CMSIS DSP library
Dependents: KL25Z_FFT_Demo Hat_Board_v5_1 KL25Z_FFT_Demo_tony KL25Z_FFT_Demo_tony ... more
Fork of mbed-dsp by
cmsis_dsp/FilteringFunctions/arm_lms_norm_q31.c@3:7a284390b0ce, 2013-11-08 (annotated)
- Committer:
- mbed_official
- Date:
- Fri Nov 08 13:45:10 2013 +0000
- Revision:
- 3:7a284390b0ce
- Parent:
- 2:da51fb522205
Synchronized with git revision e69956aba2f68a2a26ac26b051f8d349deaa1ce8
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
emilmont | 1:fdd22bb7aa52 | 1 | /* ---------------------------------------------------------------------- |
mbed_official | 3:7a284390b0ce | 2 | * Copyright (C) 2010-2013 ARM Limited. All rights reserved. |
emilmont | 1:fdd22bb7aa52 | 3 | * |
mbed_official | 3:7a284390b0ce | 4 | * $Date: 17. January 2013 |
mbed_official | 3:7a284390b0ce | 5 | * $Revision: V1.4.1 |
emilmont | 1:fdd22bb7aa52 | 6 | * |
emilmont | 2:da51fb522205 | 7 | * Project: CMSIS DSP Library |
emilmont | 2:da51fb522205 | 8 | * Title: arm_lms_norm_q31.c |
emilmont | 1:fdd22bb7aa52 | 9 | * |
emilmont | 2:da51fb522205 | 10 | * Description: Processing function for the Q31 NLMS filter. |
emilmont | 1:fdd22bb7aa52 | 11 | * |
emilmont | 1:fdd22bb7aa52 | 12 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
emilmont | 1:fdd22bb7aa52 | 13 | * |
mbed_official | 3:7a284390b0ce | 14 | * Redistribution and use in source and binary forms, with or without |
mbed_official | 3:7a284390b0ce | 15 | * modification, are permitted provided that the following conditions |
mbed_official | 3:7a284390b0ce | 16 | * are met: |
mbed_official | 3:7a284390b0ce | 17 | * - Redistributions of source code must retain the above copyright |
mbed_official | 3:7a284390b0ce | 18 | * notice, this list of conditions and the following disclaimer. |
mbed_official | 3:7a284390b0ce | 19 | * - Redistributions in binary form must reproduce the above copyright |
mbed_official | 3:7a284390b0ce | 20 | * notice, this list of conditions and the following disclaimer in |
mbed_official | 3:7a284390b0ce | 21 | * the documentation and/or other materials provided with the |
mbed_official | 3:7a284390b0ce | 22 | * distribution. |
mbed_official | 3:7a284390b0ce | 23 | * - Neither the name of ARM LIMITED nor the names of its contributors |
mbed_official | 3:7a284390b0ce | 24 | * may be used to endorse or promote products derived from this |
mbed_official | 3:7a284390b0ce | 25 | * software without specific prior written permission. |
mbed_official | 3:7a284390b0ce | 26 | * |
mbed_official | 3:7a284390b0ce | 27 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
mbed_official | 3:7a284390b0ce | 28 | * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
mbed_official | 3:7a284390b0ce | 29 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
mbed_official | 3:7a284390b0ce | 30 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
mbed_official | 3:7a284390b0ce | 31 | * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
mbed_official | 3:7a284390b0ce | 32 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, |
mbed_official | 3:7a284390b0ce | 33 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
mbed_official | 3:7a284390b0ce | 34 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
mbed_official | 3:7a284390b0ce | 35 | * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
mbed_official | 3:7a284390b0ce | 36 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
mbed_official | 3:7a284390b0ce | 37 | * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
mbed_official | 3:7a284390b0ce | 38 | * POSSIBILITY OF SUCH DAMAGE. |
emilmont | 1:fdd22bb7aa52 | 39 | * -------------------------------------------------------------------- */ |
emilmont | 1:fdd22bb7aa52 | 40 | |
emilmont | 1:fdd22bb7aa52 | 41 | #include "arm_math.h" |
emilmont | 1:fdd22bb7aa52 | 42 | |
emilmont | 1:fdd22bb7aa52 | 43 | /** |
emilmont | 1:fdd22bb7aa52 | 44 | * @ingroup groupFilters |
emilmont | 1:fdd22bb7aa52 | 45 | */ |
emilmont | 1:fdd22bb7aa52 | 46 | |
emilmont | 1:fdd22bb7aa52 | 47 | /** |
emilmont | 1:fdd22bb7aa52 | 48 | * @addtogroup LMS_NORM |
emilmont | 1:fdd22bb7aa52 | 49 | * @{ |
emilmont | 1:fdd22bb7aa52 | 50 | */ |
emilmont | 1:fdd22bb7aa52 | 51 | |
emilmont | 1:fdd22bb7aa52 | 52 | /** |
emilmont | 1:fdd22bb7aa52 | 53 | * @brief Processing function for Q31 normalized LMS filter. |
emilmont | 1:fdd22bb7aa52 | 54 | * @param[in] *S points to an instance of the Q31 normalized LMS filter structure. |
emilmont | 1:fdd22bb7aa52 | 55 | * @param[in] *pSrc points to the block of input data. |
emilmont | 1:fdd22bb7aa52 | 56 | * @param[in] *pRef points to the block of reference data. |
emilmont | 1:fdd22bb7aa52 | 57 | * @param[out] *pOut points to the block of output data. |
emilmont | 1:fdd22bb7aa52 | 58 | * @param[out] *pErr points to the block of error data. |
emilmont | 1:fdd22bb7aa52 | 59 | * @param[in] blockSize number of samples to process. |
emilmont | 1:fdd22bb7aa52 | 60 | * @return none. |
emilmont | 1:fdd22bb7aa52 | 61 | * |
emilmont | 1:fdd22bb7aa52 | 62 | * <b>Scaling and Overflow Behavior:</b> |
emilmont | 1:fdd22bb7aa52 | 63 | * \par |
emilmont | 1:fdd22bb7aa52 | 64 | * The function is implemented using an internal 64-bit accumulator. |
emilmont | 1:fdd22bb7aa52 | 65 | * The accumulator has a 2.62 format and maintains full precision of the intermediate |
emilmont | 1:fdd22bb7aa52 | 66 | * multiplication results but provides only a single guard bit. |
emilmont | 1:fdd22bb7aa52 | 67 | * Thus, if the accumulator result overflows it wraps around rather than clip. |
emilmont | 1:fdd22bb7aa52 | 68 | * In order to avoid overflows completely the input signal must be scaled down by |
emilmont | 1:fdd22bb7aa52 | 69 | * log2(numTaps) bits. The reference signal should not be scaled down. |
emilmont | 1:fdd22bb7aa52 | 70 | * After all multiply-accumulates are performed, the 2.62 accumulator is shifted |
emilmont | 1:fdd22bb7aa52 | 71 | * and saturated to 1.31 format to yield the final result. |
emilmont | 1:fdd22bb7aa52 | 72 | * The output signal and error signal are in 1.31 format. |
emilmont | 1:fdd22bb7aa52 | 73 | * |
emilmont | 1:fdd22bb7aa52 | 74 | * \par |
emilmont | 2:da51fb522205 | 75 | * In this filter, filter coefficients are updated for each sample and the |
emilmont | 1:fdd22bb7aa52 | 76 | * updation of filter cofficients are saturted. |
emilmont | 1:fdd22bb7aa52 | 77 | * |
emilmont | 1:fdd22bb7aa52 | 78 | */ |
emilmont | 1:fdd22bb7aa52 | 79 | |
emilmont | 1:fdd22bb7aa52 | 80 | void arm_lms_norm_q31( |
emilmont | 1:fdd22bb7aa52 | 81 | arm_lms_norm_instance_q31 * S, |
emilmont | 1:fdd22bb7aa52 | 82 | q31_t * pSrc, |
emilmont | 1:fdd22bb7aa52 | 83 | q31_t * pRef, |
emilmont | 1:fdd22bb7aa52 | 84 | q31_t * pOut, |
emilmont | 1:fdd22bb7aa52 | 85 | q31_t * pErr, |
emilmont | 1:fdd22bb7aa52 | 86 | uint32_t blockSize) |
emilmont | 1:fdd22bb7aa52 | 87 | { |
emilmont | 1:fdd22bb7aa52 | 88 | q31_t *pState = S->pState; /* State pointer */ |
emilmont | 1:fdd22bb7aa52 | 89 | q31_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */ |
emilmont | 1:fdd22bb7aa52 | 90 | q31_t *pStateCurnt; /* Points to the current sample of the state */ |
emilmont | 1:fdd22bb7aa52 | 91 | q31_t *px, *pb; /* Temporary pointers for state and coefficient buffers */ |
emilmont | 1:fdd22bb7aa52 | 92 | q31_t mu = S->mu; /* Adaptive factor */ |
emilmont | 1:fdd22bb7aa52 | 93 | uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */ |
emilmont | 1:fdd22bb7aa52 | 94 | uint32_t tapCnt, blkCnt; /* Loop counters */ |
emilmont | 1:fdd22bb7aa52 | 95 | q63_t energy; /* Energy of the input */ |
emilmont | 1:fdd22bb7aa52 | 96 | q63_t acc; /* Accumulator */ |
emilmont | 1:fdd22bb7aa52 | 97 | q31_t e = 0, d = 0; /* error, reference data sample */ |
emilmont | 1:fdd22bb7aa52 | 98 | q31_t w = 0, in; /* weight factor and state */ |
emilmont | 1:fdd22bb7aa52 | 99 | q31_t x0; /* temporary variable to hold input sample */ |
emilmont | 1:fdd22bb7aa52 | 100 | // uint32_t shift = 32u - ((uint32_t) S->postShift + 1u); /* Shift to be applied to the output */ |
emilmont | 1:fdd22bb7aa52 | 101 | q31_t errorXmu, oneByEnergy; /* Temporary variables to store error and mu product and reciprocal of energy */ |
emilmont | 1:fdd22bb7aa52 | 102 | q31_t postShift; /* Post shift to be applied to weight after reciprocal calculation */ |
emilmont | 1:fdd22bb7aa52 | 103 | q31_t coef; /* Temporary variable for coef */ |
emilmont | 1:fdd22bb7aa52 | 104 | q31_t acc_l, acc_h; /* temporary input */ |
emilmont | 1:fdd22bb7aa52 | 105 | uint32_t uShift = ((uint32_t) S->postShift + 1u); |
emilmont | 1:fdd22bb7aa52 | 106 | uint32_t lShift = 32u - uShift; /* Shift to be applied to the output */ |
emilmont | 1:fdd22bb7aa52 | 107 | |
emilmont | 1:fdd22bb7aa52 | 108 | energy = S->energy; |
emilmont | 1:fdd22bb7aa52 | 109 | x0 = S->x0; |
emilmont | 1:fdd22bb7aa52 | 110 | |
emilmont | 1:fdd22bb7aa52 | 111 | /* S->pState points to buffer which contains previous frame (numTaps - 1) samples */ |
emilmont | 1:fdd22bb7aa52 | 112 | /* pStateCurnt points to the location where the new input data should be written */ |
emilmont | 1:fdd22bb7aa52 | 113 | pStateCurnt = &(S->pState[(numTaps - 1u)]); |
emilmont | 1:fdd22bb7aa52 | 114 | |
emilmont | 1:fdd22bb7aa52 | 115 | /* Loop over blockSize number of values */ |
emilmont | 1:fdd22bb7aa52 | 116 | blkCnt = blockSize; |
emilmont | 1:fdd22bb7aa52 | 117 | |
emilmont | 1:fdd22bb7aa52 | 118 | |
mbed_official | 3:7a284390b0ce | 119 | #ifndef ARM_MATH_CM0_FAMILY |
emilmont | 1:fdd22bb7aa52 | 120 | |
emilmont | 1:fdd22bb7aa52 | 121 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
emilmont | 1:fdd22bb7aa52 | 122 | |
emilmont | 1:fdd22bb7aa52 | 123 | while(blkCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 124 | { |
emilmont | 1:fdd22bb7aa52 | 125 | |
emilmont | 1:fdd22bb7aa52 | 126 | /* Copy the new input sample into the state buffer */ |
emilmont | 1:fdd22bb7aa52 | 127 | *pStateCurnt++ = *pSrc; |
emilmont | 1:fdd22bb7aa52 | 128 | |
emilmont | 1:fdd22bb7aa52 | 129 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 130 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 131 | |
emilmont | 1:fdd22bb7aa52 | 132 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 133 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 134 | |
emilmont | 1:fdd22bb7aa52 | 135 | /* Read the sample from input buffer */ |
emilmont | 1:fdd22bb7aa52 | 136 | in = *pSrc++; |
emilmont | 1:fdd22bb7aa52 | 137 | |
emilmont | 1:fdd22bb7aa52 | 138 | /* Update the energy calculation */ |
emilmont | 1:fdd22bb7aa52 | 139 | energy = (q31_t) ((((q63_t) energy << 32) - |
emilmont | 1:fdd22bb7aa52 | 140 | (((q63_t) x0 * x0) << 1)) >> 32); |
emilmont | 1:fdd22bb7aa52 | 141 | energy = (q31_t) (((((q63_t) in * in) << 1) + (energy << 32)) >> 32); |
emilmont | 1:fdd22bb7aa52 | 142 | |
emilmont | 1:fdd22bb7aa52 | 143 | /* Set the accumulator to zero */ |
emilmont | 1:fdd22bb7aa52 | 144 | acc = 0; |
emilmont | 1:fdd22bb7aa52 | 145 | |
emilmont | 1:fdd22bb7aa52 | 146 | /* Loop unrolling. Process 4 taps at a time. */ |
emilmont | 1:fdd22bb7aa52 | 147 | tapCnt = numTaps >> 2; |
emilmont | 1:fdd22bb7aa52 | 148 | |
emilmont | 1:fdd22bb7aa52 | 149 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 150 | { |
emilmont | 1:fdd22bb7aa52 | 151 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 152 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 153 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 154 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 155 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 156 | |
emilmont | 1:fdd22bb7aa52 | 157 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 158 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 159 | } |
emilmont | 1:fdd22bb7aa52 | 160 | |
emilmont | 1:fdd22bb7aa52 | 161 | /* If the filter length is not a multiple of 4, compute the remaining filter taps */ |
emilmont | 1:fdd22bb7aa52 | 162 | tapCnt = numTaps % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 163 | |
emilmont | 1:fdd22bb7aa52 | 164 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 165 | { |
emilmont | 1:fdd22bb7aa52 | 166 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 167 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 168 | |
emilmont | 1:fdd22bb7aa52 | 169 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 170 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 171 | } |
emilmont | 1:fdd22bb7aa52 | 172 | |
emilmont | 1:fdd22bb7aa52 | 173 | /* Converting the result to 1.31 format */ |
emilmont | 1:fdd22bb7aa52 | 174 | /* Calc lower part of acc */ |
emilmont | 1:fdd22bb7aa52 | 175 | acc_l = acc & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 176 | |
emilmont | 1:fdd22bb7aa52 | 177 | /* Calc upper part of acc */ |
emilmont | 1:fdd22bb7aa52 | 178 | acc_h = (acc >> 32) & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 179 | |
emilmont | 1:fdd22bb7aa52 | 180 | acc = (uint32_t) acc_l >> lShift | acc_h << uShift; |
emilmont | 1:fdd22bb7aa52 | 181 | |
emilmont | 1:fdd22bb7aa52 | 182 | /* Store the result from accumulator into the destination buffer. */ |
emilmont | 1:fdd22bb7aa52 | 183 | *pOut++ = (q31_t) acc; |
emilmont | 1:fdd22bb7aa52 | 184 | |
emilmont | 1:fdd22bb7aa52 | 185 | /* Compute and store error */ |
emilmont | 1:fdd22bb7aa52 | 186 | d = *pRef++; |
emilmont | 1:fdd22bb7aa52 | 187 | e = d - (q31_t) acc; |
emilmont | 1:fdd22bb7aa52 | 188 | *pErr++ = e; |
emilmont | 1:fdd22bb7aa52 | 189 | |
emilmont | 1:fdd22bb7aa52 | 190 | /* Calculates the reciprocal of energy */ |
emilmont | 1:fdd22bb7aa52 | 191 | postShift = arm_recip_q31(energy + DELTA_Q31, |
emilmont | 1:fdd22bb7aa52 | 192 | &oneByEnergy, &S->recipTable[0]); |
emilmont | 1:fdd22bb7aa52 | 193 | |
emilmont | 1:fdd22bb7aa52 | 194 | /* Calculation of product of (e * mu) */ |
emilmont | 1:fdd22bb7aa52 | 195 | errorXmu = (q31_t) (((q63_t) e * mu) >> 31); |
emilmont | 1:fdd22bb7aa52 | 196 | |
emilmont | 1:fdd22bb7aa52 | 197 | /* Weighting factor for the normalized version */ |
emilmont | 1:fdd22bb7aa52 | 198 | w = clip_q63_to_q31(((q63_t) errorXmu * oneByEnergy) >> (31 - postShift)); |
emilmont | 1:fdd22bb7aa52 | 199 | |
emilmont | 1:fdd22bb7aa52 | 200 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 201 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 202 | |
emilmont | 1:fdd22bb7aa52 | 203 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 204 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 205 | |
emilmont | 1:fdd22bb7aa52 | 206 | /* Loop unrolling. Process 4 taps at a time. */ |
emilmont | 1:fdd22bb7aa52 | 207 | tapCnt = numTaps >> 2; |
emilmont | 1:fdd22bb7aa52 | 208 | |
emilmont | 1:fdd22bb7aa52 | 209 | /* Update filter coefficients */ |
emilmont | 1:fdd22bb7aa52 | 210 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 211 | { |
emilmont | 1:fdd22bb7aa52 | 212 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 213 | |
emilmont | 1:fdd22bb7aa52 | 214 | /* coef is in 2.30 format */ |
emilmont | 1:fdd22bb7aa52 | 215 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 216 | /* get coef in 1.31 format by left shifting */ |
emilmont | 1:fdd22bb7aa52 | 217 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 218 | /* update coefficient buffer to next coefficient */ |
emilmont | 1:fdd22bb7aa52 | 219 | pb++; |
emilmont | 1:fdd22bb7aa52 | 220 | |
emilmont | 1:fdd22bb7aa52 | 221 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 222 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 223 | pb++; |
emilmont | 1:fdd22bb7aa52 | 224 | |
emilmont | 1:fdd22bb7aa52 | 225 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 226 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 227 | pb++; |
emilmont | 1:fdd22bb7aa52 | 228 | |
emilmont | 1:fdd22bb7aa52 | 229 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 230 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 231 | pb++; |
emilmont | 1:fdd22bb7aa52 | 232 | |
emilmont | 1:fdd22bb7aa52 | 233 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 234 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 235 | } |
emilmont | 1:fdd22bb7aa52 | 236 | |
emilmont | 1:fdd22bb7aa52 | 237 | /* If the filter length is not a multiple of 4, compute the remaining filter taps */ |
emilmont | 1:fdd22bb7aa52 | 238 | tapCnt = numTaps % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 239 | |
emilmont | 1:fdd22bb7aa52 | 240 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 241 | { |
emilmont | 1:fdd22bb7aa52 | 242 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 243 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 244 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 245 | pb++; |
emilmont | 1:fdd22bb7aa52 | 246 | |
emilmont | 1:fdd22bb7aa52 | 247 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 248 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 249 | } |
emilmont | 1:fdd22bb7aa52 | 250 | |
emilmont | 1:fdd22bb7aa52 | 251 | /* Read the sample from state buffer */ |
emilmont | 1:fdd22bb7aa52 | 252 | x0 = *pState; |
emilmont | 1:fdd22bb7aa52 | 253 | |
emilmont | 1:fdd22bb7aa52 | 254 | /* Advance state pointer by 1 for the next sample */ |
emilmont | 1:fdd22bb7aa52 | 255 | pState = pState + 1; |
emilmont | 1:fdd22bb7aa52 | 256 | |
emilmont | 1:fdd22bb7aa52 | 257 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 258 | blkCnt--; |
emilmont | 1:fdd22bb7aa52 | 259 | } |
emilmont | 1:fdd22bb7aa52 | 260 | |
emilmont | 1:fdd22bb7aa52 | 261 | /* Save energy and x0 values for the next frame */ |
emilmont | 1:fdd22bb7aa52 | 262 | S->energy = (q31_t) energy; |
emilmont | 1:fdd22bb7aa52 | 263 | S->x0 = x0; |
emilmont | 1:fdd22bb7aa52 | 264 | |
emilmont | 1:fdd22bb7aa52 | 265 | /* Processing is complete. Now copy the last numTaps - 1 samples to the |
emilmont | 1:fdd22bb7aa52 | 266 | satrt of the state buffer. This prepares the state buffer for the |
emilmont | 1:fdd22bb7aa52 | 267 | next function call. */ |
emilmont | 1:fdd22bb7aa52 | 268 | |
emilmont | 1:fdd22bb7aa52 | 269 | /* Points to the start of the pState buffer */ |
emilmont | 1:fdd22bb7aa52 | 270 | pStateCurnt = S->pState; |
emilmont | 1:fdd22bb7aa52 | 271 | |
emilmont | 1:fdd22bb7aa52 | 272 | /* Loop unrolling for (numTaps - 1u) samples copy */ |
emilmont | 1:fdd22bb7aa52 | 273 | tapCnt = (numTaps - 1u) >> 2u; |
emilmont | 1:fdd22bb7aa52 | 274 | |
emilmont | 1:fdd22bb7aa52 | 275 | /* copy data */ |
emilmont | 1:fdd22bb7aa52 | 276 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 277 | { |
emilmont | 1:fdd22bb7aa52 | 278 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 279 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 280 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 281 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 282 | |
emilmont | 1:fdd22bb7aa52 | 283 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 284 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 285 | } |
emilmont | 1:fdd22bb7aa52 | 286 | |
emilmont | 1:fdd22bb7aa52 | 287 | /* Calculate remaining number of copies */ |
emilmont | 1:fdd22bb7aa52 | 288 | tapCnt = (numTaps - 1u) % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 289 | |
emilmont | 1:fdd22bb7aa52 | 290 | /* Copy the remaining q31_t data */ |
emilmont | 1:fdd22bb7aa52 | 291 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 292 | { |
emilmont | 1:fdd22bb7aa52 | 293 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 294 | |
emilmont | 1:fdd22bb7aa52 | 295 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 296 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 297 | } |
emilmont | 1:fdd22bb7aa52 | 298 | |
emilmont | 1:fdd22bb7aa52 | 299 | #else |
emilmont | 1:fdd22bb7aa52 | 300 | |
emilmont | 1:fdd22bb7aa52 | 301 | /* Run the below code for Cortex-M0 */ |
emilmont | 1:fdd22bb7aa52 | 302 | |
emilmont | 1:fdd22bb7aa52 | 303 | while(blkCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 304 | { |
emilmont | 1:fdd22bb7aa52 | 305 | |
emilmont | 1:fdd22bb7aa52 | 306 | /* Copy the new input sample into the state buffer */ |
emilmont | 1:fdd22bb7aa52 | 307 | *pStateCurnt++ = *pSrc; |
emilmont | 1:fdd22bb7aa52 | 308 | |
emilmont | 1:fdd22bb7aa52 | 309 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 310 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 311 | |
emilmont | 1:fdd22bb7aa52 | 312 | /* Initialize pCoeffs pointer */ |
emilmont | 1:fdd22bb7aa52 | 313 | pb = pCoeffs; |
emilmont | 1:fdd22bb7aa52 | 314 | |
emilmont | 1:fdd22bb7aa52 | 315 | /* Read the sample from input buffer */ |
emilmont | 1:fdd22bb7aa52 | 316 | in = *pSrc++; |
emilmont | 1:fdd22bb7aa52 | 317 | |
emilmont | 1:fdd22bb7aa52 | 318 | /* Update the energy calculation */ |
emilmont | 1:fdd22bb7aa52 | 319 | energy = |
emilmont | 1:fdd22bb7aa52 | 320 | (q31_t) ((((q63_t) energy << 32) - (((q63_t) x0 * x0) << 1)) >> 32); |
emilmont | 1:fdd22bb7aa52 | 321 | energy = (q31_t) (((((q63_t) in * in) << 1) + (energy << 32)) >> 32); |
emilmont | 1:fdd22bb7aa52 | 322 | |
emilmont | 1:fdd22bb7aa52 | 323 | /* Set the accumulator to zero */ |
emilmont | 1:fdd22bb7aa52 | 324 | acc = 0; |
emilmont | 1:fdd22bb7aa52 | 325 | |
emilmont | 1:fdd22bb7aa52 | 326 | /* Loop over numTaps number of values */ |
emilmont | 1:fdd22bb7aa52 | 327 | tapCnt = numTaps; |
emilmont | 1:fdd22bb7aa52 | 328 | |
emilmont | 1:fdd22bb7aa52 | 329 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 330 | { |
emilmont | 1:fdd22bb7aa52 | 331 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 332 | acc += ((q63_t) (*px++)) * (*pb++); |
emilmont | 1:fdd22bb7aa52 | 333 | |
emilmont | 1:fdd22bb7aa52 | 334 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 335 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 336 | } |
emilmont | 1:fdd22bb7aa52 | 337 | |
emilmont | 1:fdd22bb7aa52 | 338 | /* Converting the result to 1.31 format */ |
emilmont | 1:fdd22bb7aa52 | 339 | /* Converting the result to 1.31 format */ |
emilmont | 1:fdd22bb7aa52 | 340 | /* Calc lower part of acc */ |
emilmont | 1:fdd22bb7aa52 | 341 | acc_l = acc & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 342 | |
emilmont | 1:fdd22bb7aa52 | 343 | /* Calc upper part of acc */ |
emilmont | 1:fdd22bb7aa52 | 344 | acc_h = (acc >> 32) & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 345 | |
emilmont | 1:fdd22bb7aa52 | 346 | acc = (uint32_t) acc_l >> lShift | acc_h << uShift; |
emilmont | 1:fdd22bb7aa52 | 347 | |
emilmont | 1:fdd22bb7aa52 | 348 | |
emilmont | 1:fdd22bb7aa52 | 349 | //acc = (q31_t) (acc >> shift); |
emilmont | 1:fdd22bb7aa52 | 350 | |
emilmont | 1:fdd22bb7aa52 | 351 | /* Store the result from accumulator into the destination buffer. */ |
emilmont | 1:fdd22bb7aa52 | 352 | *pOut++ = (q31_t) acc; |
emilmont | 1:fdd22bb7aa52 | 353 | |
emilmont | 1:fdd22bb7aa52 | 354 | /* Compute and store error */ |
emilmont | 1:fdd22bb7aa52 | 355 | d = *pRef++; |
emilmont | 1:fdd22bb7aa52 | 356 | e = d - (q31_t) acc; |
emilmont | 1:fdd22bb7aa52 | 357 | *pErr++ = e; |
emilmont | 1:fdd22bb7aa52 | 358 | |
emilmont | 1:fdd22bb7aa52 | 359 | /* Calculates the reciprocal of energy */ |
emilmont | 1:fdd22bb7aa52 | 360 | postShift = |
emilmont | 1:fdd22bb7aa52 | 361 | arm_recip_q31(energy + DELTA_Q31, &oneByEnergy, &S->recipTable[0]); |
emilmont | 1:fdd22bb7aa52 | 362 | |
emilmont | 1:fdd22bb7aa52 | 363 | /* Calculation of product of (e * mu) */ |
emilmont | 1:fdd22bb7aa52 | 364 | errorXmu = (q31_t) (((q63_t) e * mu) >> 31); |
emilmont | 1:fdd22bb7aa52 | 365 | |
emilmont | 1:fdd22bb7aa52 | 366 | /* Weighting factor for the normalized version */ |
emilmont | 1:fdd22bb7aa52 | 367 | w = clip_q63_to_q31(((q63_t) errorXmu * oneByEnergy) >> (31 - postShift)); |
emilmont | 1:fdd22bb7aa52 | 368 | |
emilmont | 1:fdd22bb7aa52 | 369 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 370 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 371 | |
emilmont | 1:fdd22bb7aa52 | 372 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 373 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 374 | |
emilmont | 1:fdd22bb7aa52 | 375 | /* Loop over numTaps number of values */ |
emilmont | 1:fdd22bb7aa52 | 376 | tapCnt = numTaps; |
emilmont | 1:fdd22bb7aa52 | 377 | |
emilmont | 1:fdd22bb7aa52 | 378 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 379 | { |
emilmont | 1:fdd22bb7aa52 | 380 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 381 | /* coef is in 2.30 format */ |
emilmont | 1:fdd22bb7aa52 | 382 | coef = (q31_t) (((q63_t) w * (*px++)) >> (32)); |
emilmont | 1:fdd22bb7aa52 | 383 | /* get coef in 1.31 format by left shifting */ |
emilmont | 1:fdd22bb7aa52 | 384 | *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u)); |
emilmont | 1:fdd22bb7aa52 | 385 | /* update coefficient buffer to next coefficient */ |
emilmont | 1:fdd22bb7aa52 | 386 | pb++; |
emilmont | 1:fdd22bb7aa52 | 387 | |
emilmont | 1:fdd22bb7aa52 | 388 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 389 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 390 | } |
emilmont | 1:fdd22bb7aa52 | 391 | |
emilmont | 1:fdd22bb7aa52 | 392 | /* Read the sample from state buffer */ |
emilmont | 1:fdd22bb7aa52 | 393 | x0 = *pState; |
emilmont | 1:fdd22bb7aa52 | 394 | |
emilmont | 1:fdd22bb7aa52 | 395 | /* Advance state pointer by 1 for the next sample */ |
emilmont | 1:fdd22bb7aa52 | 396 | pState = pState + 1; |
emilmont | 1:fdd22bb7aa52 | 397 | |
emilmont | 1:fdd22bb7aa52 | 398 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 399 | blkCnt--; |
emilmont | 1:fdd22bb7aa52 | 400 | } |
emilmont | 1:fdd22bb7aa52 | 401 | |
emilmont | 1:fdd22bb7aa52 | 402 | /* Save energy and x0 values for the next frame */ |
emilmont | 1:fdd22bb7aa52 | 403 | S->energy = (q31_t) energy; |
emilmont | 1:fdd22bb7aa52 | 404 | S->x0 = x0; |
emilmont | 1:fdd22bb7aa52 | 405 | |
emilmont | 1:fdd22bb7aa52 | 406 | /* Processing is complete. Now copy the last numTaps - 1 samples to the |
emilmont | 1:fdd22bb7aa52 | 407 | start of the state buffer. This prepares the state buffer for the |
emilmont | 1:fdd22bb7aa52 | 408 | next function call. */ |
emilmont | 1:fdd22bb7aa52 | 409 | |
emilmont | 1:fdd22bb7aa52 | 410 | /* Points to the start of the pState buffer */ |
emilmont | 1:fdd22bb7aa52 | 411 | pStateCurnt = S->pState; |
emilmont | 1:fdd22bb7aa52 | 412 | |
emilmont | 1:fdd22bb7aa52 | 413 | /* Loop for (numTaps - 1u) samples copy */ |
emilmont | 1:fdd22bb7aa52 | 414 | tapCnt = (numTaps - 1u); |
emilmont | 1:fdd22bb7aa52 | 415 | |
emilmont | 1:fdd22bb7aa52 | 416 | /* Copy the remaining q31_t data */ |
emilmont | 1:fdd22bb7aa52 | 417 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 418 | { |
emilmont | 1:fdd22bb7aa52 | 419 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 420 | |
emilmont | 1:fdd22bb7aa52 | 421 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 422 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 423 | } |
emilmont | 1:fdd22bb7aa52 | 424 | |
mbed_official | 3:7a284390b0ce | 425 | #endif /* #ifndef ARM_MATH_CM0_FAMILY */ |
emilmont | 1:fdd22bb7aa52 | 426 | |
emilmont | 1:fdd22bb7aa52 | 427 | } |
emilmont | 1:fdd22bb7aa52 | 428 | |
emilmont | 1:fdd22bb7aa52 | 429 | /** |
emilmont | 1:fdd22bb7aa52 | 430 | * @} end of LMS_NORM group |
emilmont | 1:fdd22bb7aa52 | 431 | */ |