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_opt_q7.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_opt_q7.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Correlation of Q7 sequences.
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 Q7 sequences.
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] *pScratch1 points to scratch buffer(of type q15_t) of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 60 * @param[in] *pScratch2 points to scratch buffer (of type q15_t) of size min(srcALen, srcBLen).
emilmont 1:fdd22bb7aa52 61 * @return none.
emilmont 1:fdd22bb7aa52 62 *
emilmont 1:fdd22bb7aa52 63 *
emilmont 1:fdd22bb7aa52 64 * \par Restrictions
emilmont 1:fdd22bb7aa52 65 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 66 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 67 *
emilmont 1:fdd22bb7aa52 68 * @details
emilmont 1:fdd22bb7aa52 69 * <b>Scaling and Overflow Behavior:</b>
emilmont 1:fdd22bb7aa52 70 *
emilmont 1:fdd22bb7aa52 71 * \par
emilmont 1:fdd22bb7aa52 72 * The function is implemented using a 32-bit internal accumulator.
emilmont 1:fdd22bb7aa52 73 * Both the inputs are represented in 1.7 format and multiplications yield a 2.14 result.
emilmont 1:fdd22bb7aa52 74 * The 2.14 intermediate results are accumulated in a 32-bit accumulator in 18.14 format.
emilmont 1:fdd22bb7aa52 75 * This approach provides 17 guard bits and there is no risk of overflow as long as <code>max(srcALen, srcBLen)<131072</code>.
emilmont 1:fdd22bb7aa52 76 * The 18.14 result is then truncated to 18.7 format by discarding the low 7 bits and saturated to 1.7 format.
emilmont 1:fdd22bb7aa52 77 *
emilmont 1:fdd22bb7aa52 78 *
emilmont 1:fdd22bb7aa52 79 */
emilmont 1:fdd22bb7aa52 80
emilmont 1:fdd22bb7aa52 81
emilmont 1:fdd22bb7aa52 82
emilmont 1:fdd22bb7aa52 83 void arm_correlate_opt_q7(
emilmont 1:fdd22bb7aa52 84 q7_t * pSrcA,
emilmont 1:fdd22bb7aa52 85 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 86 q7_t * pSrcB,
emilmont 1:fdd22bb7aa52 87 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 88 q7_t * pDst,
emilmont 1:fdd22bb7aa52 89 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 90 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 91 {
emilmont 1:fdd22bb7aa52 92 q7_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 93 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch */
emilmont 1:fdd22bb7aa52 94 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch */
emilmont 1:fdd22bb7aa52 95 q7_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 96 q7_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 97 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 98 q31_t acc0, acc1, acc2, acc3; /* Accumulators */
emilmont 1:fdd22bb7aa52 99 uint32_t j, k = 0u, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 100 int32_t inc = 1; /* output pointer increment */
emilmont 1:fdd22bb7aa52 101 uint32_t outBlockSize; /* loop counter */
emilmont 1:fdd22bb7aa52 102 q15_t x4; /* Temporary input variable */
emilmont 1:fdd22bb7aa52 103 uint32_t tapCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 104 q31_t x1, x2, x3, y1; /* Temporary input variables */
emilmont 1:fdd22bb7aa52 105
emilmont 1:fdd22bb7aa52 106 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 107 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 108 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 109 /* But CORR(x, y) is reverse of CORR(y, x) */
emilmont 1:fdd22bb7aa52 110 /* So, when srcBLen > srcALen, output pointer is made to point to the end of the output buffer */
emilmont 1:fdd22bb7aa52 111 /* and the destination pointer modifier, inc is set to -1 */
emilmont 1:fdd22bb7aa52 112 /* If srcALen > srcBLen, zero pad has to be done to srcB to make the two inputs of same length */
emilmont 1:fdd22bb7aa52 113 /* But to improve the performance,
emilmont 1:fdd22bb7aa52 114 * we include zeroes in the output instead of zero padding either of the the inputs*/
emilmont 1:fdd22bb7aa52 115 /* If srcALen > srcBLen,
emilmont 1:fdd22bb7aa52 116 * (srcALen - srcBLen) zeroes has to included in the starting of the output buffer */
emilmont 1:fdd22bb7aa52 117 /* If srcALen < srcBLen,
emilmont 1:fdd22bb7aa52 118 * (srcALen - srcBLen) zeroes has to included in the ending of the output buffer */
emilmont 1:fdd22bb7aa52 119 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 120 {
emilmont 1:fdd22bb7aa52 121 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 122 pIn1 = (pSrcA);
emilmont 1:fdd22bb7aa52 123
emilmont 1:fdd22bb7aa52 124 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 125 pIn2 = (pSrcB);
emilmont 1:fdd22bb7aa52 126
emilmont 1:fdd22bb7aa52 127 /* Number of output samples is calculated */
emilmont 1:fdd22bb7aa52 128 outBlockSize = (2u * srcALen) - 1u;
emilmont 1:fdd22bb7aa52 129
emilmont 1:fdd22bb7aa52 130 /* When srcALen > srcBLen, zero padding is done to srcB
emilmont 1:fdd22bb7aa52 131 * to make their lengths equal.
emilmont 1:fdd22bb7aa52 132 * Instead, (outBlockSize - (srcALen + srcBLen - 1))
emilmont 1:fdd22bb7aa52 133 * number of output samples are made zero */
emilmont 1:fdd22bb7aa52 134 j = outBlockSize - (srcALen + (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 135
emilmont 1:fdd22bb7aa52 136 /* Updating the pointer position to non zero value */
emilmont 1:fdd22bb7aa52 137 pOut += j;
emilmont 1:fdd22bb7aa52 138
emilmont 1:fdd22bb7aa52 139 }
emilmont 1:fdd22bb7aa52 140 else
emilmont 1:fdd22bb7aa52 141 {
emilmont 1:fdd22bb7aa52 142 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 143 pIn1 = (pSrcB);
emilmont 1:fdd22bb7aa52 144
emilmont 1:fdd22bb7aa52 145 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 146 pIn2 = (pSrcA);
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 149 j = srcBLen;
emilmont 1:fdd22bb7aa52 150 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 151 srcALen = j;
emilmont 1:fdd22bb7aa52 152
emilmont 1:fdd22bb7aa52 153 /* CORR(x, y) = Reverse order(CORR(y, x)) */
emilmont 1:fdd22bb7aa52 154 /* Hence set the destination pointer to point to the last output sample */
emilmont 1:fdd22bb7aa52 155 pOut = pDst + ((srcALen + srcBLen) - 2u);
emilmont 1:fdd22bb7aa52 156
emilmont 1:fdd22bb7aa52 157 /* Destination address modifier is set to -1 */
emilmont 1:fdd22bb7aa52 158 inc = -1;
emilmont 1:fdd22bb7aa52 159
emilmont 1:fdd22bb7aa52 160 }
emilmont 1:fdd22bb7aa52 161
emilmont 1:fdd22bb7aa52 162
emilmont 1:fdd22bb7aa52 163 /* Copy (srcBLen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 164 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 165
emilmont 1:fdd22bb7aa52 166 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 167 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 168 while(k > 0u)
emilmont 1:fdd22bb7aa52 169 {
emilmont 1:fdd22bb7aa52 170 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 171 x4 = (q15_t) * pIn2++;
emilmont 1:fdd22bb7aa52 172 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 173 x4 = (q15_t) * pIn2++;
emilmont 1:fdd22bb7aa52 174 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 175 x4 = (q15_t) * pIn2++;
emilmont 1:fdd22bb7aa52 176 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 177 x4 = (q15_t) * pIn2++;
emilmont 1:fdd22bb7aa52 178 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 181 k--;
emilmont 1:fdd22bb7aa52 182 }
emilmont 1:fdd22bb7aa52 183
emilmont 1:fdd22bb7aa52 184 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 185 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 186 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 187
emilmont 1:fdd22bb7aa52 188 while(k > 0u)
emilmont 1:fdd22bb7aa52 189 {
emilmont 1:fdd22bb7aa52 190 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 191 x4 = (q15_t) * pIn2++;
emilmont 1:fdd22bb7aa52 192 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 193
emilmont 1:fdd22bb7aa52 194 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 195 k--;
emilmont 1:fdd22bb7aa52 196 }
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 199 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 202 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 205 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 206
emilmont 1:fdd22bb7aa52 207 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 208 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 209 while(k > 0u)
emilmont 1:fdd22bb7aa52 210 {
emilmont 1:fdd22bb7aa52 211 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 212 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 213 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 214 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 215 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 216 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 217 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 218 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 219 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 220
emilmont 1:fdd22bb7aa52 221 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 222 k--;
emilmont 1:fdd22bb7aa52 223 }
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 226 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 227 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 228
emilmont 1:fdd22bb7aa52 229 while(k > 0u)
emilmont 1:fdd22bb7aa52 230 {
emilmont 1:fdd22bb7aa52 231 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 232 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 233 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 234
emilmont 1:fdd22bb7aa52 235 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 236 k--;
emilmont 1:fdd22bb7aa52 237 }
emilmont 1:fdd22bb7aa52 238
emilmont 1:fdd22bb7aa52 239 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 240
emilmont 1:fdd22bb7aa52 241 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 242 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 243
emilmont 1:fdd22bb7aa52 244 /* Update pointer */
emilmont 1:fdd22bb7aa52 245 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 246
emilmont 1:fdd22bb7aa52 247 #else
emilmont 1:fdd22bb7aa52 248
emilmont 1:fdd22bb7aa52 249 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 250 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 251
emilmont 1:fdd22bb7aa52 252 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 253 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 254 while(k > 0u)
emilmont 1:fdd22bb7aa52 255 {
emilmont 1:fdd22bb7aa52 256 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 257 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 258 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 259 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 260 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 261
emilmont 1:fdd22bb7aa52 262 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 263 k--;
emilmont 1:fdd22bb7aa52 264 }
emilmont 1:fdd22bb7aa52 265
emilmont 1:fdd22bb7aa52 266 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 267 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 268 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 269
emilmont 1:fdd22bb7aa52 270 while(k > 0u)
emilmont 1:fdd22bb7aa52 271 {
emilmont 1:fdd22bb7aa52 272 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 273 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 276 k--;
emilmont 1:fdd22bb7aa52 277 }
emilmont 1:fdd22bb7aa52 278
emilmont 2:da51fb522205 279 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 /* Temporary pointer for second sequence */
emilmont 1:fdd22bb7aa52 282 py = pScratch2;
emilmont 1:fdd22bb7aa52 283
emilmont 1:fdd22bb7aa52 284 /* Initialization of pScr2 pointer */
emilmont 1:fdd22bb7aa52 285 pScr2 = pScratch2;
emilmont 1:fdd22bb7aa52 286
emilmont 1:fdd22bb7aa52 287 /* Actual correlation process starts here */
emilmont 1:fdd22bb7aa52 288 blkCnt = (srcALen + srcBLen - 1u) >> 2;
emilmont 1:fdd22bb7aa52 289
emilmont 1:fdd22bb7aa52 290 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 291 {
emilmont 1:fdd22bb7aa52 292 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 293 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 294
emilmont 1:fdd22bb7aa52 295 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 296 acc0 = 0;
emilmont 1:fdd22bb7aa52 297 acc1 = 0;
emilmont 1:fdd22bb7aa52 298 acc2 = 0;
emilmont 1:fdd22bb7aa52 299 acc3 = 0;
emilmont 1:fdd22bb7aa52 300
emilmont 1:fdd22bb7aa52 301 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 302 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 305 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 306
emilmont 1:fdd22bb7aa52 307 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 308
emilmont 1:fdd22bb7aa52 309 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 310 {
emilmont 1:fdd22bb7aa52 311
emilmont 1:fdd22bb7aa52 312 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 313 y1 = _SIMD32_OFFSET(pScr2);
emilmont 1:fdd22bb7aa52 314
emilmont 1:fdd22bb7aa52 315 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 316 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 317 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 318
emilmont 1:fdd22bb7aa52 319 /* pack input data */
emilmont 1:fdd22bb7aa52 320 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 321 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 322 #else
emilmont 1:fdd22bb7aa52 323 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 324 #endif
emilmont 1:fdd22bb7aa52 325
emilmont 1:fdd22bb7aa52 326 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 327 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 328
emilmont 1:fdd22bb7aa52 329 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 330 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332 /* pack input data */
emilmont 1:fdd22bb7aa52 333 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 334 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 335 #else
emilmont 1:fdd22bb7aa52 336 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 337 #endif
emilmont 1:fdd22bb7aa52 338
emilmont 1:fdd22bb7aa52 339 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 342 y1 = _SIMD32_OFFSET(pScr2 + 2u);
emilmont 1:fdd22bb7aa52 343
emilmont 1:fdd22bb7aa52 344 acc0 = __SMLAD(x2, y1, acc0);
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346 acc2 = __SMLAD(x1, y1, acc2);
emilmont 1:fdd22bb7aa52 347
emilmont 1:fdd22bb7aa52 348 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 349
emilmont 1:fdd22bb7aa52 350 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 351
emilmont 1:fdd22bb7aa52 352 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 353 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 354 #else
emilmont 1:fdd22bb7aa52 355 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 356 #endif
emilmont 1:fdd22bb7aa52 357
emilmont 1:fdd22bb7aa52 358 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 pScr2 += 4u;
emilmont 1:fdd22bb7aa52 361
emilmont 1:fdd22bb7aa52 362
emilmont 1:fdd22bb7aa52 363 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 364 tapCnt--;
emilmont 1:fdd22bb7aa52 365 }
emilmont 1:fdd22bb7aa52 366
emilmont 1:fdd22bb7aa52 367
emilmont 1:fdd22bb7aa52 368
emilmont 1:fdd22bb7aa52 369 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 370 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 371
emilmont 1:fdd22bb7aa52 372
emilmont 1:fdd22bb7aa52 373 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 374 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 375
emilmont 1:fdd22bb7aa52 376 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 377 {
emilmont 1:fdd22bb7aa52 378
emilmont 1:fdd22bb7aa52 379 /* accumlate the results */
emilmont 1:fdd22bb7aa52 380 acc0 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 381 acc1 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 382 acc2 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 383 acc3 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 384
emilmont 1:fdd22bb7aa52 385 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 386
emilmont 1:fdd22bb7aa52 387 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 388 tapCnt--;
emilmont 1:fdd22bb7aa52 389 }
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391 blkCnt--;
emilmont 1:fdd22bb7aa52 392
emilmont 1:fdd22bb7aa52 393 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 394 *pOut = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 395 pOut += inc;
emilmont 1:fdd22bb7aa52 396 *pOut = (q7_t) (__SSAT(acc1 >> 7u, 8));
emilmont 1:fdd22bb7aa52 397 pOut += inc;
emilmont 1:fdd22bb7aa52 398 *pOut = (q7_t) (__SSAT(acc2 >> 7u, 8));
emilmont 1:fdd22bb7aa52 399 pOut += inc;
emilmont 1:fdd22bb7aa52 400 *pOut = (q7_t) (__SSAT(acc3 >> 7u, 8));
emilmont 1:fdd22bb7aa52 401 pOut += inc;
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 404 pScr2 = py;
emilmont 1:fdd22bb7aa52 405
emilmont 1:fdd22bb7aa52 406 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 407
emilmont 1:fdd22bb7aa52 408 }
emilmont 1:fdd22bb7aa52 409
emilmont 1:fdd22bb7aa52 410
emilmont 1:fdd22bb7aa52 411 blkCnt = (srcALen + srcBLen - 1u) & 0x3;
emilmont 1:fdd22bb7aa52 412
emilmont 1:fdd22bb7aa52 413 /* Calculate correlation for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 414 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 415 {
emilmont 1:fdd22bb7aa52 416 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 417 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 418
emilmont 1:fdd22bb7aa52 419 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 420 acc0 = 0;
emilmont 1:fdd22bb7aa52 421
emilmont 1:fdd22bb7aa52 422 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 423
emilmont 1:fdd22bb7aa52 424 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 425 {
emilmont 1:fdd22bb7aa52 426 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 427 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 428
emilmont 1:fdd22bb7aa52 429 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 430 tapCnt--;
emilmont 1:fdd22bb7aa52 431 }
emilmont 1:fdd22bb7aa52 432
emilmont 1:fdd22bb7aa52 433 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 434
emilmont 1:fdd22bb7aa52 435 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 436 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 437 {
emilmont 1:fdd22bb7aa52 438
emilmont 1:fdd22bb7aa52 439 /* accumlate the results */
emilmont 1:fdd22bb7aa52 440 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 441
emilmont 1:fdd22bb7aa52 442 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 443 tapCnt--;
emilmont 1:fdd22bb7aa52 444 }
emilmont 1:fdd22bb7aa52 445
emilmont 1:fdd22bb7aa52 446 blkCnt--;
emilmont 1:fdd22bb7aa52 447
emilmont 1:fdd22bb7aa52 448 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 449 *pOut = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 450
emilmont 1:fdd22bb7aa52 451 pOut += inc;
emilmont 1:fdd22bb7aa52 452
emilmont 1:fdd22bb7aa52 453 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 454 pScr2 = py;
emilmont 1:fdd22bb7aa52 455
emilmont 1:fdd22bb7aa52 456 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 457
emilmont 1:fdd22bb7aa52 458 }
emilmont 1:fdd22bb7aa52 459
emilmont 1:fdd22bb7aa52 460 }
emilmont 1:fdd22bb7aa52 461
emilmont 1:fdd22bb7aa52 462 /**
emilmont 1:fdd22bb7aa52 463 * @} end of Corr group
emilmont 1:fdd22bb7aa52 464 */