of internal overflow with this approach and the full precision of intermediate multiplications is preserved.  
00055  * After all additions have been performed, the accumulator is truncated to 34.15 format by discarding low 15 bits.  
00056  * Lastly, the accumulator is saturated to yield a result in 1.15 format.  
00057  *  
00058  * \par  
00059  * Refer to the function <code>arm_fir_decimate_fast_q15()</code> for a faster but less precise implementation of this function.  
00060  */ 
00061  
00062 void arm_fir_decimate_q15( 
00063   const arm_fir_decimate_instance_q15 * S, 
00064   q15_t * pSrc, 
00065   q15_t * pDst, 
00066   uint32_t blockSize) 
00067 { 
00068   q15_t *pState = S->pState;                     /* State pointer */ 
00069   q15_t *pCoeffs = S->pCoeffs;                   /* Coefficient pointer */ 
00070   q15_t *pStateCurnt;                            /* Points to the current sample of the state */ 
00071   q15_t *px;                                     /* Temporary pointer for state buffer */ 
00072   q15_t *pb;                                     /* Temporary pointer coefficient buffer */ 
00073   q31_t x0, c0;                                  /* Temporary variables to hold state and coefficient values */ 
00074   q63_t sum0;                                    /* Accumulators */ 
00075   uint32_t numTaps = S->numTaps;                 /* Number of taps */ 
00076   uint32_t i, blkCnt, tapCnt, outBlockSize = blockSize / S->M;  /* Loop counters */ 
00077  
00078  
00079   /* S->pState buffer contains previous frame (numTaps - 1) samples */ 
00080   /* pStateCurnt points to the location where the new input data should be written */ 
00081   pStateCurnt = S->pState + (numTaps - 1u); 
00082  
00083   /* Total number of output samples to be computed */ 
00084   blkCnt = outBlockSize; 
00085  
00086   while(blkCnt > 0u) 
00087   { 
00088     /* Copy decimation factor number of new input samples into the state buffer */ 
00089     i = S->M; 
00090  
00091     do 
00092     { 
00093       *pStateCurnt++ = *pSrc++; 
00094  
00095     } while(--i); 
00096  
00097     /*Set sum to zero */ 
00098     sum0 = 0; 
00099  
00100     /* Initialize state pointer */ 
00101     px = pState; 
00102  
00103     /* Initialize coeff pointer */ 
00104     pb = pCoeffs; 
00105  
00106     /* Loop unrolling.  Process 4 taps at a time. */ 
00107     tapCnt = numTaps >> 2; 
00108  
00109     /* Loop over the number of taps.  Unroll by a factor of 4.  
00110      ** Repeat until we've computed numTaps-4 coefficients. */ 
00111     while(tapCnt > 0u) 
00112     { 
00113       /* Read the Read b[numTaps-1] and b[numTaps-2]  coefficients */ 
00114       c0 = *__SIMD32(pb)++; 
00115  
00116       /* Read x[n-numTaps-1] and x[n-numTaps-2]sample */ 
00117       x0 = *__SIMD32(px)++; 
00118  
00119       /* Perform the multiply-accumulate */ 
00120       sum0 = __SMLALD(x0, c0, sum0); 
00121  
00122       /* Read the b[numTaps-3] and b[numTaps-4] coefficient */ 
00123       c0 = *__SIMD32(pb)++; 
00124  
00125       /* Read x[n-numTaps-2] and x[n-numTaps-3] sample */ 
00126       x0 = *__SIMD32(px)++; 
00127  
00128       /* Perform the multiply-accumulate */ 
00129       sum0 = __SMLALD(x0, c0, sum0); 
00130  
00131       /* Decrement the loop counter */ 
00132       tapCnt--; 
00133     } 
00134  
00135     /* If the filter length is not a multiple of 4, compute the remaining filter taps */ 
00136     tapCnt = numTaps % 0x4u; 
00137  
00138     while(tapCnt > 0u) 
00139     { 
00140       /* Read coefficients */ 
00141       c0 = *pb++; 
00142  
00143       /* Fetch 1 state variable */ 
00144       x0 = *px++; 
00145  
00146       /* Perform the multiply-accumulate */ 
00147       sum0 = __SMLALD(x0, c0, sum0); 
00148  
00149       /* Decrement the loop counter */ 
00150       tapCnt--; 
00151     } 
00152  
00153     /* Advance the state pointer by the decimation factor  
00154      * to process the next group of decimation factor number samples */ 
00155     pState = pState + S->M; 
00156  
00157     /* Store filter output, smlad returns the values in 2.14 format */ 
00158     /* so downsacle by 15 to get output in 1.15 */ 
00159     *pDst++ = (q15_t) (__SSAT((sum0 >> 15), 16)); 
00160  
00161     /* Decrement the loop counter */ 
00162     blkCnt--; 
00163   } 
00164  
00165   /* Processing is complete.  
00166    ** Now copy the last numTaps - 1 samples to the satrt of the state buffer.  
00167    ** This prepares the state buffer for the next function call. */ 
00168  
00169   /* Points to the start of the state buffer */ 
00170   pStateCurnt = S->pState; 
00171  
00172   i = (numTaps - 1u) >> 2u; 
00173  
00174   /* copy data */ 
00175   while(i > 0u) 
00176   { 
00177     *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; 
00178     *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; 
00179  
00180     /* Decrement the loop counter */ 
00181     i--; 
00182   } 
00183  
00184   i = (numTaps - 1u) % 0x04u; 
00185  
00186   /* copy data */ 
00187   while(i > 0u) 
00188   { 
00189     *pStateCurnt++ = *pState++; 
00190  
00191     /* Decrement the loop counter */ 
00192     i--; 
00193   } 
00194 } 
00195  
00196 /**  
00197  * @} end of FIR_decimate group  
00198  */