Eli Hughes / CMSIS_DSP_401

V4.0.1 of the ARM CMSIS DSP libraries. Note that arm_bitreversal2.s, arm_cfft_f32.c and arm_rfft_fast_f32.c had to be removed. arm_bitreversal2.s will not assemble with the online tools. So, the fast f32 FFT functions are not yet available. All the other FFT functions are available.

Dependents:   MPU9150_Example fir_f32 fir_f32 MPU9150_nucleo_noni2cdev ... more

FilteringFunctions/arm_correlate_q31.c@0:3d9c67d97d6f, 2014-07-28 (annotated)

Committer:
emh203
Date:
Mon Jul 28 15:03:15 2014 +0000
Revision:
0:3d9c67d97d6f
1st working commit.   Had to remove arm_bitreversal2.s     arm_cfft_f32.c and arm_rfft_fast_f32.c.    The .s will not assemble.      For now I removed these functions so we could at least have a library for the other functions.

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emh203 0:3d9c67d97d6f 1 /* ----------------------------------------------------------------------
emh203 0:3d9c67d97d6f 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emh203 0:3d9c67d97d6f 3 *
emh203 0:3d9c67d97d6f 4 * $Date: 12. March 2014
emh203 0:3d9c67d97d6f 5 * $Revision: V1.4.3
emh203 0:3d9c67d97d6f 6 *
emh203 0:3d9c67d97d6f 7 * Project: CMSIS DSP Library
emh203 0:3d9c67d97d6f 8 * Title: arm_correlate_q31.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Correlation of Q31 sequences.
emh203 0:3d9c67d97d6f 11 *
emh203 0:3d9c67d97d6f 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emh203 0:3d9c67d97d6f 13 *
emh203 0:3d9c67d97d6f 14 * Redistribution and use in source and binary forms, with or without
emh203 0:3d9c67d97d6f 15 * modification, are permitted provided that the following conditions
emh203 0:3d9c67d97d6f 16 * are met:
emh203 0:3d9c67d97d6f 17 * - Redistributions of source code must retain the above copyright
emh203 0:3d9c67d97d6f 18 * notice, this list of conditions and the following disclaimer.
emh203 0:3d9c67d97d6f 19 * - Redistributions in binary form must reproduce the above copyright
emh203 0:3d9c67d97d6f 20 * notice, this list of conditions and the following disclaimer in
emh203 0:3d9c67d97d6f 21 * the documentation and/or other materials provided with the
emh203 0:3d9c67d97d6f 22 * distribution.
emh203 0:3d9c67d97d6f 23 * - Neither the name of ARM LIMITED nor the names of its contributors
emh203 0:3d9c67d97d6f 24 * may be used to endorse or promote products derived from this
emh203 0:3d9c67d97d6f 25 * software without specific prior written permission.
emh203 0:3d9c67d97d6f 26 *
emh203 0:3d9c67d97d6f 27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
emh203 0:3d9c67d97d6f 28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
emh203 0:3d9c67d97d6f 29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
emh203 0:3d9c67d97d6f 30 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
emh203 0:3d9c67d97d6f 31 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
emh203 0:3d9c67d97d6f 32 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
emh203 0:3d9c67d97d6f 33 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
emh203 0:3d9c67d97d6f 34 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
emh203 0:3d9c67d97d6f 35 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
emh203 0:3d9c67d97d6f 36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
emh203 0:3d9c67d97d6f 37 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
emh203 0:3d9c67d97d6f 38 * POSSIBILITY OF SUCH DAMAGE.
emh203 0:3d9c67d97d6f 39 * -------------------------------------------------------------------- */
emh203 0:3d9c67d97d6f 40
emh203 0:3d9c67d97d6f 41 #include "arm_math.h"
emh203 0:3d9c67d97d6f 42
emh203 0:3d9c67d97d6f 43 /**
emh203 0:3d9c67d97d6f 44 * @ingroup groupFilters
emh203 0:3d9c67d97d6f 45 */
emh203 0:3d9c67d97d6f 46
emh203 0:3d9c67d97d6f 47 /**
emh203 0:3d9c67d97d6f 48 * @addtogroup Corr
emh203 0:3d9c67d97d6f 49 * @{
emh203 0:3d9c67d97d6f 50 */
emh203 0:3d9c67d97d6f 51
emh203 0:3d9c67d97d6f 52 /**
emh203 0:3d9c67d97d6f 53 * @brief Correlation of Q31 sequences.
emh203 0:3d9c67d97d6f 54 * @param[in] *pSrcA points to the first input sequence.
emh203 0:3d9c67d97d6f 55 * @param[in] srcALen length of the first input sequence.
emh203 0:3d9c67d97d6f 56 * @param[in] *pSrcB points to the second input sequence.
emh203 0:3d9c67d97d6f 57 * @param[in] srcBLen length of the second input sequence.
emh203 0:3d9c67d97d6f 58 * @param[out] *pDst points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1.
emh203 0:3d9c67d97d6f 59 * @return none.
emh203 0:3d9c67d97d6f 60 *
emh203 0:3d9c67d97d6f 61 * @details
emh203 0:3d9c67d97d6f 62 * <b>Scaling and Overflow Behavior:</b>
emh203 0:3d9c67d97d6f 63 *
emh203 0:3d9c67d97d6f 64 * \par
emh203 0:3d9c67d97d6f 65 * The function is implemented using an internal 64-bit accumulator.
emh203 0:3d9c67d97d6f 66 * The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.
emh203 0:3d9c67d97d6f 67 * There is no saturation on intermediate additions.
emh203 0:3d9c67d97d6f 68 * Thus, if the accumulator overflows it wraps around and distorts the result.
emh203 0:3d9c67d97d6f 69 * The input signals should be scaled down to avoid intermediate overflows.
emh203 0:3d9c67d97d6f 70 * Scale down one of the inputs by 1/min(srcALen, srcBLen)to avoid overflows since a
emh203 0:3d9c67d97d6f 71 * maximum of min(srcALen, srcBLen) number of additions is carried internally.
emh203 0:3d9c67d97d6f 72 * The 2.62 accumulator is right shifted by 31 bits and saturated to 1.31 format to yield the final result.
emh203 0:3d9c67d97d6f 73 *
emh203 0:3d9c67d97d6f 74 * \par
emh203 0:3d9c67d97d6f 75 * See <code>arm_correlate_fast_q31()</code> for a faster but less precise implementation of this function for Cortex-M3 and Cortex-M4.
emh203 0:3d9c67d97d6f 76 */
emh203 0:3d9c67d97d6f 77
emh203 0:3d9c67d97d6f 78 void arm_correlate_q31(
emh203 0:3d9c67d97d6f 79 q31_t * pSrcA,
emh203 0:3d9c67d97d6f 80 uint32_t srcALen,
emh203 0:3d9c67d97d6f 81 q31_t * pSrcB,
emh203 0:3d9c67d97d6f 82 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 83 q31_t * pDst)
emh203 0:3d9c67d97d6f 84 {
emh203 0:3d9c67d97d6f 85
emh203 0:3d9c67d97d6f 86 #ifndef ARM_MATH_CM0_FAMILY
emh203 0:3d9c67d97d6f 87
emh203 0:3d9c67d97d6f 88 /* Run the below code for Cortex-M4 and Cortex-M3 */
emh203 0:3d9c67d97d6f 89
emh203 0:3d9c67d97d6f 90 q31_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 91 q31_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 92 q31_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 93 q31_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 94 q31_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 95 q31_t *pSrc1; /* Intermediate pointers */
emh203 0:3d9c67d97d6f 96 q63_t sum, acc0, acc1, acc2; /* Accumulators */
emh203 0:3d9c67d97d6f 97 q31_t x0, x1, x2, c0; /* temporary variables for holding input and coefficient values */
emh203 0:3d9c67d97d6f 98 uint32_t j, k = 0u, count, blkCnt, outBlockSize, blockSize1, blockSize2, blockSize3; /* loop counter */
emh203 0:3d9c67d97d6f 99 int32_t inc = 1; /* Destination address modifier */
emh203 0:3d9c67d97d6f 100
emh203 0:3d9c67d97d6f 101
emh203 0:3d9c67d97d6f 102 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 103 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 104 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 105 /* But CORR(x, y) is reverse of CORR(y, x) */
emh203 0:3d9c67d97d6f 106 /* So, when srcBLen > srcALen, output pointer is made to point to the end of the output buffer */
emh203 0:3d9c67d97d6f 107 /* and the destination pointer modifier, inc is set to -1 */
emh203 0:3d9c67d97d6f 108 /* If srcALen > srcBLen, zero pad has to be done to srcB to make the two inputs of same length */
emh203 0:3d9c67d97d6f 109 /* But to improve the performance,
emh203 0:3d9c67d97d6f 110 * we include zeroes in the output instead of zero padding either of the the inputs*/
emh203 0:3d9c67d97d6f 111 /* If srcALen > srcBLen,
emh203 0:3d9c67d97d6f 112 * (srcALen - srcBLen) zeroes has to included in the starting of the output buffer */
emh203 0:3d9c67d97d6f 113 /* If srcALen < srcBLen,
emh203 0:3d9c67d97d6f 114 * (srcALen - srcBLen) zeroes has to included in the ending of the output buffer */
emh203 0:3d9c67d97d6f 115 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 116 {
emh203 0:3d9c67d97d6f 117 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 118 pIn1 = (pSrcA);
emh203 0:3d9c67d97d6f 119
emh203 0:3d9c67d97d6f 120 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 121 pIn2 = (pSrcB);
emh203 0:3d9c67d97d6f 122
emh203 0:3d9c67d97d6f 123 /* Number of output samples is calculated */
emh203 0:3d9c67d97d6f 124 outBlockSize = (2u * srcALen) - 1u;
emh203 0:3d9c67d97d6f 125
emh203 0:3d9c67d97d6f 126 /* When srcALen > srcBLen, zero padding is done to srcB
emh203 0:3d9c67d97d6f 127 * to make their lengths equal.
emh203 0:3d9c67d97d6f 128 * Instead, (outBlockSize - (srcALen + srcBLen - 1))
emh203 0:3d9c67d97d6f 129 * number of output samples are made zero */
emh203 0:3d9c67d97d6f 130 j = outBlockSize - (srcALen + (srcBLen - 1u));
emh203 0:3d9c67d97d6f 131
emh203 0:3d9c67d97d6f 132 /* Updating the pointer position to non zero value */
emh203 0:3d9c67d97d6f 133 pOut += j;
emh203 0:3d9c67d97d6f 134
emh203 0:3d9c67d97d6f 135 }
emh203 0:3d9c67d97d6f 136 else
emh203 0:3d9c67d97d6f 137 {
emh203 0:3d9c67d97d6f 138 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 139 pIn1 = (pSrcB);
emh203 0:3d9c67d97d6f 140
emh203 0:3d9c67d97d6f 141 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 142 pIn2 = (pSrcA);
emh203 0:3d9c67d97d6f 143
emh203 0:3d9c67d97d6f 144 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 145 j = srcBLen;
emh203 0:3d9c67d97d6f 146 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 147 srcALen = j;
emh203 0:3d9c67d97d6f 148
emh203 0:3d9c67d97d6f 149 /* CORR(x, y) = Reverse order(CORR(y, x)) */
emh203 0:3d9c67d97d6f 150 /* Hence set the destination pointer to point to the last output sample */
emh203 0:3d9c67d97d6f 151 pOut = pDst + ((srcALen + srcBLen) - 2u);
emh203 0:3d9c67d97d6f 152
emh203 0:3d9c67d97d6f 153 /* Destination address modifier is set to -1 */
emh203 0:3d9c67d97d6f 154 inc = -1;
emh203 0:3d9c67d97d6f 155
emh203 0:3d9c67d97d6f 156 }
emh203 0:3d9c67d97d6f 157
emh203 0:3d9c67d97d6f 158 /* The function is internally
emh203 0:3d9c67d97d6f 159 * divided into three parts according to the number of multiplications that has to be
emh203 0:3d9c67d97d6f 160 * taken place between inputA samples and inputB samples. In the first part of the
emh203 0:3d9c67d97d6f 161 * algorithm, the multiplications increase by one for every iteration.
emh203 0:3d9c67d97d6f 162 * In the second part of the algorithm, srcBLen number of multiplications are done.
emh203 0:3d9c67d97d6f 163 * In the third part of the algorithm, the multiplications decrease by one
emh203 0:3d9c67d97d6f 164 * for every iteration.*/
emh203 0:3d9c67d97d6f 165 /* The algorithm is implemented in three stages.
emh203 0:3d9c67d97d6f 166 * The loop counters of each stage is initiated here. */
emh203 0:3d9c67d97d6f 167 blockSize1 = srcBLen - 1u;
emh203 0:3d9c67d97d6f 168 blockSize2 = srcALen - (srcBLen - 1u);
emh203 0:3d9c67d97d6f 169 blockSize3 = blockSize1;
emh203 0:3d9c67d97d6f 170
emh203 0:3d9c67d97d6f 171 /* --------------------------
emh203 0:3d9c67d97d6f 172 * Initializations of stage1
emh203 0:3d9c67d97d6f 173 * -------------------------*/
emh203 0:3d9c67d97d6f 174
emh203 0:3d9c67d97d6f 175 /* sum = x[0] * y[srcBlen - 1]
emh203 0:3d9c67d97d6f 176 * sum = x[0] * y[srcBlen - 2] + x[1] * y[srcBlen - 1]
emh203 0:3d9c67d97d6f 177 * ....
emh203 0:3d9c67d97d6f 178 * sum = x[0] * y[0] + x[1] * y[1] +...+ x[srcBLen - 1] * y[srcBLen - 1]
emh203 0:3d9c67d97d6f 179 */
emh203 0:3d9c67d97d6f 180
emh203 0:3d9c67d97d6f 181 /* In this stage the MAC operations are increased by 1 for every iteration.
emh203 0:3d9c67d97d6f 182 The count variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 183 count = 1u;
emh203 0:3d9c67d97d6f 184
emh203 0:3d9c67d97d6f 185 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 186 px = pIn1;
emh203 0:3d9c67d97d6f 187
emh203 0:3d9c67d97d6f 188 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 189 pSrc1 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 190 py = pSrc1;
emh203 0:3d9c67d97d6f 191
emh203 0:3d9c67d97d6f 192 /* ------------------------
emh203 0:3d9c67d97d6f 193 * Stage1 process
emh203 0:3d9c67d97d6f 194 * ----------------------*/
emh203 0:3d9c67d97d6f 195
emh203 0:3d9c67d97d6f 196 /* The first stage starts here */
emh203 0:3d9c67d97d6f 197 while(blockSize1 > 0u)
emh203 0:3d9c67d97d6f 198 {
emh203 0:3d9c67d97d6f 199 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 200 sum = 0;
emh203 0:3d9c67d97d6f 201
emh203 0:3d9c67d97d6f 202 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 203 k = count >> 2;
emh203 0:3d9c67d97d6f 204
emh203 0:3d9c67d97d6f 205 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 206 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 207 while(k > 0u)
emh203 0:3d9c67d97d6f 208 {
emh203 0:3d9c67d97d6f 209 /* x[0] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 210 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 211 /* x[1] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 212 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 213 /* x[2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 214 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 215 /* x[3] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 216 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 217
emh203 0:3d9c67d97d6f 218 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 219 k--;
emh203 0:3d9c67d97d6f 220 }
emh203 0:3d9c67d97d6f 221
emh203 0:3d9c67d97d6f 222 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 223 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 224 k = count % 0x4u;
emh203 0:3d9c67d97d6f 225
emh203 0:3d9c67d97d6f 226 while(k > 0u)
emh203 0:3d9c67d97d6f 227 {
emh203 0:3d9c67d97d6f 228 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 229 /* x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 230 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 231
emh203 0:3d9c67d97d6f 232 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 233 k--;
emh203 0:3d9c67d97d6f 234 }
emh203 0:3d9c67d97d6f 235
emh203 0:3d9c67d97d6f 236 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 237 *pOut = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 238 /* Destination pointer is updated according to the address modifier, inc */
emh203 0:3d9c67d97d6f 239 pOut += inc;
emh203 0:3d9c67d97d6f 240
emh203 0:3d9c67d97d6f 241 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 242 py = pSrc1 - count;
emh203 0:3d9c67d97d6f 243 px = pIn1;
emh203 0:3d9c67d97d6f 244
emh203 0:3d9c67d97d6f 245 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 246 count++;
emh203 0:3d9c67d97d6f 247
emh203 0:3d9c67d97d6f 248 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 249 blockSize1--;
emh203 0:3d9c67d97d6f 250 }
emh203 0:3d9c67d97d6f 251
emh203 0:3d9c67d97d6f 252 /* --------------------------
emh203 0:3d9c67d97d6f 253 * Initializations of stage2
emh203 0:3d9c67d97d6f 254 * ------------------------*/
emh203 0:3d9c67d97d6f 255
emh203 0:3d9c67d97d6f 256 /* sum = x[0] * y[0] + x[1] * y[1] +...+ x[srcBLen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 257 * sum = x[1] * y[0] + x[2] * y[1] +...+ x[srcBLen] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 258 * ....
emh203 0:3d9c67d97d6f 259 * sum = x[srcALen-srcBLen-2] * y[0] + x[srcALen-srcBLen-1] * y[1] +...+ x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 260 */
emh203 0:3d9c67d97d6f 261
emh203 0:3d9c67d97d6f 262 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 263 px = pIn1;
emh203 0:3d9c67d97d6f 264
emh203 0:3d9c67d97d6f 265 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 266 py = pIn2;
emh203 0:3d9c67d97d6f 267
emh203 0:3d9c67d97d6f 268 /* count is index by which the pointer pIn1 to be incremented */
emh203 0:3d9c67d97d6f 269 count = 0u;
emh203 0:3d9c67d97d6f 270
emh203 0:3d9c67d97d6f 271 /* -------------------
emh203 0:3d9c67d97d6f 272 * Stage2 process
emh203 0:3d9c67d97d6f 273 * ------------------*/
emh203 0:3d9c67d97d6f 274
emh203 0:3d9c67d97d6f 275 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emh203 0:3d9c67d97d6f 276 * So, to loop unroll over blockSize2,
emh203 0:3d9c67d97d6f 277 * srcBLen should be greater than or equal to 4 */
emh203 0:3d9c67d97d6f 278 if(srcBLen >= 4u)
emh203 0:3d9c67d97d6f 279 {
emh203 0:3d9c67d97d6f 280 /* Loop unroll by 3 */
emh203 0:3d9c67d97d6f 281 blkCnt = blockSize2 / 3;
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 284 {
emh203 0:3d9c67d97d6f 285 /* Set all accumulators to zero */
emh203 0:3d9c67d97d6f 286 acc0 = 0;
emh203 0:3d9c67d97d6f 287 acc1 = 0;
emh203 0:3d9c67d97d6f 288 acc2 = 0;
emh203 0:3d9c67d97d6f 289
emh203 0:3d9c67d97d6f 290 /* read x[0], x[1] samples */
emh203 0:3d9c67d97d6f 291 x0 = *(px++);
emh203 0:3d9c67d97d6f 292 x1 = *(px++);
emh203 0:3d9c67d97d6f 293
emh203 0:3d9c67d97d6f 294 /* Apply loop unrolling and compute 3 MACs simultaneously. */
emh203 0:3d9c67d97d6f 295 k = srcBLen / 3;
emh203 0:3d9c67d97d6f 296
emh203 0:3d9c67d97d6f 297 /* First part of the processing with loop unrolling. Compute 3 MACs at a time.
emh203 0:3d9c67d97d6f 298 ** a second loop below computes MACs for the remaining 1 to 2 samples. */
emh203 0:3d9c67d97d6f 299 do
emh203 0:3d9c67d97d6f 300 {
emh203 0:3d9c67d97d6f 301 /* Read y[0] sample */
emh203 0:3d9c67d97d6f 302 c0 = *(py);
emh203 0:3d9c67d97d6f 303
emh203 0:3d9c67d97d6f 304 /* Read x[2] sample */
emh203 0:3d9c67d97d6f 305 x2 = *(px);
emh203 0:3d9c67d97d6f 306
emh203 0:3d9c67d97d6f 307 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 308 /* acc0 += x[0] * y[0] */
emh203 0:3d9c67d97d6f 309 acc0 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 310 /* acc1 += x[1] * y[0] */
emh203 0:3d9c67d97d6f 311 acc1 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 312 /* acc2 += x[2] * y[0] */
emh203 0:3d9c67d97d6f 313 acc2 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 314
emh203 0:3d9c67d97d6f 315 /* Read y[1] sample */
emh203 0:3d9c67d97d6f 316 c0 = *(py + 1u);
emh203 0:3d9c67d97d6f 317
emh203 0:3d9c67d97d6f 318 /* Read x[3] sample */
emh203 0:3d9c67d97d6f 319 x0 = *(px + 1u);
emh203 0:3d9c67d97d6f 320
emh203 0:3d9c67d97d6f 321 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 322 /* acc0 += x[1] * y[1] */
emh203 0:3d9c67d97d6f 323 acc0 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 324 /* acc1 += x[2] * y[1] */
emh203 0:3d9c67d97d6f 325 acc1 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 326 /* acc2 += x[3] * y[1] */
emh203 0:3d9c67d97d6f 327 acc2 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 328
emh203 0:3d9c67d97d6f 329 /* Read y[2] sample */
emh203 0:3d9c67d97d6f 330 c0 = *(py + 2u);
emh203 0:3d9c67d97d6f 331
emh203 0:3d9c67d97d6f 332 /* Read x[4] sample */
emh203 0:3d9c67d97d6f 333 x1 = *(px + 2u);
emh203 0:3d9c67d97d6f 334
emh203 0:3d9c67d97d6f 335 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 336 /* acc0 += x[2] * y[2] */
emh203 0:3d9c67d97d6f 337 acc0 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 338 /* acc1 += x[3] * y[2] */
emh203 0:3d9c67d97d6f 339 acc1 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 340 /* acc2 += x[4] * y[2] */
emh203 0:3d9c67d97d6f 341 acc2 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 342
emh203 0:3d9c67d97d6f 343 /* update scratch pointers */
emh203 0:3d9c67d97d6f 344 px += 3u;
emh203 0:3d9c67d97d6f 345 py += 3u;
emh203 0:3d9c67d97d6f 346
emh203 0:3d9c67d97d6f 347 } while(--k);
emh203 0:3d9c67d97d6f 348
emh203 0:3d9c67d97d6f 349 /* If the srcBLen is not a multiple of 3, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 350 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 351 k = srcBLen - (3 * (srcBLen / 3));
emh203 0:3d9c67d97d6f 352
emh203 0:3d9c67d97d6f 353 while(k > 0u)
emh203 0:3d9c67d97d6f 354 {
emh203 0:3d9c67d97d6f 355 /* Read y[4] sample */
emh203 0:3d9c67d97d6f 356 c0 = *(py++);
emh203 0:3d9c67d97d6f 357
emh203 0:3d9c67d97d6f 358 /* Read x[7] sample */
emh203 0:3d9c67d97d6f 359 x2 = *(px++);
emh203 0:3d9c67d97d6f 360
emh203 0:3d9c67d97d6f 361 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 362 /* acc0 += x[4] * y[4] */
emh203 0:3d9c67d97d6f 363 acc0 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 364 /* acc1 += x[5] * y[4] */
emh203 0:3d9c67d97d6f 365 acc1 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 366 /* acc2 += x[6] * y[4] */
emh203 0:3d9c67d97d6f 367 acc2 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 368
emh203 0:3d9c67d97d6f 369 /* Reuse the present samples for the next MAC */
emh203 0:3d9c67d97d6f 370 x0 = x1;
emh203 0:3d9c67d97d6f 371 x1 = x2;
emh203 0:3d9c67d97d6f 372
emh203 0:3d9c67d97d6f 373 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 374 k--;
emh203 0:3d9c67d97d6f 375 }
emh203 0:3d9c67d97d6f 376
emh203 0:3d9c67d97d6f 377 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 378 *pOut = (q31_t) (acc0 >> 31);
emh203 0:3d9c67d97d6f 379 /* Destination pointer is updated according to the address modifier, inc */
emh203 0:3d9c67d97d6f 380 pOut += inc;
emh203 0:3d9c67d97d6f 381
emh203 0:3d9c67d97d6f 382 *pOut = (q31_t) (acc1 >> 31);
emh203 0:3d9c67d97d6f 383 pOut += inc;
emh203 0:3d9c67d97d6f 384
emh203 0:3d9c67d97d6f 385 *pOut = (q31_t) (acc2 >> 31);
emh203 0:3d9c67d97d6f 386 pOut += inc;
emh203 0:3d9c67d97d6f 387
emh203 0:3d9c67d97d6f 388 /* Increment the pointer pIn1 index, count by 3 */
emh203 0:3d9c67d97d6f 389 count += 3u;
emh203 0:3d9c67d97d6f 390
emh203 0:3d9c67d97d6f 391 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 392 px = pIn1 + count;
emh203 0:3d9c67d97d6f 393 py = pIn2;
emh203 0:3d9c67d97d6f 394
emh203 0:3d9c67d97d6f 395
emh203 0:3d9c67d97d6f 396 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 397 blkCnt--;
emh203 0:3d9c67d97d6f 398 }
emh203 0:3d9c67d97d6f 399
emh203 0:3d9c67d97d6f 400 /* If the blockSize2 is not a multiple of 3, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 401 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 402 blkCnt = blockSize2 - 3 * (blockSize2 / 3);
emh203 0:3d9c67d97d6f 403
emh203 0:3d9c67d97d6f 404 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 405 {
emh203 0:3d9c67d97d6f 406 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 407 sum = 0;
emh203 0:3d9c67d97d6f 408
emh203 0:3d9c67d97d6f 409 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 410 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 411
emh203 0:3d9c67d97d6f 412 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 413 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 414 while(k > 0u)
emh203 0:3d9c67d97d6f 415 {
emh203 0:3d9c67d97d6f 416 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 417 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 418 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 419 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 420 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 421
emh203 0:3d9c67d97d6f 422 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 423 k--;
emh203 0:3d9c67d97d6f 424 }
emh203 0:3d9c67d97d6f 425
emh203 0:3d9c67d97d6f 426 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 427 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 428 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 429
emh203 0:3d9c67d97d6f 430 while(k > 0u)
emh203 0:3d9c67d97d6f 431 {
emh203 0:3d9c67d97d6f 432 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 433 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 434
emh203 0:3d9c67d97d6f 435 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 436 k--;
emh203 0:3d9c67d97d6f 437 }
emh203 0:3d9c67d97d6f 438
emh203 0:3d9c67d97d6f 439 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 440 *pOut = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 441 /* Destination pointer is updated according to the address modifier, inc */
emh203 0:3d9c67d97d6f 442 pOut += inc;
emh203 0:3d9c67d97d6f 443
emh203 0:3d9c67d97d6f 444 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 445 count++;
emh203 0:3d9c67d97d6f 446
emh203 0:3d9c67d97d6f 447 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 448 px = pIn1 + count;
emh203 0:3d9c67d97d6f 449 py = pIn2;
emh203 0:3d9c67d97d6f 450
emh203 0:3d9c67d97d6f 451 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 452 blkCnt--;
emh203 0:3d9c67d97d6f 453 }
emh203 0:3d9c67d97d6f 454 }
emh203 0:3d9c67d97d6f 455 else
emh203 0:3d9c67d97d6f 456 {
emh203 0:3d9c67d97d6f 457 /* If the srcBLen is not a multiple of 4,
emh203 0:3d9c67d97d6f 458 * the blockSize2 loop cannot be unrolled by 4 */
emh203 0:3d9c67d97d6f 459 blkCnt = blockSize2;
emh203 0:3d9c67d97d6f 460
emh203 0:3d9c67d97d6f 461 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 462 {
emh203 0:3d9c67d97d6f 463 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 464 sum = 0;
emh203 0:3d9c67d97d6f 465
emh203 0:3d9c67d97d6f 466 /* Loop over srcBLen */
emh203 0:3d9c67d97d6f 467 k = srcBLen;
emh203 0:3d9c67d97d6f 468
emh203 0:3d9c67d97d6f 469 while(k > 0u)
emh203 0:3d9c67d97d6f 470 {
emh203 0:3d9c67d97d6f 471 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 472 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 473
emh203 0:3d9c67d97d6f 474 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 475 k--;
emh203 0:3d9c67d97d6f 476 }
emh203 0:3d9c67d97d6f 477
emh203 0:3d9c67d97d6f 478 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 479 *pOut = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 480 /* Destination pointer is updated according to the address modifier, inc */
emh203 0:3d9c67d97d6f 481 pOut += inc;
emh203 0:3d9c67d97d6f 482
emh203 0:3d9c67d97d6f 483 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 484 count++;
emh203 0:3d9c67d97d6f 485
emh203 0:3d9c67d97d6f 486 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 487 px = pIn1 + count;
emh203 0:3d9c67d97d6f 488 py = pIn2;
emh203 0:3d9c67d97d6f 489
emh203 0:3d9c67d97d6f 490 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 491 blkCnt--;
emh203 0:3d9c67d97d6f 492 }
emh203 0:3d9c67d97d6f 493 }
emh203 0:3d9c67d97d6f 494
emh203 0:3d9c67d97d6f 495 /* --------------------------
emh203 0:3d9c67d97d6f 496 * Initializations of stage3
emh203 0:3d9c67d97d6f 497 * -------------------------*/
emh203 0:3d9c67d97d6f 498
emh203 0:3d9c67d97d6f 499 /* sum += x[srcALen-srcBLen+1] * y[0] + x[srcALen-srcBLen+2] * y[1] +...+ x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 500 * sum += x[srcALen-srcBLen+2] * y[0] + x[srcALen-srcBLen+3] * y[1] +...+ x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 501 * ....
emh203 0:3d9c67d97d6f 502 * sum += x[srcALen-2] * y[0] + x[srcALen-1] * y[1]
emh203 0:3d9c67d97d6f 503 * sum += x[srcALen-1] * y[0]
emh203 0:3d9c67d97d6f 504 */
emh203 0:3d9c67d97d6f 505
emh203 0:3d9c67d97d6f 506 /* In this stage the MAC operations are decreased by 1 for every iteration.
emh203 0:3d9c67d97d6f 507 The count variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 508 count = srcBLen - 1u;
emh203 0:3d9c67d97d6f 509
emh203 0:3d9c67d97d6f 510 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 511 pSrc1 = pIn1 + (srcALen - (srcBLen - 1u));
emh203 0:3d9c67d97d6f 512 px = pSrc1;
emh203 0:3d9c67d97d6f 513
emh203 0:3d9c67d97d6f 514 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 515 py = pIn2;
emh203 0:3d9c67d97d6f 516
emh203 0:3d9c67d97d6f 517 /* -------------------
emh203 0:3d9c67d97d6f 518 * Stage3 process
emh203 0:3d9c67d97d6f 519 * ------------------*/
emh203 0:3d9c67d97d6f 520
emh203 0:3d9c67d97d6f 521 while(blockSize3 > 0u)
emh203 0:3d9c67d97d6f 522 {
emh203 0:3d9c67d97d6f 523 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 524 sum = 0;
emh203 0:3d9c67d97d6f 525
emh203 0:3d9c67d97d6f 526 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 527 k = count >> 2u;
emh203 0:3d9c67d97d6f 528
emh203 0:3d9c67d97d6f 529 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 530 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 531 while(k > 0u)
emh203 0:3d9c67d97d6f 532 {
emh203 0:3d9c67d97d6f 533 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 534 /* sum += x[srcALen - srcBLen + 4] * y[3] */
emh203 0:3d9c67d97d6f 535 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 536 /* sum += x[srcALen - srcBLen + 3] * y[2] */
emh203 0:3d9c67d97d6f 537 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 538 /* sum += x[srcALen - srcBLen + 2] * y[1] */
emh203 0:3d9c67d97d6f 539 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 540 /* sum += x[srcALen - srcBLen + 1] * y[0] */
emh203 0:3d9c67d97d6f 541 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 542
emh203 0:3d9c67d97d6f 543 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 544 k--;
emh203 0:3d9c67d97d6f 545 }
emh203 0:3d9c67d97d6f 546
emh203 0:3d9c67d97d6f 547 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 548 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 549 k = count % 0x4u;
emh203 0:3d9c67d97d6f 550
emh203 0:3d9c67d97d6f 551 while(k > 0u)
emh203 0:3d9c67d97d6f 552 {
emh203 0:3d9c67d97d6f 553 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 554 sum += (q63_t) * px++ * (*py++);
emh203 0:3d9c67d97d6f 555
emh203 0:3d9c67d97d6f 556 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 557 k--;
emh203 0:3d9c67d97d6f 558 }
emh203 0:3d9c67d97d6f 559
emh203 0:3d9c67d97d6f 560 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 561 *pOut = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 562 /* Destination pointer is updated according to the address modifier, inc */
emh203 0:3d9c67d97d6f 563 pOut += inc;
emh203 0:3d9c67d97d6f 564
emh203 0:3d9c67d97d6f 565 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 566 px = ++pSrc1;
emh203 0:3d9c67d97d6f 567 py = pIn2;
emh203 0:3d9c67d97d6f 568
emh203 0:3d9c67d97d6f 569 /* Decrement the MAC count */
emh203 0:3d9c67d97d6f 570 count--;
emh203 0:3d9c67d97d6f 571
emh203 0:3d9c67d97d6f 572 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 573 blockSize3--;
emh203 0:3d9c67d97d6f 574 }
emh203 0:3d9c67d97d6f 575
emh203 0:3d9c67d97d6f 576 #else
emh203 0:3d9c67d97d6f 577
emh203 0:3d9c67d97d6f 578 /* Run the below code for Cortex-M0 */
emh203 0:3d9c67d97d6f 579
emh203 0:3d9c67d97d6f 580 q31_t *pIn1 = pSrcA; /* inputA pointer */
emh203 0:3d9c67d97d6f 581 q31_t *pIn2 = pSrcB + (srcBLen - 1u); /* inputB pointer */
emh203 0:3d9c67d97d6f 582 q63_t sum; /* Accumulators */
emh203 0:3d9c67d97d6f 583 uint32_t i = 0u, j; /* loop counters */
emh203 0:3d9c67d97d6f 584 uint32_t inv = 0u; /* Reverse order flag */
emh203 0:3d9c67d97d6f 585 uint32_t tot = 0u; /* Length */
emh203 0:3d9c67d97d6f 586
emh203 0:3d9c67d97d6f 587 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 588 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 589 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 590 /* But CORR(x, y) is reverse of CORR(y, x) */
emh203 0:3d9c67d97d6f 591 /* So, when srcBLen > srcALen, output pointer is made to point to the end of the output buffer */
emh203 0:3d9c67d97d6f 592 /* and a varaible, inv is set to 1 */
emh203 0:3d9c67d97d6f 593 /* If lengths are not equal then zero pad has to be done to make the two
emh203 0:3d9c67d97d6f 594 * inputs of same length. But to improve the performance, we include zeroes
emh203 0:3d9c67d97d6f 595 * in the output instead of zero padding either of the the inputs*/
emh203 0:3d9c67d97d6f 596 /* If srcALen > srcBLen, (srcALen - srcBLen) zeroes has to included in the
emh203 0:3d9c67d97d6f 597 * starting of the output buffer */
emh203 0:3d9c67d97d6f 598 /* If srcALen < srcBLen, (srcALen - srcBLen) zeroes has to included in the
emh203 0:3d9c67d97d6f 599 * ending of the output buffer */
emh203 0:3d9c67d97d6f 600 /* Once the zero padding is done the remaining of the output is calcualted
emh203 0:3d9c67d97d6f 601 * using correlation but with the shorter signal time shifted. */
emh203 0:3d9c67d97d6f 602
emh203 0:3d9c67d97d6f 603 /* Calculate the length of the remaining sequence */
emh203 0:3d9c67d97d6f 604 tot = ((srcALen + srcBLen) - 2u);
emh203 0:3d9c67d97d6f 605
emh203 0:3d9c67d97d6f 606 if(srcALen > srcBLen)
emh203 0:3d9c67d97d6f 607 {
emh203 0:3d9c67d97d6f 608 /* Calculating the number of zeros to be padded to the output */
emh203 0:3d9c67d97d6f 609 j = srcALen - srcBLen;
emh203 0:3d9c67d97d6f 610
emh203 0:3d9c67d97d6f 611 /* Initialise the pointer after zero padding */
emh203 0:3d9c67d97d6f 612 pDst += j;
emh203 0:3d9c67d97d6f 613 }
emh203 0:3d9c67d97d6f 614
emh203 0:3d9c67d97d6f 615 else if(srcALen < srcBLen)
emh203 0:3d9c67d97d6f 616 {
emh203 0:3d9c67d97d6f 617 /* Initialization to inputB pointer */
emh203 0:3d9c67d97d6f 618 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 619
emh203 0:3d9c67d97d6f 620 /* Initialization to the end of inputA pointer */
emh203 0:3d9c67d97d6f 621 pIn2 = pSrcA + (srcALen - 1u);
emh203 0:3d9c67d97d6f 622
emh203 0:3d9c67d97d6f 623 /* Initialisation of the pointer after zero padding */
emh203 0:3d9c67d97d6f 624 pDst = pDst + tot;
emh203 0:3d9c67d97d6f 625
emh203 0:3d9c67d97d6f 626 /* Swapping the lengths */
emh203 0:3d9c67d97d6f 627 j = srcALen;
emh203 0:3d9c67d97d6f 628 srcALen = srcBLen;
emh203 0:3d9c67d97d6f 629 srcBLen = j;
emh203 0:3d9c67d97d6f 630
emh203 0:3d9c67d97d6f 631 /* Setting the reverse flag */
emh203 0:3d9c67d97d6f 632 inv = 1;
emh203 0:3d9c67d97d6f 633
emh203 0:3d9c67d97d6f 634 }
emh203 0:3d9c67d97d6f 635
emh203 0:3d9c67d97d6f 636 /* Loop to calculate correlation for output length number of times */
emh203 0:3d9c67d97d6f 637 for (i = 0u; i <= tot; i++)
emh203 0:3d9c67d97d6f 638 {
emh203 0:3d9c67d97d6f 639 /* Initialize sum with zero to carry on MAC operations */
emh203 0:3d9c67d97d6f 640 sum = 0;
emh203 0:3d9c67d97d6f 641
emh203 0:3d9c67d97d6f 642 /* Loop to perform MAC operations according to correlation equation */
emh203 0:3d9c67d97d6f 643 for (j = 0u; j <= i; j++)
emh203 0:3d9c67d97d6f 644 {
emh203 0:3d9c67d97d6f 645 /* Check the array limitations */
emh203 0:3d9c67d97d6f 646 if((((i - j) < srcBLen) && (j < srcALen)))
emh203 0:3d9c67d97d6f 647 {
emh203 0:3d9c67d97d6f 648 /* z[i] += x[i-j] * y[j] */
emh203 0:3d9c67d97d6f 649 sum += ((q63_t) pIn1[j] * pIn2[-((int32_t) i - j)]);
emh203 0:3d9c67d97d6f 650 }
emh203 0:3d9c67d97d6f 651 }
emh203 0:3d9c67d97d6f 652 /* Store the output in the destination buffer */
emh203 0:3d9c67d97d6f 653 if(inv == 1)
emh203 0:3d9c67d97d6f 654 *pDst-- = (q31_t) (sum >> 31u);
emh203 0:3d9c67d97d6f 655 else
emh203 0:3d9c67d97d6f 656 *pDst++ = (q31_t) (sum >> 31u);
emh203 0:3d9c67d97d6f 657 }
emh203 0:3d9c67d97d6f 658
emh203 0:3d9c67d97d6f 659 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emh203 0:3d9c67d97d6f 660
emh203 0:3d9c67d97d6f 661 }
emh203 0:3d9c67d97d6f 662
emh203 0:3d9c67d97d6f 663 /**
emh203 0:3d9c67d97d6f 664 * @} end of Corr group
emh203 0:3d9c67d97d6f 665 */