CMSIS DSP library
Dependents: performance_timer Surfboard_ gps2rtty Capstone ... more
Legacy Warning
This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.
cmsis_dsp/FilteringFunctions/arm_lms_norm_q15.c@1:fdd22bb7aa52, 2012-11-28 (annotated)
- Committer:
- emilmont
- Date:
- Wed Nov 28 12:30:09 2012 +0000
- Revision:
- 1:fdd22bb7aa52
- Child:
- 2:da51fb522205
DSP library code
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
emilmont | 1:fdd22bb7aa52 | 1 | /* ---------------------------------------------------------------------- |
emilmont | 1:fdd22bb7aa52 | 2 | * Copyright (C) 2010 ARM Limited. All rights reserved. |
emilmont | 1:fdd22bb7aa52 | 3 | * |
emilmont | 1:fdd22bb7aa52 | 4 | * $Date: 15. February 2012 |
emilmont | 1:fdd22bb7aa52 | 5 | * $Revision: V1.1.0 |
emilmont | 1:fdd22bb7aa52 | 6 | * |
emilmont | 1:fdd22bb7aa52 | 7 | * Project: CMSIS DSP Library |
emilmont | 1:fdd22bb7aa52 | 8 | * Title: arm_lms_norm_q15.c |
emilmont | 1:fdd22bb7aa52 | 9 | * |
emilmont | 1:fdd22bb7aa52 | 10 | * Description: Q15 NLMS filter. |
emilmont | 1:fdd22bb7aa52 | 11 | * |
emilmont | 1:fdd22bb7aa52 | 12 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
emilmont | 1:fdd22bb7aa52 | 13 | * |
emilmont | 1:fdd22bb7aa52 | 14 | * Version 1.1.0 2012/02/15 |
emilmont | 1:fdd22bb7aa52 | 15 | * Updated with more optimizations, bug fixes and minor API changes. |
emilmont | 1:fdd22bb7aa52 | 16 | * |
emilmont | 1:fdd22bb7aa52 | 17 | * Version 1.0.10 2011/7/15 |
emilmont | 1:fdd22bb7aa52 | 18 | * Big Endian support added and Merged M0 and M3/M4 Source code. |
emilmont | 1:fdd22bb7aa52 | 19 | * |
emilmont | 1:fdd22bb7aa52 | 20 | * Version 1.0.3 2010/11/29 |
emilmont | 1:fdd22bb7aa52 | 21 | * Re-organized the CMSIS folders and updated documentation. |
emilmont | 1:fdd22bb7aa52 | 22 | * |
emilmont | 1:fdd22bb7aa52 | 23 | * Version 1.0.2 2010/11/11 |
emilmont | 1:fdd22bb7aa52 | 24 | * Documentation updated. |
emilmont | 1:fdd22bb7aa52 | 25 | * |
emilmont | 1:fdd22bb7aa52 | 26 | * Version 1.0.1 2010/10/05 |
emilmont | 1:fdd22bb7aa52 | 27 | * Production release and review comments incorporated. |
emilmont | 1:fdd22bb7aa52 | 28 | * |
emilmont | 1:fdd22bb7aa52 | 29 | * Version 1.0.0 2010/09/20 |
emilmont | 1:fdd22bb7aa52 | 30 | * Production release and review comments incorporated |
emilmont | 1:fdd22bb7aa52 | 31 | * |
emilmont | 1:fdd22bb7aa52 | 32 | * Version 0.0.7 2010/06/10 |
emilmont | 1:fdd22bb7aa52 | 33 | * Misra-C changes done |
emilmont | 1:fdd22bb7aa52 | 34 | * -------------------------------------------------------------------- */ |
emilmont | 1:fdd22bb7aa52 | 35 | |
emilmont | 1:fdd22bb7aa52 | 36 | #include "arm_math.h" |
emilmont | 1:fdd22bb7aa52 | 37 | |
emilmont | 1:fdd22bb7aa52 | 38 | /** |
emilmont | 1:fdd22bb7aa52 | 39 | * @ingroup groupFilters |
emilmont | 1:fdd22bb7aa52 | 40 | */ |
emilmont | 1:fdd22bb7aa52 | 41 | |
emilmont | 1:fdd22bb7aa52 | 42 | /** |
emilmont | 1:fdd22bb7aa52 | 43 | * @addtogroup LMS_NORM |
emilmont | 1:fdd22bb7aa52 | 44 | * @{ |
emilmont | 1:fdd22bb7aa52 | 45 | */ |
emilmont | 1:fdd22bb7aa52 | 46 | |
emilmont | 1:fdd22bb7aa52 | 47 | /** |
emilmont | 1:fdd22bb7aa52 | 48 | * @brief Processing function for Q15 normalized LMS filter. |
emilmont | 1:fdd22bb7aa52 | 49 | * @param[in] *S points to an instance of the Q15 normalized LMS filter structure. |
emilmont | 1:fdd22bb7aa52 | 50 | * @param[in] *pSrc points to the block of input data. |
emilmont | 1:fdd22bb7aa52 | 51 | * @param[in] *pRef points to the block of reference data. |
emilmont | 1:fdd22bb7aa52 | 52 | * @param[out] *pOut points to the block of output data. |
emilmont | 1:fdd22bb7aa52 | 53 | * @param[out] *pErr points to the block of error data. |
emilmont | 1:fdd22bb7aa52 | 54 | * @param[in] blockSize number of samples to process. |
emilmont | 1:fdd22bb7aa52 | 55 | * @return none. |
emilmont | 1:fdd22bb7aa52 | 56 | * |
emilmont | 1:fdd22bb7aa52 | 57 | * <b>Scaling and Overflow Behavior:</b> |
emilmont | 1:fdd22bb7aa52 | 58 | * \par |
emilmont | 1:fdd22bb7aa52 | 59 | * The function is implemented using a 64-bit internal accumulator. |
emilmont | 1:fdd22bb7aa52 | 60 | * Both coefficients and state variables are represented in 1.15 format and |
emilmont | 1:fdd22bb7aa52 | 61 | * multiplications yield a 2.30 result. The 2.30 intermediate results are |
emilmont | 1:fdd22bb7aa52 | 62 | * accumulated in a 64-bit accumulator in 34.30 format. |
emilmont | 1:fdd22bb7aa52 | 63 | * There is no risk of internal overflow with this approach and the full |
emilmont | 1:fdd22bb7aa52 | 64 | * precision of intermediate multiplications is preserved. After all additions |
emilmont | 1:fdd22bb7aa52 | 65 | * have been performed, the accumulator is truncated to 34.15 format by |
emilmont | 1:fdd22bb7aa52 | 66 | * discarding low 15 bits. Lastly, the accumulator is saturated to yield a |
emilmont | 1:fdd22bb7aa52 | 67 | * result in 1.15 format. |
emilmont | 1:fdd22bb7aa52 | 68 | * |
emilmont | 1:fdd22bb7aa52 | 69 | * \par |
emilmont | 1:fdd22bb7aa52 | 70 | * In this filter, filter coefficients are updated for each sample and the updation of filter cofficients are saturted. |
emilmont | 1:fdd22bb7aa52 | 71 | * |
emilmont | 1:fdd22bb7aa52 | 72 | */ |
emilmont | 1:fdd22bb7aa52 | 73 | |
emilmont | 1:fdd22bb7aa52 | 74 | void arm_lms_norm_q15( |
emilmont | 1:fdd22bb7aa52 | 75 | arm_lms_norm_instance_q15 * S, |
emilmont | 1:fdd22bb7aa52 | 76 | q15_t * pSrc, |
emilmont | 1:fdd22bb7aa52 | 77 | q15_t * pRef, |
emilmont | 1:fdd22bb7aa52 | 78 | q15_t * pOut, |
emilmont | 1:fdd22bb7aa52 | 79 | q15_t * pErr, |
emilmont | 1:fdd22bb7aa52 | 80 | uint32_t blockSize) |
emilmont | 1:fdd22bb7aa52 | 81 | { |
emilmont | 1:fdd22bb7aa52 | 82 | q15_t *pState = S->pState; /* State pointer */ |
emilmont | 1:fdd22bb7aa52 | 83 | q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */ |
emilmont | 1:fdd22bb7aa52 | 84 | q15_t *pStateCurnt; /* Points to the current sample of the state */ |
emilmont | 1:fdd22bb7aa52 | 85 | q15_t *px, *pb; /* Temporary pointers for state and coefficient buffers */ |
emilmont | 1:fdd22bb7aa52 | 86 | q15_t mu = S->mu; /* Adaptive factor */ |
emilmont | 1:fdd22bb7aa52 | 87 | uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */ |
emilmont | 1:fdd22bb7aa52 | 88 | uint32_t tapCnt, blkCnt; /* Loop counters */ |
emilmont | 1:fdd22bb7aa52 | 89 | q31_t energy; /* Energy of the input */ |
emilmont | 1:fdd22bb7aa52 | 90 | q63_t acc; /* Accumulator */ |
emilmont | 1:fdd22bb7aa52 | 91 | q15_t e = 0, d = 0; /* error, reference data sample */ |
emilmont | 1:fdd22bb7aa52 | 92 | q15_t w = 0, in; /* weight factor and state */ |
emilmont | 1:fdd22bb7aa52 | 93 | q15_t x0; /* temporary variable to hold input sample */ |
emilmont | 1:fdd22bb7aa52 | 94 | //uint32_t shift = (uint32_t) S->postShift + 1u; /* Shift to be applied to the output */ |
emilmont | 1:fdd22bb7aa52 | 95 | q15_t errorXmu, oneByEnergy; /* Temporary variables to store error and mu product and reciprocal of energy */ |
emilmont | 1:fdd22bb7aa52 | 96 | q15_t postShift; /* Post shift to be applied to weight after reciprocal calculation */ |
emilmont | 1:fdd22bb7aa52 | 97 | q31_t coef; /* Teporary variable for coefficient */ |
emilmont | 1:fdd22bb7aa52 | 98 | q31_t acc_l, acc_h; |
emilmont | 1:fdd22bb7aa52 | 99 | int32_t lShift = (15 - (int32_t) S->postShift); /* Post shift */ |
emilmont | 1:fdd22bb7aa52 | 100 | int32_t uShift = (32 - lShift); |
emilmont | 1:fdd22bb7aa52 | 101 | |
emilmont | 1:fdd22bb7aa52 | 102 | energy = S->energy; |
emilmont | 1:fdd22bb7aa52 | 103 | x0 = S->x0; |
emilmont | 1:fdd22bb7aa52 | 104 | |
emilmont | 1:fdd22bb7aa52 | 105 | /* S->pState points to buffer which contains previous frame (numTaps - 1) samples */ |
emilmont | 1:fdd22bb7aa52 | 106 | /* pStateCurnt points to the location where the new input data should be written */ |
emilmont | 1:fdd22bb7aa52 | 107 | pStateCurnt = &(S->pState[(numTaps - 1u)]); |
emilmont | 1:fdd22bb7aa52 | 108 | |
emilmont | 1:fdd22bb7aa52 | 109 | /* Loop over blockSize number of values */ |
emilmont | 1:fdd22bb7aa52 | 110 | blkCnt = blockSize; |
emilmont | 1:fdd22bb7aa52 | 111 | |
emilmont | 1:fdd22bb7aa52 | 112 | |
emilmont | 1:fdd22bb7aa52 | 113 | #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 114 | |
emilmont | 1:fdd22bb7aa52 | 115 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
emilmont | 1:fdd22bb7aa52 | 116 | |
emilmont | 1:fdd22bb7aa52 | 117 | while(blkCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 118 | { |
emilmont | 1:fdd22bb7aa52 | 119 | /* Copy the new input sample into the state buffer */ |
emilmont | 1:fdd22bb7aa52 | 120 | *pStateCurnt++ = *pSrc; |
emilmont | 1:fdd22bb7aa52 | 121 | |
emilmont | 1:fdd22bb7aa52 | 122 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 123 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 124 | |
emilmont | 1:fdd22bb7aa52 | 125 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 126 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 127 | |
emilmont | 1:fdd22bb7aa52 | 128 | /* Read the sample from input buffer */ |
emilmont | 1:fdd22bb7aa52 | 129 | in = *pSrc++; |
emilmont | 1:fdd22bb7aa52 | 130 | |
emilmont | 1:fdd22bb7aa52 | 131 | /* Update the energy calculation */ |
emilmont | 1:fdd22bb7aa52 | 132 | energy -= (((q31_t) x0 * (x0)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 133 | energy += (((q31_t) in * (in)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 134 | |
emilmont | 1:fdd22bb7aa52 | 135 | /* Set the accumulator to zero */ |
emilmont | 1:fdd22bb7aa52 | 136 | acc = 0; |
emilmont | 1:fdd22bb7aa52 | 137 | |
emilmont | 1:fdd22bb7aa52 | 138 | /* Loop unrolling. Process 4 taps at a time. */ |
emilmont | 1:fdd22bb7aa52 | 139 | tapCnt = numTaps >> 2; |
emilmont | 1:fdd22bb7aa52 | 140 | |
emilmont | 1:fdd22bb7aa52 | 141 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 142 | { |
emilmont | 1:fdd22bb7aa52 | 143 | |
emilmont | 1:fdd22bb7aa52 | 144 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 145 | #ifndef UNALIGNED_SUPPORT_DISABLE |
emilmont | 1:fdd22bb7aa52 | 146 | |
emilmont | 1:fdd22bb7aa52 | 147 | acc = __SMLALD(*__SIMD32(px)++, (*__SIMD32(pb)++), acc); |
emilmont | 1:fdd22bb7aa52 | 148 | acc = __SMLALD(*__SIMD32(px)++, (*__SIMD32(pb)++), acc); |
emilmont | 1:fdd22bb7aa52 | 149 | |
emilmont | 1:fdd22bb7aa52 | 150 | #else |
emilmont | 1:fdd22bb7aa52 | 151 | |
emilmont | 1:fdd22bb7aa52 | 152 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 153 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 154 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 155 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 156 | |
emilmont | 1:fdd22bb7aa52 | 157 | #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */ |
emilmont | 1:fdd22bb7aa52 | 158 | |
emilmont | 1:fdd22bb7aa52 | 159 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 160 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 161 | } |
emilmont | 1:fdd22bb7aa52 | 162 | |
emilmont | 1:fdd22bb7aa52 | 163 | /* If the filter length is not a multiple of 4, compute the remaining filter taps */ |
emilmont | 1:fdd22bb7aa52 | 164 | tapCnt = numTaps % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 165 | |
emilmont | 1:fdd22bb7aa52 | 166 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 167 | { |
emilmont | 1:fdd22bb7aa52 | 168 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 169 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 170 | |
emilmont | 1:fdd22bb7aa52 | 171 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 172 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 173 | } |
emilmont | 1:fdd22bb7aa52 | 174 | |
emilmont | 1:fdd22bb7aa52 | 175 | /* Calc lower part of acc */ |
emilmont | 1:fdd22bb7aa52 | 176 | acc_l = acc & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 177 | |
emilmont | 1:fdd22bb7aa52 | 178 | /* Calc upper part of acc */ |
emilmont | 1:fdd22bb7aa52 | 179 | acc_h = (acc >> 32) & 0xffffffff; |
emilmont | 1:fdd22bb7aa52 | 180 | |
emilmont | 1:fdd22bb7aa52 | 181 | /* Apply shift for lower part of acc and upper part of acc */ |
emilmont | 1:fdd22bb7aa52 | 182 | acc = (uint32_t) acc_l >> lShift | acc_h << uShift; |
emilmont | 1:fdd22bb7aa52 | 183 | |
emilmont | 1:fdd22bb7aa52 | 184 | /* Converting the result to 1.15 format and saturate the output */ |
emilmont | 1:fdd22bb7aa52 | 185 | acc = __SSAT(acc, 16u); |
emilmont | 1:fdd22bb7aa52 | 186 | |
emilmont | 1:fdd22bb7aa52 | 187 | /* Store the result from accumulator into the destination buffer. */ |
emilmont | 1:fdd22bb7aa52 | 188 | *pOut++ = (q15_t) acc; |
emilmont | 1:fdd22bb7aa52 | 189 | |
emilmont | 1:fdd22bb7aa52 | 190 | /* Compute and store error */ |
emilmont | 1:fdd22bb7aa52 | 191 | d = *pRef++; |
emilmont | 1:fdd22bb7aa52 | 192 | e = d - (q15_t) acc; |
emilmont | 1:fdd22bb7aa52 | 193 | *pErr++ = e; |
emilmont | 1:fdd22bb7aa52 | 194 | |
emilmont | 1:fdd22bb7aa52 | 195 | /* Calculation of 1/energy */ |
emilmont | 1:fdd22bb7aa52 | 196 | postShift = arm_recip_q15((q15_t) energy + DELTA_Q15, |
emilmont | 1:fdd22bb7aa52 | 197 | &oneByEnergy, S->recipTable); |
emilmont | 1:fdd22bb7aa52 | 198 | |
emilmont | 1:fdd22bb7aa52 | 199 | /* Calculation of e * mu value */ |
emilmont | 1:fdd22bb7aa52 | 200 | errorXmu = (q15_t) (((q31_t) e * mu) >> 15); |
emilmont | 1:fdd22bb7aa52 | 201 | |
emilmont | 1:fdd22bb7aa52 | 202 | /* Calculation of (e * mu) * (1/energy) value */ |
emilmont | 1:fdd22bb7aa52 | 203 | acc = (((q31_t) errorXmu * oneByEnergy) >> (15 - postShift)); |
emilmont | 1:fdd22bb7aa52 | 204 | |
emilmont | 1:fdd22bb7aa52 | 205 | /* Weighting factor for the normalized version */ |
emilmont | 1:fdd22bb7aa52 | 206 | w = (q15_t) __SSAT((q31_t) acc, 16); |
emilmont | 1:fdd22bb7aa52 | 207 | |
emilmont | 1:fdd22bb7aa52 | 208 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 209 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 210 | |
emilmont | 1:fdd22bb7aa52 | 211 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 212 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 213 | |
emilmont | 1:fdd22bb7aa52 | 214 | /* Loop unrolling. Process 4 taps at a time. */ |
emilmont | 1:fdd22bb7aa52 | 215 | tapCnt = numTaps >> 2; |
emilmont | 1:fdd22bb7aa52 | 216 | |
emilmont | 1:fdd22bb7aa52 | 217 | /* Update filter coefficients */ |
emilmont | 1:fdd22bb7aa52 | 218 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 219 | { |
emilmont | 1:fdd22bb7aa52 | 220 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 221 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 222 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 223 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 224 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 225 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 226 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 227 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 228 | |
emilmont | 1:fdd22bb7aa52 | 229 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 230 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 231 | } |
emilmont | 1:fdd22bb7aa52 | 232 | |
emilmont | 1:fdd22bb7aa52 | 233 | /* If the filter length is not a multiple of 4, compute the remaining filter taps */ |
emilmont | 1:fdd22bb7aa52 | 234 | tapCnt = numTaps % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 235 | |
emilmont | 1:fdd22bb7aa52 | 236 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 237 | { |
emilmont | 1:fdd22bb7aa52 | 238 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 239 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 240 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 241 | |
emilmont | 1:fdd22bb7aa52 | 242 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 243 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 244 | } |
emilmont | 1:fdd22bb7aa52 | 245 | |
emilmont | 1:fdd22bb7aa52 | 246 | /* Read the sample from state buffer */ |
emilmont | 1:fdd22bb7aa52 | 247 | x0 = *pState; |
emilmont | 1:fdd22bb7aa52 | 248 | |
emilmont | 1:fdd22bb7aa52 | 249 | /* Advance state pointer by 1 for the next sample */ |
emilmont | 1:fdd22bb7aa52 | 250 | pState = pState + 1u; |
emilmont | 1:fdd22bb7aa52 | 251 | |
emilmont | 1:fdd22bb7aa52 | 252 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 253 | blkCnt--; |
emilmont | 1:fdd22bb7aa52 | 254 | } |
emilmont | 1:fdd22bb7aa52 | 255 | |
emilmont | 1:fdd22bb7aa52 | 256 | /* Save energy and x0 values for the next frame */ |
emilmont | 1:fdd22bb7aa52 | 257 | S->energy = (q15_t) energy; |
emilmont | 1:fdd22bb7aa52 | 258 | S->x0 = x0; |
emilmont | 1:fdd22bb7aa52 | 259 | |
emilmont | 1:fdd22bb7aa52 | 260 | /* Processing is complete. Now copy the last numTaps - 1 samples to the |
emilmont | 1:fdd22bb7aa52 | 261 | satrt of the state buffer. This prepares the state buffer for the |
emilmont | 1:fdd22bb7aa52 | 262 | next function call. */ |
emilmont | 1:fdd22bb7aa52 | 263 | |
emilmont | 1:fdd22bb7aa52 | 264 | /* Points to the start of the pState buffer */ |
emilmont | 1:fdd22bb7aa52 | 265 | pStateCurnt = S->pState; |
emilmont | 1:fdd22bb7aa52 | 266 | |
emilmont | 1:fdd22bb7aa52 | 267 | /* Calculation of count for copying integer writes */ |
emilmont | 1:fdd22bb7aa52 | 268 | tapCnt = (numTaps - 1u) >> 2; |
emilmont | 1:fdd22bb7aa52 | 269 | |
emilmont | 1:fdd22bb7aa52 | 270 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 271 | { |
emilmont | 1:fdd22bb7aa52 | 272 | |
emilmont | 1:fdd22bb7aa52 | 273 | #ifndef UNALIGNED_SUPPORT_DISABLE |
emilmont | 1:fdd22bb7aa52 | 274 | |
emilmont | 1:fdd22bb7aa52 | 275 | *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; |
emilmont | 1:fdd22bb7aa52 | 276 | *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; |
emilmont | 1:fdd22bb7aa52 | 277 | |
emilmont | 1:fdd22bb7aa52 | 278 | #else |
emilmont | 1:fdd22bb7aa52 | 279 | |
emilmont | 1:fdd22bb7aa52 | 280 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 281 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 282 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 283 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 284 | |
emilmont | 1:fdd22bb7aa52 | 285 | #endif |
emilmont | 1:fdd22bb7aa52 | 286 | |
emilmont | 1:fdd22bb7aa52 | 287 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 288 | |
emilmont | 1:fdd22bb7aa52 | 289 | } |
emilmont | 1:fdd22bb7aa52 | 290 | |
emilmont | 1:fdd22bb7aa52 | 291 | /* Calculation of count for remaining q15_t data */ |
emilmont | 1:fdd22bb7aa52 | 292 | tapCnt = (numTaps - 1u) % 0x4u; |
emilmont | 1:fdd22bb7aa52 | 293 | |
emilmont | 1:fdd22bb7aa52 | 294 | /* copy data */ |
emilmont | 1:fdd22bb7aa52 | 295 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 296 | { |
emilmont | 1:fdd22bb7aa52 | 297 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 298 | |
emilmont | 1:fdd22bb7aa52 | 299 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 300 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 301 | } |
emilmont | 1:fdd22bb7aa52 | 302 | |
emilmont | 1:fdd22bb7aa52 | 303 | #else |
emilmont | 1:fdd22bb7aa52 | 304 | |
emilmont | 1:fdd22bb7aa52 | 305 | /* Run the below code for Cortex-M0 */ |
emilmont | 1:fdd22bb7aa52 | 306 | |
emilmont | 1:fdd22bb7aa52 | 307 | while(blkCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 308 | { |
emilmont | 1:fdd22bb7aa52 | 309 | /* Copy the new input sample into the state buffer */ |
emilmont | 1:fdd22bb7aa52 | 310 | *pStateCurnt++ = *pSrc; |
emilmont | 1:fdd22bb7aa52 | 311 | |
emilmont | 1:fdd22bb7aa52 | 312 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 313 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 314 | |
emilmont | 1:fdd22bb7aa52 | 315 | /* Initialize pCoeffs pointer */ |
emilmont | 1:fdd22bb7aa52 | 316 | pb = pCoeffs; |
emilmont | 1:fdd22bb7aa52 | 317 | |
emilmont | 1:fdd22bb7aa52 | 318 | /* Read the sample from input buffer */ |
emilmont | 1:fdd22bb7aa52 | 319 | in = *pSrc++; |
emilmont | 1:fdd22bb7aa52 | 320 | |
emilmont | 1:fdd22bb7aa52 | 321 | /* Update the energy calculation */ |
emilmont | 1:fdd22bb7aa52 | 322 | energy -= (((q31_t) x0 * (x0)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 323 | energy += (((q31_t) in * (in)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 324 | |
emilmont | 1:fdd22bb7aa52 | 325 | /* Set the accumulator to zero */ |
emilmont | 1:fdd22bb7aa52 | 326 | acc = 0; |
emilmont | 1:fdd22bb7aa52 | 327 | |
emilmont | 1:fdd22bb7aa52 | 328 | /* Loop over numTaps number of values */ |
emilmont | 1:fdd22bb7aa52 | 329 | tapCnt = numTaps; |
emilmont | 1:fdd22bb7aa52 | 330 | |
emilmont | 1:fdd22bb7aa52 | 331 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 332 | { |
emilmont | 1:fdd22bb7aa52 | 333 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 334 | acc += (((q31_t) * px++ * (*pb++))); |
emilmont | 1:fdd22bb7aa52 | 335 | |
emilmont | 1:fdd22bb7aa52 | 336 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 337 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 338 | } |
emilmont | 1:fdd22bb7aa52 | 339 | |
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 | /* Apply shift for lower part of acc and upper part of acc */ |
emilmont | 1:fdd22bb7aa52 | 347 | acc = (uint32_t) acc_l >> lShift | acc_h << uShift; |
emilmont | 1:fdd22bb7aa52 | 348 | |
emilmont | 1:fdd22bb7aa52 | 349 | /* Converting the result to 1.15 format and saturate the output */ |
emilmont | 1:fdd22bb7aa52 | 350 | acc = __SSAT(acc, 16u); |
emilmont | 1:fdd22bb7aa52 | 351 | |
emilmont | 1:fdd22bb7aa52 | 352 | /* Converting the result to 1.15 format */ |
emilmont | 1:fdd22bb7aa52 | 353 | //acc = __SSAT((acc >> (16u - shift)), 16u); |
emilmont | 1:fdd22bb7aa52 | 354 | |
emilmont | 1:fdd22bb7aa52 | 355 | /* Store the result from accumulator into the destination buffer. */ |
emilmont | 1:fdd22bb7aa52 | 356 | *pOut++ = (q15_t) acc; |
emilmont | 1:fdd22bb7aa52 | 357 | |
emilmont | 1:fdd22bb7aa52 | 358 | /* Compute and store error */ |
emilmont | 1:fdd22bb7aa52 | 359 | d = *pRef++; |
emilmont | 1:fdd22bb7aa52 | 360 | e = d - (q15_t) acc; |
emilmont | 1:fdd22bb7aa52 | 361 | *pErr++ = e; |
emilmont | 1:fdd22bb7aa52 | 362 | |
emilmont | 1:fdd22bb7aa52 | 363 | /* Calculation of 1/energy */ |
emilmont | 1:fdd22bb7aa52 | 364 | postShift = arm_recip_q15((q15_t) energy + DELTA_Q15, |
emilmont | 1:fdd22bb7aa52 | 365 | &oneByEnergy, S->recipTable); |
emilmont | 1:fdd22bb7aa52 | 366 | |
emilmont | 1:fdd22bb7aa52 | 367 | /* Calculation of e * mu value */ |
emilmont | 1:fdd22bb7aa52 | 368 | errorXmu = (q15_t) (((q31_t) e * mu) >> 15); |
emilmont | 1:fdd22bb7aa52 | 369 | |
emilmont | 1:fdd22bb7aa52 | 370 | /* Calculation of (e * mu) * (1/energy) value */ |
emilmont | 1:fdd22bb7aa52 | 371 | acc = (((q31_t) errorXmu * oneByEnergy) >> (15 - postShift)); |
emilmont | 1:fdd22bb7aa52 | 372 | |
emilmont | 1:fdd22bb7aa52 | 373 | /* Weighting factor for the normalized version */ |
emilmont | 1:fdd22bb7aa52 | 374 | w = (q15_t) __SSAT((q31_t) acc, 16); |
emilmont | 1:fdd22bb7aa52 | 375 | |
emilmont | 1:fdd22bb7aa52 | 376 | /* Initialize pState pointer */ |
emilmont | 1:fdd22bb7aa52 | 377 | px = pState; |
emilmont | 1:fdd22bb7aa52 | 378 | |
emilmont | 1:fdd22bb7aa52 | 379 | /* Initialize coeff pointer */ |
emilmont | 1:fdd22bb7aa52 | 380 | pb = (pCoeffs); |
emilmont | 1:fdd22bb7aa52 | 381 | |
emilmont | 1:fdd22bb7aa52 | 382 | /* Loop over numTaps number of values */ |
emilmont | 1:fdd22bb7aa52 | 383 | tapCnt = numTaps; |
emilmont | 1:fdd22bb7aa52 | 384 | |
emilmont | 1:fdd22bb7aa52 | 385 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 386 | { |
emilmont | 1:fdd22bb7aa52 | 387 | /* Perform the multiply-accumulate */ |
emilmont | 1:fdd22bb7aa52 | 388 | coef = *pb + (((q31_t) w * (*px++)) >> 15); |
emilmont | 1:fdd22bb7aa52 | 389 | *pb++ = (q15_t) __SSAT((coef), 16); |
emilmont | 1:fdd22bb7aa52 | 390 | |
emilmont | 1:fdd22bb7aa52 | 391 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 392 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 393 | } |
emilmont | 1:fdd22bb7aa52 | 394 | |
emilmont | 1:fdd22bb7aa52 | 395 | /* Read the sample from state buffer */ |
emilmont | 1:fdd22bb7aa52 | 396 | x0 = *pState; |
emilmont | 1:fdd22bb7aa52 | 397 | |
emilmont | 1:fdd22bb7aa52 | 398 | /* Advance state pointer by 1 for the next sample */ |
emilmont | 1:fdd22bb7aa52 | 399 | pState = pState + 1u; |
emilmont | 1:fdd22bb7aa52 | 400 | |
emilmont | 1:fdd22bb7aa52 | 401 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 402 | blkCnt--; |
emilmont | 1:fdd22bb7aa52 | 403 | } |
emilmont | 1:fdd22bb7aa52 | 404 | |
emilmont | 1:fdd22bb7aa52 | 405 | /* Save energy and x0 values for the next frame */ |
emilmont | 1:fdd22bb7aa52 | 406 | S->energy = (q15_t) energy; |
emilmont | 1:fdd22bb7aa52 | 407 | S->x0 = x0; |
emilmont | 1:fdd22bb7aa52 | 408 | |
emilmont | 1:fdd22bb7aa52 | 409 | /* Processing is complete. Now copy the last numTaps - 1 samples to the |
emilmont | 1:fdd22bb7aa52 | 410 | satrt of the state buffer. This prepares the state buffer for the |
emilmont | 1:fdd22bb7aa52 | 411 | next function call. */ |
emilmont | 1:fdd22bb7aa52 | 412 | |
emilmont | 1:fdd22bb7aa52 | 413 | /* Points to the start of the pState buffer */ |
emilmont | 1:fdd22bb7aa52 | 414 | pStateCurnt = S->pState; |
emilmont | 1:fdd22bb7aa52 | 415 | |
emilmont | 1:fdd22bb7aa52 | 416 | /* copy (numTaps - 1u) data */ |
emilmont | 1:fdd22bb7aa52 | 417 | tapCnt = (numTaps - 1u); |
emilmont | 1:fdd22bb7aa52 | 418 | |
emilmont | 1:fdd22bb7aa52 | 419 | /* copy data */ |
emilmont | 1:fdd22bb7aa52 | 420 | while(tapCnt > 0u) |
emilmont | 1:fdd22bb7aa52 | 421 | { |
emilmont | 1:fdd22bb7aa52 | 422 | *pStateCurnt++ = *pState++; |
emilmont | 1:fdd22bb7aa52 | 423 | |
emilmont | 1:fdd22bb7aa52 | 424 | /* Decrement the loop counter */ |
emilmont | 1:fdd22bb7aa52 | 425 | tapCnt--; |
emilmont | 1:fdd22bb7aa52 | 426 | } |
emilmont | 1:fdd22bb7aa52 | 427 | |
emilmont | 1:fdd22bb7aa52 | 428 | #endif /* #ifndef ARM_MATH_CM0 */ |
emilmont | 1:fdd22bb7aa52 | 429 | |
emilmont | 1:fdd22bb7aa52 | 430 | } |
emilmont | 1:fdd22bb7aa52 | 431 | |
emilmont | 1:fdd22bb7aa52 | 432 | |
emilmont | 1:fdd22bb7aa52 | 433 | /** |
emilmont | 1:fdd22bb7aa52 | 434 | * @} end of LMS_NORM group |
emilmont | 1:fdd22bb7aa52 | 435 | */ |