mbed official / mbed-dsp

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_fir_decimate_fast_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?

UserRevisionLine numberNew 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_fir_decimate_fast_q15.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 1:fdd22bb7aa52 10 * Description: Fast Q15 FIR Decimator.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3
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
emilmont 1:fdd22bb7aa52 33 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 34
emilmont 1:fdd22bb7aa52 35 /**
emilmont 1:fdd22bb7aa52 36 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 37 */
emilmont 1:fdd22bb7aa52 38
emilmont 1:fdd22bb7aa52 39 /**
emilmont 1:fdd22bb7aa52 40 * @addtogroup FIR_decimate
emilmont 1:fdd22bb7aa52 41 * @{
emilmont 1:fdd22bb7aa52 42 */
emilmont 1:fdd22bb7aa52 43
emilmont 1:fdd22bb7aa52 44 /**
emilmont 1:fdd22bb7aa52 45 * @brief Processing function for the Q15 FIR decimator (fast variant) for Cortex-M3 and Cortex-M4.
emilmont 1:fdd22bb7aa52 46 * @param[in] *S points to an instance of the Q15 FIR decimator structure.
emilmont 1:fdd22bb7aa52 47 * @param[in] *pSrc points to the block of input data.
emilmont 1:fdd22bb7aa52 48 * @param[out] *pDst points to the block of output data
emilmont 1:fdd22bb7aa52 49 * @param[in] blockSize number of input samples to process per call.
emilmont 1:fdd22bb7aa52 50 * @return none
emilmont 1:fdd22bb7aa52 51 *
emilmont 1:fdd22bb7aa52 52 * \par Restrictions
emilmont 1:fdd22bb7aa52 53 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 54 * In this case input, output, state buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 55 *
emilmont 1:fdd22bb7aa52 56 * <b>Scaling and Overflow Behavior:</b>
emilmont 1:fdd22bb7aa52 57 * \par
emilmont 1:fdd22bb7aa52 58 * This fast version uses a 32-bit accumulator with 2.30 format.
emilmont 1:fdd22bb7aa52 59 * The accumulator maintains full precision of the intermediate multiplication results but provides only a single guard bit.
emilmont 1:fdd22bb7aa52 60 * Thus, if the accumulator result overflows it wraps around and distorts the result.
emilmont 1:fdd22bb7aa52 61 * In order to avoid overflows completely the input signal must be scaled down by log2(numTaps) bits (log2 is read as log to the base 2).
emilmont 1:fdd22bb7aa52 62 * The 2.30 accumulator is then truncated to 2.15 format and saturated to yield the 1.15 result.
emilmont 1:fdd22bb7aa52 63 *
emilmont 1:fdd22bb7aa52 64 * \par
emilmont 1:fdd22bb7aa52 65 * Refer to the function <code>arm_fir_decimate_q15()</code> for a slower implementation of this function which uses 64-bit accumulation to avoid wrap around distortion.
emilmont 1:fdd22bb7aa52 66 * Both the slow and the fast versions use the same instance structure.
emilmont 1:fdd22bb7aa52 67 * Use the function <code>arm_fir_decimate_init_q15()</code> to initialize the filter structure.
emilmont 1:fdd22bb7aa52 68 */
emilmont 1:fdd22bb7aa52 69
emilmont 1:fdd22bb7aa52 70 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 71
emilmont 1:fdd22bb7aa52 72 void arm_fir_decimate_fast_q15(
emilmont 1:fdd22bb7aa52 73 const arm_fir_decimate_instance_q15 * S,
emilmont 1:fdd22bb7aa52 74 q15_t * pSrc,
emilmont 1:fdd22bb7aa52 75 q15_t * pDst,
emilmont 1:fdd22bb7aa52 76 uint32_t blockSize)
emilmont 1:fdd22bb7aa52 77 {
emilmont 1:fdd22bb7aa52 78 q15_t *pState = S->pState; /* State pointer */
emilmont 1:fdd22bb7aa52 79 q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
emilmont 1:fdd22bb7aa52 80 q15_t *pStateCurnt; /* Points to the current sample of the state */
emilmont 1:fdd22bb7aa52 81 q15_t *px; /* Temporary pointer for state buffer */
emilmont 1:fdd22bb7aa52 82 q15_t *pb; /* Temporary pointer coefficient buffer */
emilmont 1:fdd22bb7aa52 83 q31_t x0, x1, c0, c1; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 84 q31_t sum0; /* Accumulators */
emilmont 1:fdd22bb7aa52 85 q31_t acc0, acc1;
emilmont 1:fdd22bb7aa52 86 q15_t *px0, *px1;
emilmont 1:fdd22bb7aa52 87 uint32_t blkCntN3;
emilmont 1:fdd22bb7aa52 88 uint32_t numTaps = S->numTaps; /* Number of taps */
emilmont 1:fdd22bb7aa52 89 uint32_t i, blkCnt, tapCnt, outBlockSize = blockSize / S->M; /* Loop counters */
emilmont 1:fdd22bb7aa52 90
emilmont 1:fdd22bb7aa52 91
emilmont 1:fdd22bb7aa52 92 /* S->pState buffer contains previous frame (numTaps - 1) samples */
emilmont 1:fdd22bb7aa52 93 /* pStateCurnt points to the location where the new input data should be written */
emilmont 1:fdd22bb7aa52 94 pStateCurnt = S->pState + (numTaps - 1u);
emilmont 1:fdd22bb7aa52 95
emilmont 1:fdd22bb7aa52 96
emilmont 1:fdd22bb7aa52 97 /* Total number of output samples to be computed */
emilmont 1:fdd22bb7aa52 98 blkCnt = outBlockSize / 2;
emilmont 1:fdd22bb7aa52 99 blkCntN3 = outBlockSize - (2 * blkCnt);
emilmont 1:fdd22bb7aa52 100
emilmont 1:fdd22bb7aa52 101
emilmont 1:fdd22bb7aa52 102 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 103 {
emilmont 1:fdd22bb7aa52 104 /* Copy decimation factor number of new input samples into the state buffer */
emilmont 1:fdd22bb7aa52 105 i = 2 * S->M;
emilmont 1:fdd22bb7aa52 106
emilmont 1:fdd22bb7aa52 107 do
emilmont 1:fdd22bb7aa52 108 {
emilmont 1:fdd22bb7aa52 109 *pStateCurnt++ = *pSrc++;
emilmont 1:fdd22bb7aa52 110
emilmont 1:fdd22bb7aa52 111 } while(--i);
emilmont 1:fdd22bb7aa52 112
emilmont 1:fdd22bb7aa52 113 /* Set accumulator to zero */
emilmont 1:fdd22bb7aa52 114 acc0 = 0;
emilmont 1:fdd22bb7aa52 115 acc1 = 0;
emilmont 1:fdd22bb7aa52 116
emilmont 1:fdd22bb7aa52 117 /* Initialize state pointer */
emilmont 1:fdd22bb7aa52 118 px0 = pState;
emilmont 1:fdd22bb7aa52 119
emilmont 1:fdd22bb7aa52 120 px1 = pState + S->M;
emilmont 1:fdd22bb7aa52 121
emilmont 1:fdd22bb7aa52 122
emilmont 1:fdd22bb7aa52 123 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 124 pb = pCoeffs;
emilmont 1:fdd22bb7aa52 125
emilmont 1:fdd22bb7aa52 126 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 127 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 128
emilmont 1:fdd22bb7aa52 129 /* Loop over the number of taps. Unroll by a factor of 4.
emilmont 1:fdd22bb7aa52 130 ** Repeat until we've computed numTaps-4 coefficients. */
emilmont 1:fdd22bb7aa52 131 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 132 {
emilmont 1:fdd22bb7aa52 133 /* Read the Read b[numTaps-1] and b[numTaps-2] coefficients */
emilmont 1:fdd22bb7aa52 134 c0 = *__SIMD32(pb)++;
emilmont 1:fdd22bb7aa52 135
emilmont 1:fdd22bb7aa52 136 /* Read x[n-numTaps-1] and x[n-numTaps-2]sample */
emilmont 1:fdd22bb7aa52 137 x0 = *__SIMD32(px0)++;
emilmont 1:fdd22bb7aa52 138
emilmont 1:fdd22bb7aa52 139 x1 = *__SIMD32(px1)++;
emilmont 1:fdd22bb7aa52 140
emilmont 1:fdd22bb7aa52 141 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 142 acc0 = __SMLAD(x0, c0, acc0);
emilmont 1:fdd22bb7aa52 143
emilmont 1:fdd22bb7aa52 144 acc1 = __SMLAD(x1, c0, acc1);
emilmont 1:fdd22bb7aa52 145
emilmont 1:fdd22bb7aa52 146 /* Read the b[numTaps-3] and b[numTaps-4] coefficient */
emilmont 1:fdd22bb7aa52 147 c0 = *__SIMD32(pb)++;
emilmont 1:fdd22bb7aa52 148
emilmont 1:fdd22bb7aa52 149 /* Read x[n-numTaps-2] and x[n-numTaps-3] sample */
emilmont 1:fdd22bb7aa52 150 x0 = *__SIMD32(px0)++;
emilmont 1:fdd22bb7aa52 151
emilmont 1:fdd22bb7aa52 152 x1 = *__SIMD32(px1)++;
emilmont 1:fdd22bb7aa52 153
emilmont 1:fdd22bb7aa52 154 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 155 acc0 = __SMLAD(x0, c0, acc0);
emilmont 1:fdd22bb7aa52 156
emilmont 1:fdd22bb7aa52 157 acc1 = __SMLAD(x1, c0, acc1);
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 /* Read coefficients */
emilmont 1:fdd22bb7aa52 169 c0 = *pb++;
emilmont 1:fdd22bb7aa52 170
emilmont 1:fdd22bb7aa52 171 /* Fetch 1 state variable */
emilmont 1:fdd22bb7aa52 172 x0 = *px0++;
emilmont 1:fdd22bb7aa52 173
emilmont 1:fdd22bb7aa52 174 x1 = *px1++;
emilmont 1:fdd22bb7aa52 175
emilmont 1:fdd22bb7aa52 176 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 177 acc0 = __SMLAD(x0, c0, acc0);
emilmont 1:fdd22bb7aa52 178 acc1 = __SMLAD(x1, c0, acc1);
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 181 tapCnt--;
emilmont 1:fdd22bb7aa52 182 }
emilmont 1:fdd22bb7aa52 183
emilmont 1:fdd22bb7aa52 184 /* Advance the state pointer by the decimation factor
emilmont 1:fdd22bb7aa52 185 * to process the next group of decimation factor number samples */
emilmont 1:fdd22bb7aa52 186 pState = pState + S->M * 2;
emilmont 1:fdd22bb7aa52 187
emilmont 1:fdd22bb7aa52 188 /* Store filter output, smlad returns the values in 2.14 format */
emilmont 1:fdd22bb7aa52 189 /* so downsacle by 15 to get output in 1.15 */
emilmont 1:fdd22bb7aa52 190 *pDst++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 191 *pDst++ = (q15_t) (__SSAT((acc1 >> 15), 16));
emilmont 1:fdd22bb7aa52 192
emilmont 1:fdd22bb7aa52 193 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 194 blkCnt--;
emilmont 1:fdd22bb7aa52 195 }
emilmont 1:fdd22bb7aa52 196
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198
emilmont 1:fdd22bb7aa52 199 while(blkCntN3 > 0u)
emilmont 1:fdd22bb7aa52 200 {
emilmont 1:fdd22bb7aa52 201 /* Copy decimation factor number of new input samples into the state buffer */
emilmont 1:fdd22bb7aa52 202 i = S->M;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 do
emilmont 1:fdd22bb7aa52 205 {
emilmont 1:fdd22bb7aa52 206 *pStateCurnt++ = *pSrc++;
emilmont 1:fdd22bb7aa52 207
emilmont 1:fdd22bb7aa52 208 } while(--i);
emilmont 1:fdd22bb7aa52 209
emilmont 1:fdd22bb7aa52 210 /*Set sum to zero */
emilmont 1:fdd22bb7aa52 211 sum0 = 0;
emilmont 1:fdd22bb7aa52 212
emilmont 1:fdd22bb7aa52 213 /* Initialize state pointer */
emilmont 1:fdd22bb7aa52 214 px = pState;
emilmont 1:fdd22bb7aa52 215
emilmont 1:fdd22bb7aa52 216 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 217 pb = pCoeffs;
emilmont 1:fdd22bb7aa52 218
emilmont 1:fdd22bb7aa52 219 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 220 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 221
emilmont 1:fdd22bb7aa52 222 /* Loop over the number of taps. Unroll by a factor of 4.
emilmont 1:fdd22bb7aa52 223 ** Repeat until we've computed numTaps-4 coefficients. */
emilmont 1:fdd22bb7aa52 224 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 225 {
emilmont 1:fdd22bb7aa52 226 /* Read the Read b[numTaps-1] and b[numTaps-2] coefficients */
emilmont 1:fdd22bb7aa52 227 c0 = *__SIMD32(pb)++;
emilmont 1:fdd22bb7aa52 228
emilmont 1:fdd22bb7aa52 229 /* Read x[n-numTaps-1] and x[n-numTaps-2]sample */
emilmont 1:fdd22bb7aa52 230 x0 = *__SIMD32(px)++;
emilmont 1:fdd22bb7aa52 231
emilmont 1:fdd22bb7aa52 232 /* Read the b[numTaps-3] and b[numTaps-4] coefficient */
emilmont 1:fdd22bb7aa52 233 c1 = *__SIMD32(pb)++;
emilmont 1:fdd22bb7aa52 234
emilmont 1:fdd22bb7aa52 235 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 236 sum0 = __SMLAD(x0, c0, sum0);
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* Read x[n-numTaps-2] and x[n-numTaps-3] sample */
emilmont 1:fdd22bb7aa52 239 x0 = *__SIMD32(px)++;
emilmont 1:fdd22bb7aa52 240
emilmont 1:fdd22bb7aa52 241 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 242 sum0 = __SMLAD(x0, c1, sum0);
emilmont 1:fdd22bb7aa52 243
emilmont 1:fdd22bb7aa52 244 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 245 tapCnt--;
emilmont 1:fdd22bb7aa52 246 }
emilmont 1:fdd22bb7aa52 247
emilmont 1:fdd22bb7aa52 248 /* If the filter length is not a multiple of 4, compute the remaining filter taps */
emilmont 1:fdd22bb7aa52 249 tapCnt = numTaps % 0x4u;
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 252 {
emilmont 1:fdd22bb7aa52 253 /* Read coefficients */
emilmont 1:fdd22bb7aa52 254 c0 = *pb++;
emilmont 1:fdd22bb7aa52 255
emilmont 1:fdd22bb7aa52 256 /* Fetch 1 state variable */
emilmont 1:fdd22bb7aa52 257 x0 = *px++;
emilmont 1:fdd22bb7aa52 258
emilmont 1:fdd22bb7aa52 259 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 260 sum0 = __SMLAD(x0, c0, sum0);
emilmont 1:fdd22bb7aa52 261
emilmont 1:fdd22bb7aa52 262 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 263 tapCnt--;
emilmont 1:fdd22bb7aa52 264 }
emilmont 1:fdd22bb7aa52 265
emilmont 1:fdd22bb7aa52 266 /* Advance the state pointer by the decimation factor
emilmont 1:fdd22bb7aa52 267 * to process the next group of decimation factor number samples */
emilmont 1:fdd22bb7aa52 268 pState = pState + S->M;
emilmont 1:fdd22bb7aa52 269
emilmont 1:fdd22bb7aa52 270 /* Store filter output, smlad returns the values in 2.14 format */
emilmont 1:fdd22bb7aa52 271 /* so downsacle by 15 to get output in 1.15 */
emilmont 1:fdd22bb7aa52 272 *pDst++ = (q15_t) (__SSAT((sum0 >> 15), 16));
emilmont 1:fdd22bb7aa52 273
emilmont 1:fdd22bb7aa52 274 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 275 blkCntN3--;
emilmont 1:fdd22bb7aa52 276 }
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* Processing is complete.
emilmont 1:fdd22bb7aa52 279 ** Now copy the last numTaps - 1 samples to the satrt of the state buffer.
emilmont 1:fdd22bb7aa52 280 ** This prepares the state buffer for the next function call. */
emilmont 1:fdd22bb7aa52 281
emilmont 1:fdd22bb7aa52 282 /* Points to the start of the state buffer */
emilmont 1:fdd22bb7aa52 283 pStateCurnt = S->pState;
emilmont 1:fdd22bb7aa52 284
emilmont 1:fdd22bb7aa52 285 i = (numTaps - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 286
emilmont 1:fdd22bb7aa52 287 /* copy data */
emilmont 1:fdd22bb7aa52 288 while(i > 0u)
emilmont 1:fdd22bb7aa52 289 {
emilmont 1:fdd22bb7aa52 290 *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++;
emilmont 1:fdd22bb7aa52 291 *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++;
emilmont 1:fdd22bb7aa52 292
emilmont 1:fdd22bb7aa52 293 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 294 i--;
emilmont 1:fdd22bb7aa52 295 }
emilmont 1:fdd22bb7aa52 296
emilmont 1:fdd22bb7aa52 297 i = (numTaps - 1u) % 0x04u;
emilmont 1:fdd22bb7aa52 298
emilmont 1:fdd22bb7aa52 299 /* copy data */
emilmont 1:fdd22bb7aa52 300 while(i > 0u)
emilmont 1:fdd22bb7aa52 301 {
emilmont 1:fdd22bb7aa52 302 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 305 i--;
emilmont 1:fdd22bb7aa52 306 }
emilmont 1:fdd22bb7aa52 307 }
emilmont 1:fdd22bb7aa52 308
emilmont 1:fdd22bb7aa52 309 #else
emilmont 1:fdd22bb7aa52 310
emilmont 1:fdd22bb7aa52 311
emilmont 1:fdd22bb7aa52 312 void arm_fir_decimate_fast_q15(
emilmont 1:fdd22bb7aa52 313 const arm_fir_decimate_instance_q15 * S,
emilmont 1:fdd22bb7aa52 314 q15_t * pSrc,
emilmont 1:fdd22bb7aa52 315 q15_t * pDst,
emilmont 1:fdd22bb7aa52 316 uint32_t blockSize)
emilmont 1:fdd22bb7aa52 317 {
emilmont 1:fdd22bb7aa52 318 q15_t *pState = S->pState; /* State pointer */
emilmont 1:fdd22bb7aa52 319 q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
emilmont 1:fdd22bb7aa52 320 q15_t *pStateCurnt; /* Points to the current sample of the state */
emilmont 1:fdd22bb7aa52 321 q15_t *px; /* Temporary pointer for state buffer */
emilmont 1:fdd22bb7aa52 322 q15_t *pb; /* Temporary pointer coefficient buffer */
emilmont 1:fdd22bb7aa52 323 q15_t x0, x1, c0; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 324 q31_t sum0; /* Accumulators */
emilmont 1:fdd22bb7aa52 325 q31_t acc0, acc1;
emilmont 1:fdd22bb7aa52 326 q15_t *px0, *px1;
emilmont 1:fdd22bb7aa52 327 uint32_t blkCntN3;
emilmont 1:fdd22bb7aa52 328 uint32_t numTaps = S->numTaps; /* Number of taps */
emilmont 1:fdd22bb7aa52 329 uint32_t i, blkCnt, tapCnt, outBlockSize = blockSize / S->M; /* Loop counters */
emilmont 1:fdd22bb7aa52 330
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332 /* S->pState buffer contains previous frame (numTaps - 1) samples */
emilmont 1:fdd22bb7aa52 333 /* pStateCurnt points to the location where the new input data should be written */
emilmont 1:fdd22bb7aa52 334 pStateCurnt = S->pState + (numTaps - 1u);
emilmont 1:fdd22bb7aa52 335
emilmont 1:fdd22bb7aa52 336
emilmont 1:fdd22bb7aa52 337 /* Total number of output samples to be computed */
emilmont 1:fdd22bb7aa52 338 blkCnt = outBlockSize / 2;
emilmont 1:fdd22bb7aa52 339 blkCntN3 = outBlockSize - (2 * blkCnt);
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 342 {
emilmont 1:fdd22bb7aa52 343 /* Copy decimation factor number of new input samples into the state buffer */
emilmont 1:fdd22bb7aa52 344 i = 2 * S->M;
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346 do
emilmont 1:fdd22bb7aa52 347 {
emilmont 1:fdd22bb7aa52 348 *pStateCurnt++ = *pSrc++;
emilmont 1:fdd22bb7aa52 349
emilmont 1:fdd22bb7aa52 350 } while(--i);
emilmont 1:fdd22bb7aa52 351
emilmont 1:fdd22bb7aa52 352 /* Set accumulator to zero */
emilmont 1:fdd22bb7aa52 353 acc0 = 0;
emilmont 1:fdd22bb7aa52 354 acc1 = 0;
emilmont 1:fdd22bb7aa52 355
emilmont 1:fdd22bb7aa52 356 /* Initialize state pointer */
emilmont 1:fdd22bb7aa52 357 px0 = pState;
emilmont 1:fdd22bb7aa52 358
emilmont 1:fdd22bb7aa52 359 px1 = pState + S->M;
emilmont 1:fdd22bb7aa52 360
emilmont 1:fdd22bb7aa52 361
emilmont 1:fdd22bb7aa52 362 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 363 pb = pCoeffs;
emilmont 1:fdd22bb7aa52 364
emilmont 1:fdd22bb7aa52 365 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 366 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 367
emilmont 1:fdd22bb7aa52 368 /* Loop over the number of taps. Unroll by a factor of 4.
emilmont 1:fdd22bb7aa52 369 ** Repeat until we've computed numTaps-4 coefficients. */
emilmont 1:fdd22bb7aa52 370 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 371 {
emilmont 1:fdd22bb7aa52 372 /* Read the Read b[numTaps-1] coefficients */
emilmont 1:fdd22bb7aa52 373 c0 = *pb++;
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 /* Read x[n-numTaps-1] for sample 0 and for sample 1 */
emilmont 1:fdd22bb7aa52 376 x0 = *px0++;
emilmont 1:fdd22bb7aa52 377 x1 = *px1++;
emilmont 1:fdd22bb7aa52 378
emilmont 1:fdd22bb7aa52 379 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 380 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 381 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 382
emilmont 1:fdd22bb7aa52 383 /* Read the b[numTaps-2] coefficient */
emilmont 1:fdd22bb7aa52 384 c0 = *pb++;
emilmont 1:fdd22bb7aa52 385
emilmont 1:fdd22bb7aa52 386 /* Read x[n-numTaps-2] for sample 0 and sample 1 */
emilmont 1:fdd22bb7aa52 387 x0 = *px0++;
emilmont 1:fdd22bb7aa52 388 x1 = *px1++;
emilmont 1:fdd22bb7aa52 389
emilmont 1:fdd22bb7aa52 390 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 391 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 392 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 393
emilmont 1:fdd22bb7aa52 394 /* Read the b[numTaps-3] coefficients */
emilmont 1:fdd22bb7aa52 395 c0 = *pb++;
emilmont 1:fdd22bb7aa52 396
emilmont 1:fdd22bb7aa52 397 /* Read x[n-numTaps-3] for sample 0 and sample 1 */
emilmont 1:fdd22bb7aa52 398 x0 = *px0++;
emilmont 1:fdd22bb7aa52 399 x1 = *px1++;
emilmont 1:fdd22bb7aa52 400
emilmont 1:fdd22bb7aa52 401 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 402 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 403 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 404
emilmont 1:fdd22bb7aa52 405 /* Read the b[numTaps-4] coefficient */
emilmont 1:fdd22bb7aa52 406 c0 = *pb++;
emilmont 1:fdd22bb7aa52 407
emilmont 1:fdd22bb7aa52 408 /* Read x[n-numTaps-4] for sample 0 and sample 1 */
emilmont 1:fdd22bb7aa52 409 x0 = *px0++;
emilmont 1:fdd22bb7aa52 410 x1 = *px1++;
emilmont 1:fdd22bb7aa52 411
emilmont 1:fdd22bb7aa52 412 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 413 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 414 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 415
emilmont 1:fdd22bb7aa52 416 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 417 tapCnt--;
emilmont 1:fdd22bb7aa52 418 }
emilmont 1:fdd22bb7aa52 419
emilmont 1:fdd22bb7aa52 420 /* If the filter length is not a multiple of 4, compute the remaining filter taps */
emilmont 1:fdd22bb7aa52 421 tapCnt = numTaps % 0x4u;
emilmont 1:fdd22bb7aa52 422
emilmont 1:fdd22bb7aa52 423 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 424 {
emilmont 1:fdd22bb7aa52 425 /* Read coefficients */
emilmont 1:fdd22bb7aa52 426 c0 = *pb++;
emilmont 1:fdd22bb7aa52 427
emilmont 1:fdd22bb7aa52 428 /* Fetch 1 state variable */
emilmont 1:fdd22bb7aa52 429 x0 = *px0++;
emilmont 1:fdd22bb7aa52 430 x1 = *px1++;
emilmont 1:fdd22bb7aa52 431
emilmont 1:fdd22bb7aa52 432 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 433 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 434 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 435
emilmont 1:fdd22bb7aa52 436 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 437 tapCnt--;
emilmont 1:fdd22bb7aa52 438 }
emilmont 1:fdd22bb7aa52 439
emilmont 1:fdd22bb7aa52 440 /* Advance the state pointer by the decimation factor
emilmont 1:fdd22bb7aa52 441 * to process the next group of decimation factor number samples */
emilmont 1:fdd22bb7aa52 442 pState = pState + S->M * 2;
emilmont 1:fdd22bb7aa52 443
emilmont 1:fdd22bb7aa52 444 /* Store filter output, smlad returns the values in 2.14 format */
emilmont 1:fdd22bb7aa52 445 /* so downsacle by 15 to get output in 1.15 */
emilmont 1:fdd22bb7aa52 446
emilmont 1:fdd22bb7aa52 447 *pDst++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 448 *pDst++ = (q15_t) (__SSAT((acc1 >> 15), 16));
emilmont 1:fdd22bb7aa52 449
emilmont 1:fdd22bb7aa52 450
emilmont 1:fdd22bb7aa52 451 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 452 blkCnt--;
emilmont 1:fdd22bb7aa52 453 }
emilmont 1:fdd22bb7aa52 454
emilmont 1:fdd22bb7aa52 455 while(blkCntN3 > 0u)
emilmont 1:fdd22bb7aa52 456 {
emilmont 1:fdd22bb7aa52 457 /* Copy decimation factor number of new input samples into the state buffer */
emilmont 1:fdd22bb7aa52 458 i = S->M;
emilmont 1:fdd22bb7aa52 459
emilmont 1:fdd22bb7aa52 460 do
emilmont 1:fdd22bb7aa52 461 {
emilmont 1:fdd22bb7aa52 462 *pStateCurnt++ = *pSrc++;
emilmont 1:fdd22bb7aa52 463
emilmont 1:fdd22bb7aa52 464 } while(--i);
emilmont 1:fdd22bb7aa52 465
emilmont 1:fdd22bb7aa52 466 /*Set sum to zero */
emilmont 1:fdd22bb7aa52 467 sum0 = 0;
emilmont 1:fdd22bb7aa52 468
emilmont 1:fdd22bb7aa52 469 /* Initialize state pointer */
emilmont 1:fdd22bb7aa52 470 px = pState;
emilmont 1:fdd22bb7aa52 471
emilmont 1:fdd22bb7aa52 472 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 473 pb = pCoeffs;
emilmont 1:fdd22bb7aa52 474
emilmont 1:fdd22bb7aa52 475 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 476 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 477
emilmont 1:fdd22bb7aa52 478 /* Loop over the number of taps. Unroll by a factor of 4.
emilmont 1:fdd22bb7aa52 479 ** Repeat until we've computed numTaps-4 coefficients. */
emilmont 1:fdd22bb7aa52 480 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 481 {
emilmont 1:fdd22bb7aa52 482 /* Read the Read b[numTaps-1] coefficients */
emilmont 1:fdd22bb7aa52 483 c0 = *pb++;
emilmont 1:fdd22bb7aa52 484
emilmont 1:fdd22bb7aa52 485 /* Read x[n-numTaps-1] and sample */
emilmont 1:fdd22bb7aa52 486 x0 = *px++;
emilmont 1:fdd22bb7aa52 487
emilmont 1:fdd22bb7aa52 488 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 489 sum0 += x0 * c0;
emilmont 1:fdd22bb7aa52 490
emilmont 1:fdd22bb7aa52 491 /* Read the b[numTaps-2] coefficient */
emilmont 1:fdd22bb7aa52 492 c0 = *pb++;
emilmont 1:fdd22bb7aa52 493
emilmont 1:fdd22bb7aa52 494 /* Read x[n-numTaps-2] and sample */
emilmont 1:fdd22bb7aa52 495 x0 = *px++;
emilmont 1:fdd22bb7aa52 496
emilmont 1:fdd22bb7aa52 497 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 498 sum0 += x0 * c0;
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 /* Read the b[numTaps-3] coefficients */
emilmont 1:fdd22bb7aa52 501 c0 = *pb++;
emilmont 1:fdd22bb7aa52 502
emilmont 1:fdd22bb7aa52 503 /* Read x[n-numTaps-3] sample */
emilmont 1:fdd22bb7aa52 504 x0 = *px++;
emilmont 1:fdd22bb7aa52 505
emilmont 1:fdd22bb7aa52 506 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 507 sum0 += x0 * c0;
emilmont 1:fdd22bb7aa52 508
emilmont 1:fdd22bb7aa52 509 /* Read the b[numTaps-4] coefficient */
emilmont 1:fdd22bb7aa52 510 c0 = *pb++;
emilmont 1:fdd22bb7aa52 511
emilmont 1:fdd22bb7aa52 512 /* Read x[n-numTaps-4] sample */
emilmont 1:fdd22bb7aa52 513 x0 = *px++;
emilmont 1:fdd22bb7aa52 514
emilmont 1:fdd22bb7aa52 515 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 516 sum0 += x0 * c0;
emilmont 1:fdd22bb7aa52 517
emilmont 1:fdd22bb7aa52 518 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 519 tapCnt--;
emilmont 1:fdd22bb7aa52 520 }
emilmont 1:fdd22bb7aa52 521
emilmont 1:fdd22bb7aa52 522 /* If the filter length is not a multiple of 4, compute the remaining filter taps */
emilmont 1:fdd22bb7aa52 523 tapCnt = numTaps % 0x4u;
emilmont 1:fdd22bb7aa52 524
emilmont 1:fdd22bb7aa52 525 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 526 {
emilmont 1:fdd22bb7aa52 527 /* Read coefficients */
emilmont 1:fdd22bb7aa52 528 c0 = *pb++;
emilmont 1:fdd22bb7aa52 529
emilmont 1:fdd22bb7aa52 530 /* Fetch 1 state variable */
emilmont 1:fdd22bb7aa52 531 x0 = *px++;
emilmont 1:fdd22bb7aa52 532
emilmont 1:fdd22bb7aa52 533 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 534 sum0 += x0 * c0;
emilmont 1:fdd22bb7aa52 535
emilmont 1:fdd22bb7aa52 536 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 537 tapCnt--;
emilmont 1:fdd22bb7aa52 538 }
emilmont 1:fdd22bb7aa52 539
emilmont 1:fdd22bb7aa52 540 /* Advance the state pointer by the decimation factor
emilmont 1:fdd22bb7aa52 541 * to process the next group of decimation factor number samples */
emilmont 1:fdd22bb7aa52 542 pState = pState + S->M;
emilmont 1:fdd22bb7aa52 543
emilmont 1:fdd22bb7aa52 544 /* Store filter output, smlad returns the values in 2.14 format */
emilmont 1:fdd22bb7aa52 545 /* so downsacle by 15 to get output in 1.15 */
emilmont 1:fdd22bb7aa52 546 *pDst++ = (q15_t) (__SSAT((sum0 >> 15), 16));
emilmont 1:fdd22bb7aa52 547
emilmont 1:fdd22bb7aa52 548 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 549 blkCntN3--;
emilmont 1:fdd22bb7aa52 550 }
emilmont 1:fdd22bb7aa52 551
emilmont 1:fdd22bb7aa52 552 /* Processing is complete.
emilmont 1:fdd22bb7aa52 553 ** Now copy the last numTaps - 1 samples to the satrt of the state buffer.
emilmont 1:fdd22bb7aa52 554 ** This prepares the state buffer for the next function call. */
emilmont 1:fdd22bb7aa52 555
emilmont 1:fdd22bb7aa52 556 /* Points to the start of the state buffer */
emilmont 1:fdd22bb7aa52 557 pStateCurnt = S->pState;
emilmont 1:fdd22bb7aa52 558
emilmont 1:fdd22bb7aa52 559 i = (numTaps - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 560
emilmont 1:fdd22bb7aa52 561 /* copy data */
emilmont 1:fdd22bb7aa52 562 while(i > 0u)
emilmont 1:fdd22bb7aa52 563 {
emilmont 1:fdd22bb7aa52 564 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 565 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 566 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 567 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 568
emilmont 1:fdd22bb7aa52 569 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 570 i--;
emilmont 1:fdd22bb7aa52 571 }
emilmont 1:fdd22bb7aa52 572
emilmont 1:fdd22bb7aa52 573 i = (numTaps - 1u) % 0x04u;
emilmont 1:fdd22bb7aa52 574
emilmont 1:fdd22bb7aa52 575 /* copy data */
emilmont 1:fdd22bb7aa52 576 while(i > 0u)
emilmont 1:fdd22bb7aa52 577 {
emilmont 1:fdd22bb7aa52 578 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 579
emilmont 1:fdd22bb7aa52 580 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 581 i--;
emilmont 1:fdd22bb7aa52 582 }
emilmont 1:fdd22bb7aa52 583 }
emilmont 1:fdd22bb7aa52 584
emilmont 1:fdd22bb7aa52 585
emilmont 1:fdd22bb7aa52 586 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 587
emilmont 1:fdd22bb7aa52 588 /**
emilmont 1:fdd22bb7aa52 589 * @} end of FIR_decimate group
emilmont 1:fdd22bb7aa52 590 */