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_conv_f32.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_conv_f32.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Convolution of floating-point sequences.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
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 * @defgroup Conv Convolution
emilmont 1:fdd22bb7aa52 49 *
emilmont 1:fdd22bb7aa52 50 * Convolution is a mathematical operation that operates on two finite length vectors to generate a finite length output vector.
emilmont 1:fdd22bb7aa52 51 * Convolution is similar to correlation and is frequently used in filtering and data analysis.
emilmont 1:fdd22bb7aa52 52 * The CMSIS DSP library contains functions for convolving Q7, Q15, Q31, and floating-point data types.
emilmont 1:fdd22bb7aa52 53 * The library also provides fast versions of the Q15 and Q31 functions on Cortex-M4 and Cortex-M3.
emilmont 1:fdd22bb7aa52 54 *
emilmont 1:fdd22bb7aa52 55 * \par Algorithm
emilmont 1:fdd22bb7aa52 56 * Let <code>a[n]</code> and <code>b[n]</code> be sequences of length <code>srcALen</code> and <code>srcBLen</code> samples respectively.
emilmont 1:fdd22bb7aa52 57 * Then the convolution
emilmont 1:fdd22bb7aa52 58 *
emilmont 1:fdd22bb7aa52 59 * <pre>
emilmont 1:fdd22bb7aa52 60 * c[n] = a[n] * b[n]
emilmont 1:fdd22bb7aa52 61 * </pre>
emilmont 1:fdd22bb7aa52 62 *
emilmont 1:fdd22bb7aa52 63 * \par
emilmont 1:fdd22bb7aa52 64 * is defined as
emilmont 1:fdd22bb7aa52 65 * \image html ConvolutionEquation.gif
emilmont 1:fdd22bb7aa52 66 * \par
emilmont 1:fdd22bb7aa52 67 * Note that <code>c[n]</code> is of length <code>srcALen + srcBLen - 1</code> and is defined over the interval <code>n=0, 1, 2, ..., srcALen + srcBLen - 2</code>.
emilmont 1:fdd22bb7aa52 68 * <code>pSrcA</code> points to the first input vector of length <code>srcALen</code> and
emilmont 1:fdd22bb7aa52 69 * <code>pSrcB</code> points to the second input vector of length <code>srcBLen</code>.
emilmont 1:fdd22bb7aa52 70 * The output result is written to <code>pDst</code> and the calling function must allocate <code>srcALen+srcBLen-1</code> words for the result.
emilmont 1:fdd22bb7aa52 71 *
emilmont 1:fdd22bb7aa52 72 * \par
emilmont 1:fdd22bb7aa52 73 * Conceptually, when two signals <code>a[n]</code> and <code>b[n]</code> are convolved,
emilmont 1:fdd22bb7aa52 74 * the signal <code>b[n]</code> slides over <code>a[n]</code>.
emilmont 1:fdd22bb7aa52 75 * For each offset \c n, the overlapping portions of a[n] and b[n] are multiplied and summed together.
emilmont 1:fdd22bb7aa52 76 *
emilmont 1:fdd22bb7aa52 77 * \par
emilmont 1:fdd22bb7aa52 78 * Note that convolution is a commutative operation:
emilmont 1:fdd22bb7aa52 79 *
emilmont 1:fdd22bb7aa52 80 * <pre>
emilmont 1:fdd22bb7aa52 81 * a[n] * b[n] = b[n] * a[n].
emilmont 1:fdd22bb7aa52 82 * </pre>
emilmont 1:fdd22bb7aa52 83 *
emilmont 1:fdd22bb7aa52 84 * \par
emilmont 1:fdd22bb7aa52 85 * This means that switching the A and B arguments to the convolution functions has no effect.
emilmont 1:fdd22bb7aa52 86 *
emilmont 1:fdd22bb7aa52 87 * <b>Fixed-Point Behavior</b>
emilmont 1:fdd22bb7aa52 88 *
emilmont 1:fdd22bb7aa52 89 * \par
emilmont 1:fdd22bb7aa52 90 * Convolution requires summing up a large number of intermediate products.
emilmont 1:fdd22bb7aa52 91 * As such, the Q7, Q15, and Q31 functions run a risk of overflow and saturation.
emilmont 1:fdd22bb7aa52 92 * Refer to the function specific documentation below for further details of the particular algorithm used.
emilmont 1:fdd22bb7aa52 93 *
emilmont 1:fdd22bb7aa52 94 *
emilmont 1:fdd22bb7aa52 95 * <b>Fast Versions</b>
emilmont 1:fdd22bb7aa52 96 *
emilmont 1:fdd22bb7aa52 97 * \par
emilmont 1:fdd22bb7aa52 98 * Fast versions are supported for Q31 and Q15. Cycles for Fast versions are less compared to Q31 and Q15 of conv and the design requires
emilmont 1:fdd22bb7aa52 99 * the input signals should be scaled down to avoid intermediate overflows.
emilmont 1:fdd22bb7aa52 100 *
emilmont 1:fdd22bb7aa52 101 *
emilmont 1:fdd22bb7aa52 102 * <b>Opt Versions</b>
emilmont 1:fdd22bb7aa52 103 *
emilmont 1:fdd22bb7aa52 104 * \par
emilmont 1:fdd22bb7aa52 105 * Opt versions are supported for Q15 and Q7. Design uses internal scratch buffer for getting good optimisation.
emilmont 1:fdd22bb7aa52 106 * These versions are optimised in cycles and consumes more memory(Scratch memory) compared to Q15 and Q7 versions
emilmont 1:fdd22bb7aa52 107 */
emilmont 1:fdd22bb7aa52 108
emilmont 1:fdd22bb7aa52 109 /**
emilmont 1:fdd22bb7aa52 110 * @addtogroup Conv
emilmont 1:fdd22bb7aa52 111 * @{
emilmont 1:fdd22bb7aa52 112 */
emilmont 1:fdd22bb7aa52 113
emilmont 1:fdd22bb7aa52 114 /**
emilmont 1:fdd22bb7aa52 115 * @brief Convolution of floating-point sequences.
emilmont 1:fdd22bb7aa52 116 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 117 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 118 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 119 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 120 * @param[out] *pDst points to the location where the output result is written. Length srcALen+srcBLen-1.
emilmont 1:fdd22bb7aa52 121 * @return none.
emilmont 1:fdd22bb7aa52 122 */
emilmont 1:fdd22bb7aa52 123
emilmont 1:fdd22bb7aa52 124 void arm_conv_f32(
emilmont 1:fdd22bb7aa52 125 float32_t * pSrcA,
emilmont 1:fdd22bb7aa52 126 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 127 float32_t * pSrcB,
emilmont 1:fdd22bb7aa52 128 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 129 float32_t * pDst)
emilmont 1:fdd22bb7aa52 130 {
emilmont 1:fdd22bb7aa52 131
emilmont 1:fdd22bb7aa52 132
mbed_official 3:7a284390b0ce 133 #ifndef ARM_MATH_CM0_FAMILY
emilmont 1:fdd22bb7aa52 134
emilmont 1:fdd22bb7aa52 135 /* Run the below code for Cortex-M4 and Cortex-M3 */
emilmont 1:fdd22bb7aa52 136
emilmont 1:fdd22bb7aa52 137 float32_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 138 float32_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 139 float32_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 140 float32_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 141 float32_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 142 float32_t *pSrc1, *pSrc2; /* Intermediate pointers */
emilmont 1:fdd22bb7aa52 143 float32_t sum, acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 144 float32_t x0, x1, x2, x3, c0; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 145 uint32_t j, k, count, blkCnt, blockSize1, blockSize2, blockSize3; /* loop counters */
emilmont 1:fdd22bb7aa52 146
emilmont 1:fdd22bb7aa52 147 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 148 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 149 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 150 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 151 {
emilmont 1:fdd22bb7aa52 152 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 153 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 154
emilmont 1:fdd22bb7aa52 155 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 156 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 157 }
emilmont 1:fdd22bb7aa52 158 else
emilmont 1:fdd22bb7aa52 159 {
emilmont 1:fdd22bb7aa52 160 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 161 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 162
emilmont 1:fdd22bb7aa52 163 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 164 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 165
emilmont 1:fdd22bb7aa52 166 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 167 j = srcBLen;
emilmont 1:fdd22bb7aa52 168 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 169 srcALen = j;
emilmont 1:fdd22bb7aa52 170 }
emilmont 1:fdd22bb7aa52 171
emilmont 1:fdd22bb7aa52 172 /* conv(x,y) at n = x[n] * y[0] + x[n-1] * y[1] + x[n-2] * y[2] + ...+ x[n-N+1] * y[N -1] */
emilmont 1:fdd22bb7aa52 173 /* The function is internally
emilmont 1:fdd22bb7aa52 174 * divided into three stages according to the number of multiplications that has to be
emilmont 1:fdd22bb7aa52 175 * taken place between inputA samples and inputB samples. In the first stage of the
emilmont 1:fdd22bb7aa52 176 * algorithm, the multiplications increase by one for every iteration.
emilmont 1:fdd22bb7aa52 177 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emilmont 1:fdd22bb7aa52 178 * In the third stage of the algorithm, the multiplications decrease by one
emilmont 1:fdd22bb7aa52 179 * for every iteration. */
emilmont 1:fdd22bb7aa52 180
emilmont 1:fdd22bb7aa52 181 /* The algorithm is implemented in three stages.
emilmont 1:fdd22bb7aa52 182 The loop counters of each stage is initiated here. */
emilmont 1:fdd22bb7aa52 183 blockSize1 = srcBLen - 1u;
emilmont 1:fdd22bb7aa52 184 blockSize2 = srcALen - (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 185 blockSize3 = blockSize1;
emilmont 1:fdd22bb7aa52 186
emilmont 1:fdd22bb7aa52 187 /* --------------------------
emilmont 1:fdd22bb7aa52 188 * initializations of stage1
emilmont 1:fdd22bb7aa52 189 * -------------------------*/
emilmont 1:fdd22bb7aa52 190
emilmont 1:fdd22bb7aa52 191 /* sum = x[0] * y[0]
emilmont 1:fdd22bb7aa52 192 * sum = x[0] * y[1] + x[1] * y[0]
emilmont 1:fdd22bb7aa52 193 * ....
emilmont 1:fdd22bb7aa52 194 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emilmont 1:fdd22bb7aa52 195 */
emilmont 1:fdd22bb7aa52 196
emilmont 1:fdd22bb7aa52 197 /* In this stage the MAC operations are increased by 1 for every iteration.
emilmont 1:fdd22bb7aa52 198 The count variable holds the number of MAC operations performed */
emilmont 1:fdd22bb7aa52 199 count = 1u;
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Working pointer of inputA */
emilmont 1:fdd22bb7aa52 202 px = pIn1;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 205 py = pIn2;
emilmont 1:fdd22bb7aa52 206
emilmont 1:fdd22bb7aa52 207
emilmont 1:fdd22bb7aa52 208 /* ------------------------
emilmont 1:fdd22bb7aa52 209 * Stage1 process
emilmont 1:fdd22bb7aa52 210 * ----------------------*/
emilmont 1:fdd22bb7aa52 211
emilmont 1:fdd22bb7aa52 212 /* The first stage starts here */
emilmont 1:fdd22bb7aa52 213 while(blockSize1 > 0u)
emilmont 1:fdd22bb7aa52 214 {
emilmont 1:fdd22bb7aa52 215 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 216 sum = 0.0f;
emilmont 1:fdd22bb7aa52 217
emilmont 1:fdd22bb7aa52 218 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 219 k = count >> 2u;
emilmont 1:fdd22bb7aa52 220
emilmont 1:fdd22bb7aa52 221 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 222 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 223 while(k > 0u)
emilmont 1:fdd22bb7aa52 224 {
emilmont 1:fdd22bb7aa52 225 /* x[0] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 226 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* x[1] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 229 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 230
emilmont 1:fdd22bb7aa52 231 /* x[2] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 232 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 /* x[3] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 235 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 236
emilmont 1:fdd22bb7aa52 237 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 238 k--;
emilmont 1:fdd22bb7aa52 239 }
emilmont 1:fdd22bb7aa52 240
emilmont 1:fdd22bb7aa52 241 /* If the count is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 242 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 243 k = count % 0x4u;
emilmont 1:fdd22bb7aa52 244
emilmont 1:fdd22bb7aa52 245 while(k > 0u)
emilmont 1:fdd22bb7aa52 246 {
emilmont 1:fdd22bb7aa52 247 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 248 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 249
emilmont 1:fdd22bb7aa52 250 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 251 k--;
emilmont 1:fdd22bb7aa52 252 }
emilmont 1:fdd22bb7aa52 253
emilmont 1:fdd22bb7aa52 254 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 255 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 256
emilmont 1:fdd22bb7aa52 257 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 258 py = pIn2 + count;
emilmont 1:fdd22bb7aa52 259 px = pIn1;
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 262 count++;
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 265 blockSize1--;
emilmont 1:fdd22bb7aa52 266 }
emilmont 1:fdd22bb7aa52 267
emilmont 1:fdd22bb7aa52 268 /* --------------------------
emilmont 1:fdd22bb7aa52 269 * Initializations of stage2
emilmont 1:fdd22bb7aa52 270 * ------------------------*/
emilmont 1:fdd22bb7aa52 271
emilmont 1:fdd22bb7aa52 272 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emilmont 1:fdd22bb7aa52 273 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emilmont 1:fdd22bb7aa52 274 * ....
emilmont 1:fdd22bb7aa52 275 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emilmont 1:fdd22bb7aa52 276 */
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* Working pointer of inputA */
emilmont 1:fdd22bb7aa52 279 px = pIn1;
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 282 pSrc2 = pIn2 + (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 283 py = pSrc2;
emilmont 1:fdd22bb7aa52 284
emilmont 1:fdd22bb7aa52 285 /* count is index by which the pointer pIn1 to be incremented */
emilmont 1:fdd22bb7aa52 286 count = 0u;
emilmont 1:fdd22bb7aa52 287
emilmont 1:fdd22bb7aa52 288 /* -------------------
emilmont 1:fdd22bb7aa52 289 * Stage2 process
emilmont 1:fdd22bb7aa52 290 * ------------------*/
emilmont 1:fdd22bb7aa52 291
emilmont 1:fdd22bb7aa52 292 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emilmont 1:fdd22bb7aa52 293 * So, to loop unroll over blockSize2,
emilmont 1:fdd22bb7aa52 294 * srcBLen should be greater than or equal to 4 */
emilmont 1:fdd22bb7aa52 295 if(srcBLen >= 4u)
emilmont 1:fdd22bb7aa52 296 {
emilmont 1:fdd22bb7aa52 297 /* Loop unroll over blockSize2, by 4 */
emilmont 1:fdd22bb7aa52 298 blkCnt = blockSize2 >> 2u;
emilmont 1:fdd22bb7aa52 299
emilmont 1:fdd22bb7aa52 300 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 301 {
emilmont 1:fdd22bb7aa52 302 /* Set all accumulators to zero */
emilmont 1:fdd22bb7aa52 303 acc0 = 0.0f;
emilmont 1:fdd22bb7aa52 304 acc1 = 0.0f;
emilmont 1:fdd22bb7aa52 305 acc2 = 0.0f;
emilmont 1:fdd22bb7aa52 306 acc3 = 0.0f;
emilmont 1:fdd22bb7aa52 307
emilmont 1:fdd22bb7aa52 308 /* read x[0], x[1], x[2] samples */
emilmont 1:fdd22bb7aa52 309 x0 = *(px++);
emilmont 1:fdd22bb7aa52 310 x1 = *(px++);
emilmont 1:fdd22bb7aa52 311 x2 = *(px++);
emilmont 1:fdd22bb7aa52 312
emilmont 1:fdd22bb7aa52 313 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 314 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 315
emilmont 1:fdd22bb7aa52 316 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 317 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 318 do
emilmont 1:fdd22bb7aa52 319 {
emilmont 1:fdd22bb7aa52 320 /* Read y[srcBLen - 1] sample */
emilmont 1:fdd22bb7aa52 321 c0 = *(py--);
emilmont 1:fdd22bb7aa52 322
emilmont 1:fdd22bb7aa52 323 /* Read x[3] sample */
emilmont 1:fdd22bb7aa52 324 x3 = *(px);
emilmont 1:fdd22bb7aa52 325
emilmont 1:fdd22bb7aa52 326 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 327 /* acc0 += x[0] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 328 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 329
emilmont 1:fdd22bb7aa52 330 /* acc1 += x[1] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 331 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 332
emilmont 1:fdd22bb7aa52 333 /* acc2 += x[2] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 334 acc2 += x2 * c0;
emilmont 1:fdd22bb7aa52 335
emilmont 1:fdd22bb7aa52 336 /* acc3 += x[3] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 337 acc3 += x3 * c0;
emilmont 1:fdd22bb7aa52 338
emilmont 1:fdd22bb7aa52 339 /* Read y[srcBLen - 2] sample */
emilmont 1:fdd22bb7aa52 340 c0 = *(py--);
emilmont 1:fdd22bb7aa52 341
emilmont 1:fdd22bb7aa52 342 /* Read x[4] sample */
emilmont 1:fdd22bb7aa52 343 x0 = *(px + 1u);
emilmont 1:fdd22bb7aa52 344
emilmont 1:fdd22bb7aa52 345 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 346 /* acc0 += x[1] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 347 acc0 += x1 * c0;
emilmont 1:fdd22bb7aa52 348 /* acc1 += x[2] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 349 acc1 += x2 * c0;
emilmont 1:fdd22bb7aa52 350 /* acc2 += x[3] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 351 acc2 += x3 * c0;
emilmont 1:fdd22bb7aa52 352 /* acc3 += x[4] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 353 acc3 += x0 * c0;
emilmont 1:fdd22bb7aa52 354
emilmont 1:fdd22bb7aa52 355 /* Read y[srcBLen - 3] sample */
emilmont 1:fdd22bb7aa52 356 c0 = *(py--);
emilmont 1:fdd22bb7aa52 357
emilmont 1:fdd22bb7aa52 358 /* Read x[5] sample */
emilmont 1:fdd22bb7aa52 359 x1 = *(px + 2u);
emilmont 1:fdd22bb7aa52 360
emilmont 1:fdd22bb7aa52 361 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 362 /* acc0 += x[2] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 363 acc0 += x2 * c0;
emilmont 1:fdd22bb7aa52 364 /* acc1 += x[3] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 365 acc1 += x3 * c0;
emilmont 1:fdd22bb7aa52 366 /* acc2 += x[4] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 367 acc2 += x0 * c0;
emilmont 1:fdd22bb7aa52 368 /* acc3 += x[5] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 369 acc3 += x1 * c0;
emilmont 1:fdd22bb7aa52 370
emilmont 1:fdd22bb7aa52 371 /* Read y[srcBLen - 4] sample */
emilmont 1:fdd22bb7aa52 372 c0 = *(py--);
emilmont 1:fdd22bb7aa52 373
emilmont 1:fdd22bb7aa52 374 /* Read x[6] sample */
emilmont 1:fdd22bb7aa52 375 x2 = *(px + 3u);
emilmont 1:fdd22bb7aa52 376 px += 4u;
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 379 /* acc0 += x[3] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 380 acc0 += x3 * c0;
emilmont 1:fdd22bb7aa52 381 /* acc1 += x[4] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 382 acc1 += x0 * c0;
emilmont 1:fdd22bb7aa52 383 /* acc2 += x[5] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 384 acc2 += x1 * c0;
emilmont 1:fdd22bb7aa52 385 /* acc3 += x[6] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 386 acc3 += x2 * c0;
emilmont 1:fdd22bb7aa52 387
emilmont 1:fdd22bb7aa52 388
emilmont 1:fdd22bb7aa52 389 } while(--k);
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 392 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 393 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 394
emilmont 1:fdd22bb7aa52 395 while(k > 0u)
emilmont 1:fdd22bb7aa52 396 {
emilmont 1:fdd22bb7aa52 397 /* Read y[srcBLen - 5] sample */
emilmont 1:fdd22bb7aa52 398 c0 = *(py--);
emilmont 1:fdd22bb7aa52 399
emilmont 1:fdd22bb7aa52 400 /* Read x[7] sample */
emilmont 1:fdd22bb7aa52 401 x3 = *(px++);
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 404 /* acc0 += x[4] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 405 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 406 /* acc1 += x[5] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 407 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 408 /* acc2 += x[6] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 409 acc2 += x2 * c0;
emilmont 1:fdd22bb7aa52 410 /* acc3 += x[7] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 411 acc3 += x3 * c0;
emilmont 1:fdd22bb7aa52 412
emilmont 1:fdd22bb7aa52 413 /* Reuse the present samples for the next MAC */
emilmont 1:fdd22bb7aa52 414 x0 = x1;
emilmont 1:fdd22bb7aa52 415 x1 = x2;
emilmont 1:fdd22bb7aa52 416 x2 = x3;
emilmont 1:fdd22bb7aa52 417
emilmont 1:fdd22bb7aa52 418 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 419 k--;
emilmont 1:fdd22bb7aa52 420 }
emilmont 1:fdd22bb7aa52 421
emilmont 1:fdd22bb7aa52 422 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 423 *pOut++ = acc0;
emilmont 1:fdd22bb7aa52 424 *pOut++ = acc1;
emilmont 1:fdd22bb7aa52 425 *pOut++ = acc2;
emilmont 1:fdd22bb7aa52 426 *pOut++ = acc3;
emilmont 1:fdd22bb7aa52 427
emilmont 1:fdd22bb7aa52 428 /* Increment the pointer pIn1 index, count by 4 */
emilmont 1:fdd22bb7aa52 429 count += 4u;
emilmont 1:fdd22bb7aa52 430
emilmont 1:fdd22bb7aa52 431 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 432 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 433 py = pSrc2;
emilmont 1:fdd22bb7aa52 434
emilmont 1:fdd22bb7aa52 435
emilmont 1:fdd22bb7aa52 436 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 437 blkCnt--;
emilmont 1:fdd22bb7aa52 438 }
emilmont 1:fdd22bb7aa52 439
emilmont 1:fdd22bb7aa52 440
emilmont 1:fdd22bb7aa52 441 /* If the blockSize2 is not a multiple of 4, compute any remaining output samples here.
emilmont 1:fdd22bb7aa52 442 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 443 blkCnt = blockSize2 % 0x4u;
emilmont 1:fdd22bb7aa52 444
emilmont 1:fdd22bb7aa52 445 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 446 {
emilmont 1:fdd22bb7aa52 447 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 448 sum = 0.0f;
emilmont 1:fdd22bb7aa52 449
emilmont 1:fdd22bb7aa52 450 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 451 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 452
emilmont 1:fdd22bb7aa52 453 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 454 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 455 while(k > 0u)
emilmont 1:fdd22bb7aa52 456 {
emilmont 1:fdd22bb7aa52 457 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 458 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 459 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 460 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 461 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 462
emilmont 1:fdd22bb7aa52 463 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 464 k--;
emilmont 1:fdd22bb7aa52 465 }
emilmont 1:fdd22bb7aa52 466
emilmont 1:fdd22bb7aa52 467 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 468 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 469 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 470
emilmont 1:fdd22bb7aa52 471 while(k > 0u)
emilmont 1:fdd22bb7aa52 472 {
emilmont 1:fdd22bb7aa52 473 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 474 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 475
emilmont 1:fdd22bb7aa52 476 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 477 k--;
emilmont 1:fdd22bb7aa52 478 }
emilmont 1:fdd22bb7aa52 479
emilmont 1:fdd22bb7aa52 480 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 481 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 482
emilmont 1:fdd22bb7aa52 483 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 484 count++;
emilmont 1:fdd22bb7aa52 485
emilmont 1:fdd22bb7aa52 486 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 487 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 488 py = pSrc2;
emilmont 1:fdd22bb7aa52 489
emilmont 1:fdd22bb7aa52 490 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 491 blkCnt--;
emilmont 1:fdd22bb7aa52 492 }
emilmont 1:fdd22bb7aa52 493 }
emilmont 1:fdd22bb7aa52 494 else
emilmont 1:fdd22bb7aa52 495 {
emilmont 1:fdd22bb7aa52 496 /* If the srcBLen is not a multiple of 4,
emilmont 1:fdd22bb7aa52 497 * the blockSize2 loop cannot be unrolled by 4 */
emilmont 1:fdd22bb7aa52 498 blkCnt = blockSize2;
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 501 {
emilmont 1:fdd22bb7aa52 502 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 503 sum = 0.0f;
emilmont 1:fdd22bb7aa52 504
emilmont 1:fdd22bb7aa52 505 /* srcBLen number of MACS should be performed */
emilmont 1:fdd22bb7aa52 506 k = srcBLen;
emilmont 1:fdd22bb7aa52 507
emilmont 1:fdd22bb7aa52 508 while(k > 0u)
emilmont 1:fdd22bb7aa52 509 {
emilmont 1:fdd22bb7aa52 510 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 511 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 512
emilmont 1:fdd22bb7aa52 513 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 514 k--;
emilmont 1:fdd22bb7aa52 515 }
emilmont 1:fdd22bb7aa52 516
emilmont 1:fdd22bb7aa52 517 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 518 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 519
emilmont 1:fdd22bb7aa52 520 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 521 count++;
emilmont 1:fdd22bb7aa52 522
emilmont 1:fdd22bb7aa52 523 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 524 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 525 py = pSrc2;
emilmont 1:fdd22bb7aa52 526
emilmont 1:fdd22bb7aa52 527 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 528 blkCnt--;
emilmont 1:fdd22bb7aa52 529 }
emilmont 1:fdd22bb7aa52 530 }
emilmont 1:fdd22bb7aa52 531
emilmont 1:fdd22bb7aa52 532
emilmont 1:fdd22bb7aa52 533 /* --------------------------
emilmont 1:fdd22bb7aa52 534 * Initializations of stage3
emilmont 1:fdd22bb7aa52 535 * -------------------------*/
emilmont 1:fdd22bb7aa52 536
emilmont 1:fdd22bb7aa52 537 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emilmont 1:fdd22bb7aa52 538 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emilmont 1:fdd22bb7aa52 539 * ....
emilmont 1:fdd22bb7aa52 540 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emilmont 1:fdd22bb7aa52 541 * sum += x[srcALen-1] * y[srcBLen-1]
emilmont 1:fdd22bb7aa52 542 */
emilmont 1:fdd22bb7aa52 543
emilmont 1:fdd22bb7aa52 544 /* In this stage the MAC operations are decreased by 1 for every iteration.
emilmont 1:fdd22bb7aa52 545 The blockSize3 variable holds the number of MAC operations performed */
emilmont 1:fdd22bb7aa52 546
emilmont 1:fdd22bb7aa52 547 /* Working pointer of inputA */
emilmont 1:fdd22bb7aa52 548 pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 549 px = pSrc1;
emilmont 1:fdd22bb7aa52 550
emilmont 1:fdd22bb7aa52 551 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 552 pSrc2 = pIn2 + (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 553 py = pSrc2;
emilmont 1:fdd22bb7aa52 554
emilmont 1:fdd22bb7aa52 555 /* -------------------
emilmont 1:fdd22bb7aa52 556 * Stage3 process
emilmont 1:fdd22bb7aa52 557 * ------------------*/
emilmont 1:fdd22bb7aa52 558
emilmont 1:fdd22bb7aa52 559 while(blockSize3 > 0u)
emilmont 1:fdd22bb7aa52 560 {
emilmont 1:fdd22bb7aa52 561 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 562 sum = 0.0f;
emilmont 1:fdd22bb7aa52 563
emilmont 1:fdd22bb7aa52 564 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 565 k = blockSize3 >> 2u;
emilmont 1:fdd22bb7aa52 566
emilmont 1:fdd22bb7aa52 567 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 568 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 569 while(k > 0u)
emilmont 1:fdd22bb7aa52 570 {
emilmont 1:fdd22bb7aa52 571 /* sum += x[srcALen - srcBLen + 1] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 572 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 573
emilmont 1:fdd22bb7aa52 574 /* sum += x[srcALen - srcBLen + 2] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 575 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 576
emilmont 1:fdd22bb7aa52 577 /* sum += x[srcALen - srcBLen + 3] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 578 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 579
emilmont 1:fdd22bb7aa52 580 /* sum += x[srcALen - srcBLen + 4] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 581 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 582
emilmont 1:fdd22bb7aa52 583 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 584 k--;
emilmont 1:fdd22bb7aa52 585 }
emilmont 1:fdd22bb7aa52 586
emilmont 1:fdd22bb7aa52 587 /* If the blockSize3 is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 588 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 589 k = blockSize3 % 0x4u;
emilmont 1:fdd22bb7aa52 590
emilmont 1:fdd22bb7aa52 591 while(k > 0u)
emilmont 1:fdd22bb7aa52 592 {
emilmont 1:fdd22bb7aa52 593 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 594 /* sum += x[srcALen-1] * y[srcBLen-1] */
emilmont 1:fdd22bb7aa52 595 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 596
emilmont 1:fdd22bb7aa52 597 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 598 k--;
emilmont 1:fdd22bb7aa52 599 }
emilmont 1:fdd22bb7aa52 600
emilmont 1:fdd22bb7aa52 601 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 602 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 603
emilmont 1:fdd22bb7aa52 604 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 605 px = ++pSrc1;
emilmont 1:fdd22bb7aa52 606 py = pSrc2;
emilmont 1:fdd22bb7aa52 607
emilmont 1:fdd22bb7aa52 608 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 609 blockSize3--;
emilmont 1:fdd22bb7aa52 610 }
emilmont 1:fdd22bb7aa52 611
emilmont 1:fdd22bb7aa52 612 #else
emilmont 1:fdd22bb7aa52 613
emilmont 1:fdd22bb7aa52 614 /* Run the below code for Cortex-M0 */
emilmont 1:fdd22bb7aa52 615
emilmont 1:fdd22bb7aa52 616 float32_t *pIn1 = pSrcA; /* inputA pointer */
emilmont 1:fdd22bb7aa52 617 float32_t *pIn2 = pSrcB; /* inputB pointer */
emilmont 1:fdd22bb7aa52 618 float32_t sum; /* Accumulator */
emilmont 1:fdd22bb7aa52 619 uint32_t i, j; /* loop counters */
emilmont 1:fdd22bb7aa52 620
emilmont 1:fdd22bb7aa52 621 /* Loop to calculate convolution for output length number of times */
emilmont 1:fdd22bb7aa52 622 for (i = 0u; i < ((srcALen + srcBLen) - 1u); i++)
emilmont 1:fdd22bb7aa52 623 {
emilmont 1:fdd22bb7aa52 624 /* Initialize sum with zero to carry out MAC operations */
emilmont 1:fdd22bb7aa52 625 sum = 0.0f;
emilmont 1:fdd22bb7aa52 626
emilmont 1:fdd22bb7aa52 627 /* Loop to perform MAC operations according to convolution equation */
emilmont 1:fdd22bb7aa52 628 for (j = 0u; j <= i; j++)
emilmont 1:fdd22bb7aa52 629 {
emilmont 1:fdd22bb7aa52 630 /* Check the array limitations */
emilmont 1:fdd22bb7aa52 631 if((((i - j) < srcBLen) && (j < srcALen)))
emilmont 1:fdd22bb7aa52 632 {
emilmont 1:fdd22bb7aa52 633 /* z[i] += x[i-j] * y[j] */
emilmont 1:fdd22bb7aa52 634 sum += pIn1[j] * pIn2[i - j];
emilmont 1:fdd22bb7aa52 635 }
emilmont 1:fdd22bb7aa52 636 }
emilmont 1:fdd22bb7aa52 637 /* Store the output in the destination buffer */
emilmont 1:fdd22bb7aa52 638 pDst[i] = sum;
emilmont 1:fdd22bb7aa52 639 }
emilmont 1:fdd22bb7aa52 640
mbed_official 3:7a284390b0ce 641 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emilmont 1:fdd22bb7aa52 642
emilmont 1:fdd22bb7aa52 643 }
emilmont 1:fdd22bb7aa52 644
emilmont 1:fdd22bb7aa52 645 /**
emilmont 1:fdd22bb7aa52 646 * @} end of Conv group
emilmont 1:fdd22bb7aa52 647 */