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_correlate_fast_opt_q15.c@5:3762170b6d4d, 2015-11-20 (annotated)

Committer:
mbed_official
Date:
Fri Nov 20 08:45:18 2015 +0000
Revision:
5:3762170b6d4d
Parent:
3:7a284390b0ce 
Synchronized with git revision 2eb940b9a73af188d3004a2575fdfbb05febe62b



Full URL: https://github.com/mbedmicro/mbed/commit/2eb940b9a73af188d3004a2575fdfbb05febe62b/



Added option to build rpc library. closes #1426

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
mbed_official 5:3762170b6d4d 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
mbed_official 5:3762170b6d4d 4 * $Date: 19. March 2015
mbed_official 5:3762170b6d4d 5 * $Revision: V.1.4.5
emilmont 1:fdd22bb7aa52 6 *
emilmont 2:da51fb522205 7 * Project: CMSIS DSP Library
emilmont 2:da51fb522205 8 * Title: arm_correlate_fast_opt_q15.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Fast Q15 Correlation.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3
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 Corr
emilmont 1:fdd22bb7aa52 49 * @{
emilmont 1:fdd22bb7aa52 50 */
emilmont 1:fdd22bb7aa52 51
emilmont 1:fdd22bb7aa52 52 /**
emilmont 1:fdd22bb7aa52 53 * @brief Correlation of Q15 sequences (fast version) for Cortex-M3 and Cortex-M4.
emilmont 1:fdd22bb7aa52 54 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 55 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 56 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 57 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 58 * @param[out] *pDst points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1.
emilmont 1:fdd22bb7aa52 59 * @param[in] *pScratch points to scratch buffer of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 60 * @return none.
emilmont 1:fdd22bb7aa52 61 *
emilmont 1:fdd22bb7aa52 62 *
emilmont 1:fdd22bb7aa52 63 * \par Restrictions
emilmont 1:fdd22bb7aa52 64 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 65 * In this case input, output, scratch buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 66 *
emilmont 1:fdd22bb7aa52 67 *
emilmont 1:fdd22bb7aa52 68 * <b>Scaling and Overflow Behavior:</b>
emilmont 1:fdd22bb7aa52 69 *
emilmont 1:fdd22bb7aa52 70 * \par
emilmont 1:fdd22bb7aa52 71 * This fast version uses a 32-bit accumulator with 2.30 format.
emilmont 1:fdd22bb7aa52 72 * The accumulator maintains full precision of the intermediate multiplication results but provides only a single guard bit.
emilmont 1:fdd22bb7aa52 73 * There is no saturation on intermediate additions.
emilmont 1:fdd22bb7aa52 74 * Thus, if the accumulator overflows it wraps around and distorts the result.
emilmont 1:fdd22bb7aa52 75 * The input signals should be scaled down to avoid intermediate overflows.
emilmont 1:fdd22bb7aa52 76 * Scale down one of the inputs by 1/min(srcALen, srcBLen) to avoid overflow since a
emilmont 1:fdd22bb7aa52 77 * maximum of min(srcALen, srcBLen) number of additions is carried internally.
emilmont 1:fdd22bb7aa52 78 * The 2.30 accumulator is right shifted by 15 bits and then saturated to 1.15 format to yield the final result.
emilmont 1:fdd22bb7aa52 79 *
emilmont 1:fdd22bb7aa52 80 * \par
emilmont 1:fdd22bb7aa52 81 * See <code>arm_correlate_q15()</code> for a slower implementation of this function which uses a 64-bit accumulator to avoid wrap around distortion.
emilmont 1:fdd22bb7aa52 82 */
emilmont 1:fdd22bb7aa52 83
emilmont 1:fdd22bb7aa52 84 void arm_correlate_fast_opt_q15(
emilmont 1:fdd22bb7aa52 85 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 86 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 87 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 88 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 89 q15_t * pDst,
emilmont 1:fdd22bb7aa52 90 q15_t * pScratch)
emilmont 1:fdd22bb7aa52 91 {
emilmont 1:fdd22bb7aa52 92 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 93 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 94 q31_t acc0, acc1, acc2, acc3; /* Accumulators */
emilmont 1:fdd22bb7aa52 95 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 96 q31_t x1, x2, x3; /* temporary variables for holding input and coefficient values */
emilmont 1:fdd22bb7aa52 97 uint32_t j, blkCnt, outBlockSize; /* loop counter */
emilmont 1:fdd22bb7aa52 98 int32_t inc = 1; /* Destination address modifier */
emilmont 1:fdd22bb7aa52 99 uint32_t tapCnt;
emilmont 1:fdd22bb7aa52 100 q31_t y1, y2;
emilmont 1:fdd22bb7aa52 101 q15_t *pScr; /* Intermediate pointers */
emilmont 1:fdd22bb7aa52 102 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 103 #ifdef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 104
emilmont 1:fdd22bb7aa52 105 q15_t a, b;
emilmont 1:fdd22bb7aa52 106
emilmont 2:da51fb522205 107 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 108
emilmont 1:fdd22bb7aa52 109 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 110 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 111 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 112 /* But CORR(x, y) is reverse of CORR(y, x) */
emilmont 1:fdd22bb7aa52 113 /* So, when srcBLen > srcALen, output pointer is made to point to the end of the output buffer */
emilmont 1:fdd22bb7aa52 114 /* and the destination pointer modifier, inc is set to -1 */
emilmont 1:fdd22bb7aa52 115 /* If srcALen > srcBLen, zero pad has to be done to srcB to make the two inputs of same length */
emilmont 1:fdd22bb7aa52 116 /* But to improve the performance,
emilmont 1:fdd22bb7aa52 117 * we include zeroes in the output instead of zero padding either of the the inputs*/
emilmont 1:fdd22bb7aa52 118 /* If srcALen > srcBLen,
emilmont 1:fdd22bb7aa52 119 * (srcALen - srcBLen) zeroes has to included in the starting of the output buffer */
emilmont 1:fdd22bb7aa52 120 /* If srcALen < srcBLen,
emilmont 1:fdd22bb7aa52 121 * (srcALen - srcBLen) zeroes has to included in the ending of the output buffer */
emilmont 1:fdd22bb7aa52 122 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 123 {
emilmont 1:fdd22bb7aa52 124 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 125 pIn1 = (pSrcA);
emilmont 1:fdd22bb7aa52 126
emilmont 1:fdd22bb7aa52 127 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 128 pIn2 = (pSrcB);
emilmont 1:fdd22bb7aa52 129
emilmont 1:fdd22bb7aa52 130 /* Number of output samples is calculated */
emilmont 1:fdd22bb7aa52 131 outBlockSize = (2u * srcALen) - 1u;
emilmont 1:fdd22bb7aa52 132
emilmont 1:fdd22bb7aa52 133 /* When srcALen > srcBLen, zero padding is done to srcB
emilmont 1:fdd22bb7aa52 134 * to make their lengths equal.
emilmont 1:fdd22bb7aa52 135 * Instead, (outBlockSize - (srcALen + srcBLen - 1))
emilmont 1:fdd22bb7aa52 136 * number of output samples are made zero */
emilmont 1:fdd22bb7aa52 137 j = outBlockSize - (srcALen + (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 138
emilmont 1:fdd22bb7aa52 139 /* Updating the pointer position to non zero value */
emilmont 1:fdd22bb7aa52 140 pOut += j;
emilmont 1:fdd22bb7aa52 141
emilmont 1:fdd22bb7aa52 142 }
emilmont 1:fdd22bb7aa52 143 else
emilmont 1:fdd22bb7aa52 144 {
emilmont 1:fdd22bb7aa52 145 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 146 pIn1 = (pSrcB);
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 149 pIn2 = (pSrcA);
emilmont 1:fdd22bb7aa52 150
emilmont 1:fdd22bb7aa52 151 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 152 j = srcBLen;
emilmont 1:fdd22bb7aa52 153 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 154 srcALen = j;
emilmont 1:fdd22bb7aa52 155
emilmont 1:fdd22bb7aa52 156 /* CORR(x, y) = Reverse order(CORR(y, x)) */
emilmont 1:fdd22bb7aa52 157 /* Hence set the destination pointer to point to the last output sample */
emilmont 1:fdd22bb7aa52 158 pOut = pDst + ((srcALen + srcBLen) - 2u);
emilmont 1:fdd22bb7aa52 159
emilmont 1:fdd22bb7aa52 160 /* Destination address modifier is set to -1 */
emilmont 1:fdd22bb7aa52 161 inc = -1;
emilmont 1:fdd22bb7aa52 162
emilmont 1:fdd22bb7aa52 163 }
emilmont 1:fdd22bb7aa52 164
emilmont 1:fdd22bb7aa52 165 pScr = pScratch;
emilmont 1:fdd22bb7aa52 166
emilmont 1:fdd22bb7aa52 167 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 168 arm_fill_q15(0, pScr, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 169
emilmont 1:fdd22bb7aa52 170 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 171 pScr += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 172
emilmont 1:fdd22bb7aa52 173 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 176 arm_copy_q15(pIn1, pScr, srcALen);
emilmont 1:fdd22bb7aa52 177
emilmont 1:fdd22bb7aa52 178 /* Update pointers */
emilmont 1:fdd22bb7aa52 179 pScr += srcALen;
emilmont 1:fdd22bb7aa52 180
emilmont 1:fdd22bb7aa52 181 #else
emilmont 1:fdd22bb7aa52 182
emilmont 1:fdd22bb7aa52 183 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 184 j = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 185
emilmont 1:fdd22bb7aa52 186 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 187 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 188 while(j > 0u)
emilmont 1:fdd22bb7aa52 189 {
emilmont 1:fdd22bb7aa52 190 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 191 *pScr++ = *pIn1++;
emilmont 1:fdd22bb7aa52 192 *pScr++ = *pIn1++;
emilmont 1:fdd22bb7aa52 193 *pScr++ = *pIn1++;
emilmont 1:fdd22bb7aa52 194 *pScr++ = *pIn1++;
emilmont 1:fdd22bb7aa52 195
emilmont 1:fdd22bb7aa52 196 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 197 j--;
emilmont 1:fdd22bb7aa52 198 }
emilmont 1:fdd22bb7aa52 199
emilmont 1:fdd22bb7aa52 200 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 201 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 202 j = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 while(j > 0u)
emilmont 1:fdd22bb7aa52 205 {
emilmont 1:fdd22bb7aa52 206 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 207 *pScr++ = *pIn1++;
emilmont 1:fdd22bb7aa52 208
emilmont 1:fdd22bb7aa52 209 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 210 j--;
emilmont 1:fdd22bb7aa52 211 }
emilmont 1:fdd22bb7aa52 212
emilmont 2:da51fb522205 213 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 214
emilmont 1:fdd22bb7aa52 215 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 216
emilmont 1:fdd22bb7aa52 217 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 218 arm_fill_q15(0, pScr, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 219
emilmont 1:fdd22bb7aa52 220 /* Update pointer */
emilmont 1:fdd22bb7aa52 221 pScr += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 222
emilmont 1:fdd22bb7aa52 223 #else
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 226 j = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 229 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 230 while(j > 0u)
emilmont 1:fdd22bb7aa52 231 {
emilmont 1:fdd22bb7aa52 232 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 233 *pScr++ = 0;
emilmont 1:fdd22bb7aa52 234 *pScr++ = 0;
emilmont 1:fdd22bb7aa52 235 *pScr++ = 0;
emilmont 1:fdd22bb7aa52 236 *pScr++ = 0;
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 239 j--;
emilmont 1:fdd22bb7aa52 240 }
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 243 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 244 j = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 245
emilmont 1:fdd22bb7aa52 246 while(j > 0u)
emilmont 1:fdd22bb7aa52 247 {
emilmont 1:fdd22bb7aa52 248 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 249 *pScr++ = 0;
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 252 j--;
emilmont 1:fdd22bb7aa52 253 }
emilmont 1:fdd22bb7aa52 254
emilmont 2:da51fb522205 255 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 256
emilmont 1:fdd22bb7aa52 257 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 258 py = pIn2;
emilmont 1:fdd22bb7aa52 259
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 /* Actual correlation process starts here */
emilmont 1:fdd22bb7aa52 262 blkCnt = (srcALen + srcBLen - 1u) >> 2;
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 265 {
emilmont 1:fdd22bb7aa52 266 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 267 pScr = pScratch;
emilmont 1:fdd22bb7aa52 268
emilmont 1:fdd22bb7aa52 269 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 270 acc0 = 0;
emilmont 1:fdd22bb7aa52 271 acc1 = 0;
emilmont 1:fdd22bb7aa52 272 acc2 = 0;
emilmont 1:fdd22bb7aa52 273 acc3 = 0;
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* Read four samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 276 x1 = *__SIMD32(pScr)++;
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* Read next four samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 279 x2 = *__SIMD32(pScr)++;
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 282
emilmont 1:fdd22bb7aa52 283 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 284 {
emilmont 1:fdd22bb7aa52 285
emilmont 1:fdd22bb7aa52 286 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 287
emilmont 1:fdd22bb7aa52 288 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 289 y1 = _SIMD32_OFFSET(pIn2);
emilmont 1:fdd22bb7aa52 290 y2 = _SIMD32_OFFSET(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 291
emilmont 1:fdd22bb7aa52 292 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 293
emilmont 1:fdd22bb7aa52 294 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 295
emilmont 1:fdd22bb7aa52 296 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 297 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 298 #else
emilmont 1:fdd22bb7aa52 299 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 300 #endif
emilmont 1:fdd22bb7aa52 301
emilmont 1:fdd22bb7aa52 302 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 x1 = _SIMD32_OFFSET(pScr);
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 acc0 = __SMLAD(x2, y2, acc0);
emilmont 1:fdd22bb7aa52 307
emilmont 1:fdd22bb7aa52 308 acc2 = __SMLAD(x1, y2, acc2);
emilmont 1:fdd22bb7aa52 309
emilmont 1:fdd22bb7aa52 310 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 311 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 312 #else
emilmont 1:fdd22bb7aa52 313 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 314 #endif
emilmont 1:fdd22bb7aa52 315
emilmont 1:fdd22bb7aa52 316 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 317
emilmont 1:fdd22bb7aa52 318 acc1 = __SMLADX(x3, y2, acc1);
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320 x2 = _SIMD32_OFFSET(pScr + 2u);
emilmont 1:fdd22bb7aa52 321
emilmont 1:fdd22bb7aa52 322 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 323 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 324 #else
emilmont 1:fdd22bb7aa52 325 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 326 #endif
emilmont 1:fdd22bb7aa52 327
emilmont 1:fdd22bb7aa52 328 acc3 = __SMLADX(x3, y2, acc3);
emilmont 2:da51fb522205 329 #else
emilmont 1:fdd22bb7aa52 330
emilmont 1:fdd22bb7aa52 331 /* Read four samples from smaller buffer */
emilmont 2:da51fb522205 332 a = *pIn2;
emilmont 2:da51fb522205 333 b = *(pIn2 + 1);
emilmont 1:fdd22bb7aa52 334
emilmont 1:fdd22bb7aa52 335 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 336 y1 = __PKHBT(a, b, 16);
emilmont 1:fdd22bb7aa52 337 #else
emilmont 1:fdd22bb7aa52 338 y1 = __PKHBT(b, a, 16);
emilmont 1:fdd22bb7aa52 339 #endif
emilmont 2:da51fb522205 340
emilmont 2:da51fb522205 341 a = *(pIn2 + 2);
emilmont 2:da51fb522205 342 b = *(pIn2 + 3);
emilmont 1:fdd22bb7aa52 343 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 344 y2 = __PKHBT(a, b, 16);
emilmont 1:fdd22bb7aa52 345 #else
emilmont 1:fdd22bb7aa52 346 y2 = __PKHBT(b, a, 16);
emilmont 2:da51fb522205 347 #endif
emilmont 1:fdd22bb7aa52 348
emilmont 1:fdd22bb7aa52 349 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 350
emilmont 1:fdd22bb7aa52 351 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 352
emilmont 1:fdd22bb7aa52 353 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 354 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 355 #else
emilmont 1:fdd22bb7aa52 356 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 357 #endif
emilmont 1:fdd22bb7aa52 358
emilmont 1:fdd22bb7aa52 359 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 360
emilmont 2:da51fb522205 361 a = *pScr;
emilmont 2:da51fb522205 362 b = *(pScr + 1);
emilmont 1:fdd22bb7aa52 363
emilmont 1:fdd22bb7aa52 364 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 365 x1 = __PKHBT(a, b, 16);
emilmont 1:fdd22bb7aa52 366 #else
emilmont 1:fdd22bb7aa52 367 x1 = __PKHBT(b, a, 16);
emilmont 1:fdd22bb7aa52 368 #endif
emilmont 1:fdd22bb7aa52 369
emilmont 1:fdd22bb7aa52 370 acc0 = __SMLAD(x2, y2, acc0);
emilmont 1:fdd22bb7aa52 371
emilmont 1:fdd22bb7aa52 372 acc2 = __SMLAD(x1, y2, acc2);
emilmont 1:fdd22bb7aa52 373
emilmont 1:fdd22bb7aa52 374 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 375 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 376 #else
emilmont 1:fdd22bb7aa52 377 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 378 #endif
emilmont 1:fdd22bb7aa52 379
emilmont 1:fdd22bb7aa52 380 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382 acc1 = __SMLADX(x3, y2, acc1);
emilmont 1:fdd22bb7aa52 383
emilmont 2:da51fb522205 384 a = *(pScr + 2);
emilmont 2:da51fb522205 385 b = *(pScr + 3);
emilmont 1:fdd22bb7aa52 386
emilmont 1:fdd22bb7aa52 387 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 388 x2 = __PKHBT(a, b, 16);
emilmont 1:fdd22bb7aa52 389 #else
emilmont 1:fdd22bb7aa52 390 x2 = __PKHBT(b, a, 16);
emilmont 1:fdd22bb7aa52 391 #endif
emilmont 1:fdd22bb7aa52 392
emilmont 1:fdd22bb7aa52 393 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 394 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 395 #else
emilmont 1:fdd22bb7aa52 396 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 397 #endif
emilmont 1:fdd22bb7aa52 398
emilmont 1:fdd22bb7aa52 399 acc3 = __SMLADX(x3, y2, acc3);
emilmont 1:fdd22bb7aa52 400
emilmont 2:da51fb522205 401 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 404
emilmont 1:fdd22bb7aa52 405 pScr += 4u;
emilmont 1:fdd22bb7aa52 406
emilmont 1:fdd22bb7aa52 407
emilmont 1:fdd22bb7aa52 408 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 409 tapCnt--;
emilmont 1:fdd22bb7aa52 410 }
emilmont 1:fdd22bb7aa52 411
emilmont 1:fdd22bb7aa52 412
emilmont 1:fdd22bb7aa52 413
emilmont 1:fdd22bb7aa52 414 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 415 pScr -= 4u;
emilmont 1:fdd22bb7aa52 416
emilmont 1:fdd22bb7aa52 417
emilmont 1:fdd22bb7aa52 418 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 419 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 420
emilmont 1:fdd22bb7aa52 421 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 422 {
emilmont 1:fdd22bb7aa52 423
emilmont 1:fdd22bb7aa52 424 /* accumlate the results */
emilmont 1:fdd22bb7aa52 425 acc0 += (*pScr++ * *pIn2);
emilmont 1:fdd22bb7aa52 426 acc1 += (*pScr++ * *pIn2);
emilmont 1:fdd22bb7aa52 427 acc2 += (*pScr++ * *pIn2);
emilmont 1:fdd22bb7aa52 428 acc3 += (*pScr++ * *pIn2++);
emilmont 1:fdd22bb7aa52 429
emilmont 1:fdd22bb7aa52 430 pScr -= 3u;
emilmont 1:fdd22bb7aa52 431
emilmont 1:fdd22bb7aa52 432 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 433 tapCnt--;
emilmont 1:fdd22bb7aa52 434 }
emilmont 1:fdd22bb7aa52 435
emilmont 1:fdd22bb7aa52 436 blkCnt--;
emilmont 1:fdd22bb7aa52 437
emilmont 1:fdd22bb7aa52 438
emilmont 1:fdd22bb7aa52 439 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 440 *pOut = (__SSAT(acc0 >> 15u, 16));
emilmont 1:fdd22bb7aa52 441 pOut += inc;
emilmont 1:fdd22bb7aa52 442 *pOut = (__SSAT(acc1 >> 15u, 16));
emilmont 1:fdd22bb7aa52 443 pOut += inc;
emilmont 1:fdd22bb7aa52 444 *pOut = (__SSAT(acc2 >> 15u, 16));
emilmont 1:fdd22bb7aa52 445 pOut += inc;
emilmont 1:fdd22bb7aa52 446 *pOut = (__SSAT(acc3 >> 15u, 16));
emilmont 1:fdd22bb7aa52 447 pOut += inc;
emilmont 1:fdd22bb7aa52 448
emilmont 1:fdd22bb7aa52 449
emilmont 1:fdd22bb7aa52 450 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 451 pIn2 = py;
emilmont 1:fdd22bb7aa52 452
emilmont 1:fdd22bb7aa52 453 pScratch += 4u;
emilmont 1:fdd22bb7aa52 454
emilmont 1:fdd22bb7aa52 455 }
emilmont 1:fdd22bb7aa52 456
emilmont 1:fdd22bb7aa52 457
emilmont 1:fdd22bb7aa52 458 blkCnt = (srcALen + srcBLen - 1u) & 0x3;
emilmont 1:fdd22bb7aa52 459
emilmont 1:fdd22bb7aa52 460 /* Calculate correlation for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 461 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 462 {
emilmont 1:fdd22bb7aa52 463 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 464 pScr = pScratch;
emilmont 1:fdd22bb7aa52 465
emilmont 1:fdd22bb7aa52 466 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 467 acc0 = 0;
emilmont 1:fdd22bb7aa52 468
emilmont 1:fdd22bb7aa52 469 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 470
emilmont 1:fdd22bb7aa52 471 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 472 {
emilmont 1:fdd22bb7aa52 473
emilmont 1:fdd22bb7aa52 474 acc0 += (*pScr++ * *pIn2++);
emilmont 1:fdd22bb7aa52 475 acc0 += (*pScr++ * *pIn2++);
emilmont 1:fdd22bb7aa52 476
emilmont 1:fdd22bb7aa52 477 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 478 tapCnt--;
emilmont 1:fdd22bb7aa52 479 }
emilmont 1:fdd22bb7aa52 480
emilmont 1:fdd22bb7aa52 481 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 482
emilmont 1:fdd22bb7aa52 483 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 484 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 485 {
emilmont 1:fdd22bb7aa52 486
emilmont 1:fdd22bb7aa52 487 /* accumlate the results */
emilmont 1:fdd22bb7aa52 488 acc0 += (*pScr++ * *pIn2++);
emilmont 1:fdd22bb7aa52 489
emilmont 1:fdd22bb7aa52 490 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 491 tapCnt--;
emilmont 1:fdd22bb7aa52 492 }
emilmont 1:fdd22bb7aa52 493
emilmont 1:fdd22bb7aa52 494 blkCnt--;
emilmont 1:fdd22bb7aa52 495
emilmont 1:fdd22bb7aa52 496 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 497
emilmont 1:fdd22bb7aa52 498 *pOut = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 pOut += inc;
emilmont 1:fdd22bb7aa52 501
emilmont 1:fdd22bb7aa52 502 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 503 pIn2 = py;
emilmont 1:fdd22bb7aa52 504
emilmont 1:fdd22bb7aa52 505 pScratch += 1u;
emilmont 1:fdd22bb7aa52 506
emilmont 1:fdd22bb7aa52 507 }
emilmont 1:fdd22bb7aa52 508 }
emilmont 1:fdd22bb7aa52 509
emilmont 1:fdd22bb7aa52 510 /**
emilmont 1:fdd22bb7aa52 511 * @} end of Corr group
emilmont 1:fdd22bb7aa52 512 */