arm_std_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_std_q15.c  
00009 *  
00010 * Description:  Standard deviation of an array of Q15 type.  
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  
00027 #include "arm_math.h" 
00028  
00029 /**  
00030  * @ingroup groupStats  
00031  */ 
00032  
00033 /**  
00034  * @addtogroup STD  
00035  * @{  
00036  */ 
00037  
00038 /**  
00039  * @brief Standard deviation of the elements of a Q15 vector.  
00040  * @param[in]       *pSrc points to the input vector  
00041  * @param[in]       blockSize length of the input vector  
00042  * @param[out]      *pResult standard deviation value returned here  
00043  * @return none.  
00044  *  
00045  * @details  
00046  * <b>Scaling and Overflow Behavior:</b>  
00047  *  
00048  * \par  
00049  * The function is implemented using a 64-bit internal accumulator.  
00050  * The input is represented in 1.15 format. 
00051  * Intermediate multiplication yields a 2.30 format, and this  
00052  * result is added without saturation to a 64-bit accumulator in 34.30 format.  
00053  * With 33 guard bits in the accumulator, there is no risk of overflow, and the  
00054  * full precision of the intermediate multiplication is preserved.  
00055  * Finally, the 34.30 result is truncated to 34.15 format by discarding the lower   
00056  * 15 bits, and then saturated to yield a result in 1.15 format.  
00057  */ 
00058  
00059 void arm_std_q15( 
00060   q15_t * pSrc, 
00061   uint32_t blockSize, 
00062   q15_t * pResult) 
00063 { 
00064   q63_t sum = 0;                                 /* Accumulator */ 
00065   q31_t meanOfSquares, squareOfMean;             /* square of mean and mean of square */ 
00066   q15_t mean;                                    /* mean */ 
00067   q31_t in;                                      /* input value */ 
00068   q15_t in1;                                     /* input value */ 
00069   uint32_t blkCnt;                               /* loop counter */ 
00070   q15_t t;                                       /* Temporary variable */ 
00071   q15_t *pIn;                                    /* Temporary pointer */ 
00072  
00073   pIn = pSrc; 
00074  
00075   /*loop Unrolling */ 
00076   blkCnt = blockSize >> 2u; 
00077  
00078   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.  
00079    ** a second loop below computes the remaining 1 to 3 samples. */ 
00080   while(blkCnt > 0u) 
00081   { 
00082     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1])  */ 
00083     /* Compute Sum of squares of the input samples  
00084      * and then store the result in a temporary variable, sum. */ 
00085     in = *__SIMD32(pSrc)++; 
00086     sum = __SMLALD(in, in, sum); 
00087     in = *__SIMD32(pSrc)++; 
00088     sum = __SMLALD(in, in, sum); 
00089  
00090     /* Decrement the loop counter */ 
00091     blkCnt--; 
00092   } 
00093  
00094   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.  
00095    ** No loop unrolling is used. */ 
00096   blkCnt = blockSize % 0x4u; 
00097  
00098   while(blkCnt > 0u) 
00099   { 
00100     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */ 
00101     /* Compute Sum of squares of the input samples  
00102      * and then store the result in a temporary variable, sum. */ 
00103     in1 = *pSrc++; 
00104     sum = __SMLALD(in1, in1, sum); 
00105  
00106     /* Decrement the loop counter */ 
00107     blkCnt--; 
00108   } 
00109  
00110   /* Compute Mean of squares of the input samples  
00111    * and then store the result in a temporary variable, meanOfSquares. */ 
00112   t = (q15_t) ((1.0 / (blockSize - 1)) * 16384LL); 
00113   sum = __SSAT((sum >> 15u), 16u); 
00114  
00115   meanOfSquares = (q31_t) ((sum * t) >> 14u); 
00116  
00117   /* Reset the accumulator */ 
00118   sum = 0; 
00119  
00120   /*loop Unrolling */ 
00121   blkCnt = blockSize >> 2u; 
00122  
00123   /* Reset the input working pointer */ 
00124   pSrc = pIn; 
00125  
00126   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.  
00127    ** a second loop below computes the remaining 1 to 3 samples. */ 
00128   while(blkCnt > 0u) 
00129   { 
00130     /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */ 
00131     /* Compute sum of all input values and then store the result in a temporary variable, sum. */ 
00132     sum += *pSrc++; 
00133     sum += *pSrc++; 
00134     sum += *pSrc++; 
00135     sum += *pSrc++; 
00136  
00137     /* Decrement the loop counter */ 
00138     blkCnt--; 
00139   } 
00140  
00141   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.  
00142    ** No loop unrolling is used. */ 
00143   blkCnt = blockSize % 0x4u; 
00144  
00145   while(blkCnt > 0u) 
00146   { 
00147     /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */ 
00148     /* Compute sum of all input values and then store the result in a temporary variable, sum. */ 
00149     sum += *pSrc++; 
00150  
00151     /* Decrement the loop counter */ 
00152     blkCnt--; 
00153   } 
00154   /* Compute mean of all input values */ 
00155   t = (q15_t) ((1.0 / (blockSize * (blockSize - 1))) * 32768LL); 
00156   mean = (q15_t) __SSAT(sum, 16u); 
00157  
00158   /* Compute square of mean */ 
00159   squareOfMean = ((q31_t) mean * mean) >> 15; 
00160   squareOfMean = (q31_t) (((q63_t) squareOfMean * t) >> 15); 
00161  
00162   /* mean of the squares minus the square of the mean. */ 
00163   in1 = (q15_t) (meanOfSquares - squareOfMean); 
00164  
00165   /* Compute standard deviation and store the result to the destination */ 
00166   arm_sqrt_q15(in1, pResult); 
00167 } 
00168  
00169 /**  
00170  * @} end of STD group  
00171  */