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_partial_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_partial_f32.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Partial 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 PartialConv Partial Convolution
emilmont 1:fdd22bb7aa52 49 *
emilmont 1:fdd22bb7aa52 50 * Partial Convolution is equivalent to Convolution except that a subset of the output samples is generated.
emilmont 1:fdd22bb7aa52 51 * Each function has two additional arguments.
emilmont 1:fdd22bb7aa52 52 * <code>firstIndex</code> specifies the starting index of the subset of output samples.
emilmont 1:fdd22bb7aa52 53 * <code>numPoints</code> is the number of output samples to compute.
emilmont 1:fdd22bb7aa52 54 * The function computes the output in the range
emilmont 1:fdd22bb7aa52 55 * <code>[firstIndex, ..., firstIndex+numPoints-1]</code>.
emilmont 1:fdd22bb7aa52 56 * The output array <code>pDst</code> contains <code>numPoints</code> values.
emilmont 1:fdd22bb7aa52 57 *
emilmont 1:fdd22bb7aa52 58 * The allowable range of output indices is [0 srcALen+srcBLen-2].
emilmont 1:fdd22bb7aa52 59 * If the requested subset does not fall in this range then the functions return ARM_MATH_ARGUMENT_ERROR.
emilmont 1:fdd22bb7aa52 60 * Otherwise the functions return ARM_MATH_SUCCESS.
emilmont 1:fdd22bb7aa52 61 * \note Refer arm_conv_f32() for details on fixed point behavior.
emilmont 1:fdd22bb7aa52 62 *
emilmont 1:fdd22bb7aa52 63 *
emilmont 1:fdd22bb7aa52 64 * <b>Fast Versions</b>
emilmont 1:fdd22bb7aa52 65 *
emilmont 1:fdd22bb7aa52 66 * \par
emilmont 1:fdd22bb7aa52 67 * Fast versions are supported for Q31 and Q15 of partial convolution. Cycles for Fast versions are less compared to Q31 and Q15 of partial conv and the design requires
emilmont 1:fdd22bb7aa52 68 * the input signals should be scaled down to avoid intermediate overflows.
emilmont 1:fdd22bb7aa52 69 *
emilmont 1:fdd22bb7aa52 70 *
emilmont 1:fdd22bb7aa52 71 * <b>Opt Versions</b>
emilmont 1:fdd22bb7aa52 72 *
emilmont 1:fdd22bb7aa52 73 * \par
emilmont 1:fdd22bb7aa52 74 * Opt versions are supported for Q15 and Q7. Design uses internal scratch buffer for getting good optimisation.
emilmont 1:fdd22bb7aa52 75 * These versions are optimised in cycles and consumes more memory(Scratch memory) compared to Q15 and Q7 versions of partial convolution
emilmont 1:fdd22bb7aa52 76 */
emilmont 1:fdd22bb7aa52 77
emilmont 1:fdd22bb7aa52 78 /**
emilmont 1:fdd22bb7aa52 79 * @addtogroup PartialConv
emilmont 1:fdd22bb7aa52 80 * @{
emilmont 1:fdd22bb7aa52 81 */
emilmont 1:fdd22bb7aa52 82
emilmont 1:fdd22bb7aa52 83 /**
emilmont 1:fdd22bb7aa52 84 * @brief Partial convolution of floating-point sequences.
emilmont 1:fdd22bb7aa52 85 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 86 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 87 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 88 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 89 * @param[out] *pDst points to the location where the output result is written.
emilmont 1:fdd22bb7aa52 90 * @param[in] firstIndex is the first output sample to start with.
emilmont 1:fdd22bb7aa52 91 * @param[in] numPoints is the number of output points to be computed.
emilmont 1:fdd22bb7aa52 92 * @return Returns either ARM_MATH_SUCCESS if the function completed correctly or ARM_MATH_ARGUMENT_ERROR if the requested subset is not in the range [0 srcALen+srcBLen-2].
emilmont 1:fdd22bb7aa52 93 */
emilmont 1:fdd22bb7aa52 94
emilmont 1:fdd22bb7aa52 95 arm_status arm_conv_partial_f32(
emilmont 1:fdd22bb7aa52 96 float32_t * pSrcA,
emilmont 1:fdd22bb7aa52 97 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 98 float32_t * pSrcB,
emilmont 1:fdd22bb7aa52 99 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 100 float32_t * pDst,
emilmont 1:fdd22bb7aa52 101 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 102 uint32_t numPoints)
emilmont 1:fdd22bb7aa52 103 {
emilmont 1:fdd22bb7aa52 104
emilmont 1:fdd22bb7aa52 105
mbed_official 3:7a284390b0ce 106 #ifndef ARM_MATH_CM0_FAMILY
emilmont 1:fdd22bb7aa52 107
emilmont 1:fdd22bb7aa52 108 /* Run the below code for Cortex-M4 and Cortex-M3 */
emilmont 1:fdd22bb7aa52 109
emilmont 1:fdd22bb7aa52 110 float32_t *pIn1 = pSrcA; /* inputA pointer */
emilmont 1:fdd22bb7aa52 111 float32_t *pIn2 = pSrcB; /* inputB pointer */
emilmont 1:fdd22bb7aa52 112 float32_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 113 float32_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 114 float32_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 115 float32_t *pSrc1, *pSrc2; /* Intermediate pointers */
emilmont 1:fdd22bb7aa52 116 float32_t sum, acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 117 float32_t x0, x1, x2, x3, c0; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 118 uint32_t j, k, count = 0u, blkCnt, check;
emilmont 1:fdd22bb7aa52 119 int32_t blockSize1, blockSize2, blockSize3; /* loop counters */
emilmont 1:fdd22bb7aa52 120 arm_status status; /* status of Partial convolution */
emilmont 1:fdd22bb7aa52 121
emilmont 1:fdd22bb7aa52 122
emilmont 1:fdd22bb7aa52 123 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 124 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 125 {
emilmont 1:fdd22bb7aa52 126 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 127 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 128 }
emilmont 1:fdd22bb7aa52 129 else
emilmont 1:fdd22bb7aa52 130 {
emilmont 1:fdd22bb7aa52 131
emilmont 1:fdd22bb7aa52 132 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 133 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 134 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 135 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 136 {
emilmont 1:fdd22bb7aa52 137 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 138 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 139
emilmont 1:fdd22bb7aa52 140 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 141 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 142 }
emilmont 1:fdd22bb7aa52 143 else
emilmont 1:fdd22bb7aa52 144 {
emilmont 1:fdd22bb7aa52 145 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 146 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 149 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 150
emilmont 1:fdd22bb7aa52 151 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 152 j = srcBLen;
emilmont 1:fdd22bb7aa52 153 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 154 srcALen = j;
emilmont 1:fdd22bb7aa52 155 }
emilmont 1:fdd22bb7aa52 156
emilmont 1:fdd22bb7aa52 157 /* Conditions to check which loopCounter holds
emilmont 1:fdd22bb7aa52 158 * the first and last indices of the output samples to be calculated. */
emilmont 1:fdd22bb7aa52 159 check = firstIndex + numPoints;
mbed_official 5:3762170b6d4d 160 blockSize3 = ((int32_t)check > (int32_t)srcALen) ? (int32_t)check - (int32_t)srcALen : 0;
mbed_official 5:3762170b6d4d 161 blockSize3 = ((int32_t)firstIndex > (int32_t)srcALen - 1) ? blockSize3 - (int32_t)firstIndex + (int32_t)srcALen : blockSize3;
emilmont 1:fdd22bb7aa52 162 blockSize1 = ((int32_t) srcBLen - 1) - (int32_t) firstIndex;
emilmont 1:fdd22bb7aa52 163 blockSize1 = (blockSize1 > 0) ? ((check > (srcBLen - 1u)) ? blockSize1 :
emilmont 1:fdd22bb7aa52 164 (int32_t) numPoints) : 0;
emilmont 1:fdd22bb7aa52 165 blockSize2 = ((int32_t) check - blockSize3) -
emilmont 1:fdd22bb7aa52 166 (blockSize1 + (int32_t) firstIndex);
emilmont 1:fdd22bb7aa52 167 blockSize2 = (blockSize2 > 0) ? blockSize2 : 0;
emilmont 1:fdd22bb7aa52 168
emilmont 1:fdd22bb7aa52 169 /* 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 170 /* The function is internally
emilmont 1:fdd22bb7aa52 171 * divided into three stages according to the number of multiplications that has to be
emilmont 1:fdd22bb7aa52 172 * taken place between inputA samples and inputB samples. In the first stage of the
emilmont 1:fdd22bb7aa52 173 * algorithm, the multiplications increase by one for every iteration.
emilmont 1:fdd22bb7aa52 174 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emilmont 1:fdd22bb7aa52 175 * In the third stage of the algorithm, the multiplications decrease by one
emilmont 1:fdd22bb7aa52 176 * for every iteration. */
emilmont 1:fdd22bb7aa52 177
emilmont 1:fdd22bb7aa52 178 /* Set the output pointer to point to the firstIndex
emilmont 1:fdd22bb7aa52 179 * of the output sample to be calculated. */
emilmont 1:fdd22bb7aa52 180 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 181
emilmont 1:fdd22bb7aa52 182 /* --------------------------
emilmont 1:fdd22bb7aa52 183 * Initializations of stage1
emilmont 1:fdd22bb7aa52 184 * -------------------------*/
emilmont 1:fdd22bb7aa52 185
emilmont 1:fdd22bb7aa52 186 /* sum = x[0] * y[0]
emilmont 1:fdd22bb7aa52 187 * sum = x[0] * y[1] + x[1] * y[0]
emilmont 1:fdd22bb7aa52 188 * ....
emilmont 1:fdd22bb7aa52 189 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emilmont 1:fdd22bb7aa52 190 */
emilmont 1:fdd22bb7aa52 191
emilmont 1:fdd22bb7aa52 192 /* In this stage the MAC operations are increased by 1 for every iteration.
emilmont 1:fdd22bb7aa52 193 The count variable holds the number of MAC operations performed.
emilmont 1:fdd22bb7aa52 194 Since the partial convolution starts from from firstIndex
emilmont 1:fdd22bb7aa52 195 Number of Macs to be performed is firstIndex + 1 */
emilmont 1:fdd22bb7aa52 196 count = 1u + firstIndex;
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198 /* Working pointer of inputA */
emilmont 1:fdd22bb7aa52 199 px = pIn1;
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 202 pSrc1 = pIn2 + firstIndex;
emilmont 1:fdd22bb7aa52 203 py = pSrc1;
emilmont 1:fdd22bb7aa52 204
emilmont 1:fdd22bb7aa52 205 /* ------------------------
emilmont 1:fdd22bb7aa52 206 * Stage1 process
emilmont 1:fdd22bb7aa52 207 * ----------------------*/
emilmont 1:fdd22bb7aa52 208
emilmont 1:fdd22bb7aa52 209 /* The first stage starts here */
emilmont 1:fdd22bb7aa52 210 while(blockSize1 > 0)
emilmont 1:fdd22bb7aa52 211 {
emilmont 1:fdd22bb7aa52 212 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 213 sum = 0.0f;
emilmont 1:fdd22bb7aa52 214
emilmont 1:fdd22bb7aa52 215 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 216 k = count >> 2u;
emilmont 1:fdd22bb7aa52 217
emilmont 1:fdd22bb7aa52 218 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 219 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 220 while(k > 0u)
emilmont 1:fdd22bb7aa52 221 {
emilmont 1:fdd22bb7aa52 222 /* x[0] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 223 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225 /* x[1] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 226 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* x[2] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 229 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 230
emilmont 1:fdd22bb7aa52 231 /* x[3] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 232 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 235 k--;
emilmont 1:fdd22bb7aa52 236 }
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* If the count is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 239 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 240 k = count % 0x4u;
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 while(k > 0u)
emilmont 1:fdd22bb7aa52 243 {
emilmont 1:fdd22bb7aa52 244 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 245 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 246
emilmont 1:fdd22bb7aa52 247 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 248 k--;
emilmont 1:fdd22bb7aa52 249 }
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 252 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 253
emilmont 1:fdd22bb7aa52 254 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 255 py = ++pSrc1;
emilmont 1:fdd22bb7aa52 256 px = pIn1;
emilmont 1:fdd22bb7aa52 257
emilmont 1:fdd22bb7aa52 258 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 259 count++;
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 262 blockSize1--;
emilmont 1:fdd22bb7aa52 263 }
emilmont 1:fdd22bb7aa52 264
emilmont 1:fdd22bb7aa52 265 /* --------------------------
emilmont 1:fdd22bb7aa52 266 * Initializations of stage2
emilmont 1:fdd22bb7aa52 267 * ------------------------*/
emilmont 1:fdd22bb7aa52 268
emilmont 1:fdd22bb7aa52 269 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emilmont 1:fdd22bb7aa52 270 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emilmont 1:fdd22bb7aa52 271 * ....
emilmont 1:fdd22bb7aa52 272 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emilmont 1:fdd22bb7aa52 273 */
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* Working pointer of inputA */
mbed_official 5:3762170b6d4d 276 if((int32_t)firstIndex - (int32_t)srcBLen + 1 > 0)
mbed_official 5:3762170b6d4d 277 {
mbed_official 5:3762170b6d4d 278 px = pIn1 + firstIndex - srcBLen + 1;
mbed_official 5:3762170b6d4d 279 }
mbed_official 5:3762170b6d4d 280 else
mbed_official 5:3762170b6d4d 281 {
mbed_official 5:3762170b6d4d 282 px = pIn1;
mbed_official 5:3762170b6d4d 283 }
emilmont 1:fdd22bb7aa52 284
emilmont 1:fdd22bb7aa52 285 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 286 pSrc2 = pIn2 + (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 287 py = pSrc2;
emilmont 1:fdd22bb7aa52 288
emilmont 1:fdd22bb7aa52 289 /* count is index by which the pointer pIn1 to be incremented */
emilmont 1:fdd22bb7aa52 290 count = 0u;
emilmont 1:fdd22bb7aa52 291
emilmont 1:fdd22bb7aa52 292 /* -------------------
emilmont 1:fdd22bb7aa52 293 * Stage2 process
emilmont 1:fdd22bb7aa52 294 * ------------------*/
emilmont 1:fdd22bb7aa52 295
emilmont 1:fdd22bb7aa52 296 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emilmont 1:fdd22bb7aa52 297 * So, to loop unroll over blockSize2,
emilmont 1:fdd22bb7aa52 298 * srcBLen should be greater than or equal to 4 */
emilmont 1:fdd22bb7aa52 299 if(srcBLen >= 4u)
emilmont 1:fdd22bb7aa52 300 {
emilmont 1:fdd22bb7aa52 301 /* Loop unroll over blockSize2, by 4 */
emilmont 1:fdd22bb7aa52 302 blkCnt = ((uint32_t) blockSize2 >> 2u);
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 305 {
emilmont 1:fdd22bb7aa52 306 /* Set all accumulators to zero */
emilmont 1:fdd22bb7aa52 307 acc0 = 0.0f;
emilmont 1:fdd22bb7aa52 308 acc1 = 0.0f;
emilmont 1:fdd22bb7aa52 309 acc2 = 0.0f;
emilmont 1:fdd22bb7aa52 310 acc3 = 0.0f;
emilmont 1:fdd22bb7aa52 311
emilmont 1:fdd22bb7aa52 312 /* read x[0], x[1], x[2] samples */
emilmont 1:fdd22bb7aa52 313 x0 = *(px++);
emilmont 1:fdd22bb7aa52 314 x1 = *(px++);
emilmont 1:fdd22bb7aa52 315 x2 = *(px++);
emilmont 1:fdd22bb7aa52 316
emilmont 1:fdd22bb7aa52 317 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 318 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 321 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 322 do
emilmont 1:fdd22bb7aa52 323 {
emilmont 1:fdd22bb7aa52 324 /* Read y[srcBLen - 1] sample */
emilmont 1:fdd22bb7aa52 325 c0 = *(py--);
emilmont 1:fdd22bb7aa52 326
emilmont 1:fdd22bb7aa52 327 /* Read x[3] sample */
emilmont 1:fdd22bb7aa52 328 x3 = *(px++);
emilmont 1:fdd22bb7aa52 329
emilmont 1:fdd22bb7aa52 330 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 331 /* acc0 += x[0] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 332 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 333
emilmont 1:fdd22bb7aa52 334 /* acc1 += x[1] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 335 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 336
emilmont 1:fdd22bb7aa52 337 /* acc2 += x[2] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 338 acc2 += x2 * c0;
emilmont 1:fdd22bb7aa52 339
emilmont 1:fdd22bb7aa52 340 /* acc3 += x[3] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 341 acc3 += x3 * c0;
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 /* Read y[srcBLen - 2] sample */
emilmont 1:fdd22bb7aa52 344 c0 = *(py--);
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346 /* Read x[4] sample */
emilmont 1:fdd22bb7aa52 347 x0 = *(px++);
emilmont 1:fdd22bb7aa52 348
emilmont 1:fdd22bb7aa52 349 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 350 /* acc0 += x[1] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 351 acc0 += x1 * c0;
emilmont 1:fdd22bb7aa52 352 /* acc1 += x[2] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 353 acc1 += x2 * c0;
emilmont 1:fdd22bb7aa52 354 /* acc2 += x[3] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 355 acc2 += x3 * c0;
emilmont 1:fdd22bb7aa52 356 /* acc3 += x[4] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 357 acc3 += x0 * c0;
emilmont 1:fdd22bb7aa52 358
emilmont 1:fdd22bb7aa52 359 /* Read y[srcBLen - 3] sample */
emilmont 1:fdd22bb7aa52 360 c0 = *(py--);
emilmont 1:fdd22bb7aa52 361
emilmont 1:fdd22bb7aa52 362 /* Read x[5] sample */
emilmont 1:fdd22bb7aa52 363 x1 = *(px++);
emilmont 1:fdd22bb7aa52 364
emilmont 1:fdd22bb7aa52 365 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 366 /* acc0 += x[2] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 367 acc0 += x2 * c0;
emilmont 1:fdd22bb7aa52 368 /* acc1 += x[3] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 369 acc1 += x3 * c0;
emilmont 1:fdd22bb7aa52 370 /* acc2 += x[4] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 371 acc2 += x0 * c0;
emilmont 1:fdd22bb7aa52 372 /* acc3 += x[5] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 373 acc3 += x1 * c0;
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 /* Read y[srcBLen - 4] sample */
emilmont 1:fdd22bb7aa52 376 c0 = *(py--);
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 /* Read x[6] sample */
emilmont 1:fdd22bb7aa52 379 x2 = *(px++);
emilmont 1:fdd22bb7aa52 380
emilmont 1:fdd22bb7aa52 381 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 382 /* acc0 += x[3] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 383 acc0 += x3 * c0;
emilmont 1:fdd22bb7aa52 384 /* acc1 += x[4] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 385 acc1 += x0 * c0;
emilmont 1:fdd22bb7aa52 386 /* acc2 += x[5] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 387 acc2 += x1 * c0;
emilmont 1:fdd22bb7aa52 388 /* acc3 += x[6] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 389 acc3 += x2 * c0;
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391
emilmont 1:fdd22bb7aa52 392 } while(--k);
emilmont 1:fdd22bb7aa52 393
emilmont 1:fdd22bb7aa52 394 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 395 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 396 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 397
emilmont 1:fdd22bb7aa52 398 while(k > 0u)
emilmont 1:fdd22bb7aa52 399 {
emilmont 1:fdd22bb7aa52 400 /* Read y[srcBLen - 5] sample */
emilmont 1:fdd22bb7aa52 401 c0 = *(py--);
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 /* Read x[7] sample */
emilmont 1:fdd22bb7aa52 404 x3 = *(px++);
emilmont 1:fdd22bb7aa52 405
emilmont 1:fdd22bb7aa52 406 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 407 /* acc0 += x[4] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 408 acc0 += x0 * c0;
emilmont 1:fdd22bb7aa52 409 /* acc1 += x[5] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 410 acc1 += x1 * c0;
emilmont 1:fdd22bb7aa52 411 /* acc2 += x[6] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 412 acc2 += x2 * c0;
emilmont 1:fdd22bb7aa52 413 /* acc3 += x[7] * y[srcBLen - 5] */
emilmont 1:fdd22bb7aa52 414 acc3 += x3 * c0;
emilmont 1:fdd22bb7aa52 415
emilmont 1:fdd22bb7aa52 416 /* Reuse the present samples for the next MAC */
emilmont 1:fdd22bb7aa52 417 x0 = x1;
emilmont 1:fdd22bb7aa52 418 x1 = x2;
emilmont 1:fdd22bb7aa52 419 x2 = x3;
emilmont 1:fdd22bb7aa52 420
emilmont 1:fdd22bb7aa52 421 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 422 k--;
emilmont 1:fdd22bb7aa52 423 }
emilmont 1:fdd22bb7aa52 424
emilmont 1:fdd22bb7aa52 425 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 426 *pOut++ = acc0;
emilmont 1:fdd22bb7aa52 427 *pOut++ = acc1;
emilmont 1:fdd22bb7aa52 428 *pOut++ = acc2;
emilmont 1:fdd22bb7aa52 429 *pOut++ = acc3;
emilmont 1:fdd22bb7aa52 430
emilmont 1:fdd22bb7aa52 431 /* Increment the pointer pIn1 index, count by 1 */
emilmont 1:fdd22bb7aa52 432 count += 4u;
emilmont 1:fdd22bb7aa52 433
emilmont 1:fdd22bb7aa52 434 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 435 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 436 py = pSrc2;
emilmont 1:fdd22bb7aa52 437
emilmont 1:fdd22bb7aa52 438 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 439 blkCnt--;
emilmont 1:fdd22bb7aa52 440 }
emilmont 1:fdd22bb7aa52 441
emilmont 1:fdd22bb7aa52 442 /* If the blockSize2 is not a multiple of 4, compute any remaining output samples here.
emilmont 1:fdd22bb7aa52 443 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 444 blkCnt = (uint32_t) blockSize2 % 0x4u;
emilmont 1:fdd22bb7aa52 445
emilmont 1:fdd22bb7aa52 446 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 447 {
emilmont 1:fdd22bb7aa52 448 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 449 sum = 0.0f;
emilmont 1:fdd22bb7aa52 450
emilmont 1:fdd22bb7aa52 451 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 452 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 453
emilmont 1:fdd22bb7aa52 454 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 455 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 456 while(k > 0u)
emilmont 1:fdd22bb7aa52 457 {
emilmont 1:fdd22bb7aa52 458 /* Perform the multiply-accumulates */
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 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 463
emilmont 1:fdd22bb7aa52 464 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 465 k--;
emilmont 1:fdd22bb7aa52 466 }
emilmont 1:fdd22bb7aa52 467
emilmont 1:fdd22bb7aa52 468 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 469 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 470 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 471
emilmont 1:fdd22bb7aa52 472 while(k > 0u)
emilmont 1:fdd22bb7aa52 473 {
emilmont 1:fdd22bb7aa52 474 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 475 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 476
emilmont 1:fdd22bb7aa52 477 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 478 k--;
emilmont 1:fdd22bb7aa52 479 }
emilmont 1:fdd22bb7aa52 480
emilmont 1:fdd22bb7aa52 481 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 482 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 483
emilmont 1:fdd22bb7aa52 484 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 485 count++;
emilmont 1:fdd22bb7aa52 486
emilmont 1:fdd22bb7aa52 487 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 488 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 489 py = pSrc2;
emilmont 1:fdd22bb7aa52 490
emilmont 1:fdd22bb7aa52 491 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 492 blkCnt--;
emilmont 1:fdd22bb7aa52 493 }
emilmont 1:fdd22bb7aa52 494 }
emilmont 1:fdd22bb7aa52 495 else
emilmont 1:fdd22bb7aa52 496 {
emilmont 1:fdd22bb7aa52 497 /* If the srcBLen is not a multiple of 4,
emilmont 1:fdd22bb7aa52 498 * the blockSize2 loop cannot be unrolled by 4 */
emilmont 1:fdd22bb7aa52 499 blkCnt = (uint32_t) blockSize2;
emilmont 1:fdd22bb7aa52 500
emilmont 1:fdd22bb7aa52 501 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 502 {
emilmont 1:fdd22bb7aa52 503 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 504 sum = 0.0f;
emilmont 1:fdd22bb7aa52 505
emilmont 1:fdd22bb7aa52 506 /* srcBLen number of MACS should be performed */
emilmont 1:fdd22bb7aa52 507 k = srcBLen;
emilmont 1:fdd22bb7aa52 508
emilmont 1:fdd22bb7aa52 509 while(k > 0u)
emilmont 1:fdd22bb7aa52 510 {
emilmont 1:fdd22bb7aa52 511 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 512 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 513
emilmont 1:fdd22bb7aa52 514 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 515 k--;
emilmont 1:fdd22bb7aa52 516 }
emilmont 1:fdd22bb7aa52 517
emilmont 1:fdd22bb7aa52 518 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 519 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 520
emilmont 1:fdd22bb7aa52 521 /* Increment the MAC count */
emilmont 1:fdd22bb7aa52 522 count++;
emilmont 1:fdd22bb7aa52 523
emilmont 1:fdd22bb7aa52 524 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 525 px = pIn1 + count;
emilmont 1:fdd22bb7aa52 526 py = pSrc2;
emilmont 1:fdd22bb7aa52 527
emilmont 1:fdd22bb7aa52 528 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 529 blkCnt--;
emilmont 1:fdd22bb7aa52 530 }
emilmont 1:fdd22bb7aa52 531 }
emilmont 1:fdd22bb7aa52 532
emilmont 1:fdd22bb7aa52 533
emilmont 1:fdd22bb7aa52 534 /* --------------------------
emilmont 1:fdd22bb7aa52 535 * Initializations of stage3
emilmont 1:fdd22bb7aa52 536 * -------------------------*/
emilmont 1:fdd22bb7aa52 537
emilmont 1:fdd22bb7aa52 538 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emilmont 1:fdd22bb7aa52 539 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emilmont 1:fdd22bb7aa52 540 * ....
emilmont 1:fdd22bb7aa52 541 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emilmont 1:fdd22bb7aa52 542 * sum += x[srcALen-1] * y[srcBLen-1]
emilmont 1:fdd22bb7aa52 543 */
emilmont 1:fdd22bb7aa52 544
emilmont 1:fdd22bb7aa52 545 /* In this stage the MAC operations are decreased by 1 for every iteration.
emilmont 1:fdd22bb7aa52 546 The count variable holds the number of MAC operations performed */
emilmont 1:fdd22bb7aa52 547 count = srcBLen - 1u;
emilmont 1:fdd22bb7aa52 548
emilmont 1:fdd22bb7aa52 549 /* Working pointer of inputA */
emilmont 1:fdd22bb7aa52 550 pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 551 px = pSrc1;
emilmont 1:fdd22bb7aa52 552
emilmont 1:fdd22bb7aa52 553 /* Working pointer of inputB */
emilmont 1:fdd22bb7aa52 554 pSrc2 = pIn2 + (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 555 py = pSrc2;
emilmont 1:fdd22bb7aa52 556
emilmont 1:fdd22bb7aa52 557 while(blockSize3 > 0)
emilmont 1:fdd22bb7aa52 558 {
emilmont 1:fdd22bb7aa52 559 /* Accumulator is made zero for every iteration */
emilmont 1:fdd22bb7aa52 560 sum = 0.0f;
emilmont 1:fdd22bb7aa52 561
emilmont 1:fdd22bb7aa52 562 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emilmont 1:fdd22bb7aa52 563 k = count >> 2u;
emilmont 1:fdd22bb7aa52 564
emilmont 1:fdd22bb7aa52 565 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emilmont 1:fdd22bb7aa52 566 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 567 while(k > 0u)
emilmont 1:fdd22bb7aa52 568 {
emilmont 1:fdd22bb7aa52 569 /* sum += x[srcALen - srcBLen + 1] * y[srcBLen - 1] */
emilmont 1:fdd22bb7aa52 570 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 571
emilmont 1:fdd22bb7aa52 572 /* sum += x[srcALen - srcBLen + 2] * y[srcBLen - 2] */
emilmont 1:fdd22bb7aa52 573 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 574
emilmont 1:fdd22bb7aa52 575 /* sum += x[srcALen - srcBLen + 3] * y[srcBLen - 3] */
emilmont 1:fdd22bb7aa52 576 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 577
emilmont 1:fdd22bb7aa52 578 /* sum += x[srcALen - srcBLen + 4] * y[srcBLen - 4] */
emilmont 1:fdd22bb7aa52 579 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 580
emilmont 1:fdd22bb7aa52 581 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 582 k--;
emilmont 1:fdd22bb7aa52 583 }
emilmont 1:fdd22bb7aa52 584
emilmont 1:fdd22bb7aa52 585 /* If the count is not a multiple of 4, compute any remaining MACs here.
emilmont 1:fdd22bb7aa52 586 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 587 k = count % 0x4u;
emilmont 1:fdd22bb7aa52 588
emilmont 1:fdd22bb7aa52 589 while(k > 0u)
emilmont 1:fdd22bb7aa52 590 {
emilmont 1:fdd22bb7aa52 591 /* Perform the multiply-accumulates */
emilmont 1:fdd22bb7aa52 592 /* sum += x[srcALen-1] * y[srcBLen-1] */
emilmont 1:fdd22bb7aa52 593 sum += *px++ * *py--;
emilmont 1:fdd22bb7aa52 594
emilmont 1:fdd22bb7aa52 595 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 596 k--;
emilmont 1:fdd22bb7aa52 597 }
emilmont 1:fdd22bb7aa52 598
emilmont 1:fdd22bb7aa52 599 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 600 *pOut++ = sum;
emilmont 1:fdd22bb7aa52 601
emilmont 1:fdd22bb7aa52 602 /* Update the inputA and inputB pointers for next MAC calculation */
emilmont 1:fdd22bb7aa52 603 px = ++pSrc1;
emilmont 1:fdd22bb7aa52 604 py = pSrc2;
emilmont 1:fdd22bb7aa52 605
emilmont 1:fdd22bb7aa52 606 /* Decrement the MAC count */
emilmont 1:fdd22bb7aa52 607 count--;
emilmont 1:fdd22bb7aa52 608
emilmont 1:fdd22bb7aa52 609 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 610 blockSize3--;
emilmont 1:fdd22bb7aa52 611
emilmont 1:fdd22bb7aa52 612 }
emilmont 1:fdd22bb7aa52 613
emilmont 1:fdd22bb7aa52 614 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 615 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 616 }
emilmont 1:fdd22bb7aa52 617
emilmont 1:fdd22bb7aa52 618 /* Return to application */
emilmont 1:fdd22bb7aa52 619 return (status);
emilmont 1:fdd22bb7aa52 620
emilmont 1:fdd22bb7aa52 621 #else
emilmont 1:fdd22bb7aa52 622
emilmont 1:fdd22bb7aa52 623 /* Run the below code for Cortex-M0 */
emilmont 1:fdd22bb7aa52 624
emilmont 1:fdd22bb7aa52 625 float32_t *pIn1 = pSrcA; /* inputA pointer */
emilmont 1:fdd22bb7aa52 626 float32_t *pIn2 = pSrcB; /* inputB pointer */
emilmont 1:fdd22bb7aa52 627 float32_t sum; /* Accumulator */
emilmont 1:fdd22bb7aa52 628 uint32_t i, j; /* loop counters */
emilmont 1:fdd22bb7aa52 629 arm_status status; /* status of Partial convolution */
emilmont 1:fdd22bb7aa52 630
emilmont 1:fdd22bb7aa52 631 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 632 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 633 {
emilmont 1:fdd22bb7aa52 634 /* Set status as ARM_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 635 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 636 }
emilmont 1:fdd22bb7aa52 637 else
emilmont 1:fdd22bb7aa52 638 {
emilmont 1:fdd22bb7aa52 639 /* Loop to calculate convolution for output length number of values */
emilmont 1:fdd22bb7aa52 640 for (i = firstIndex; i <= (firstIndex + numPoints - 1); i++)
emilmont 1:fdd22bb7aa52 641 {
emilmont 1:fdd22bb7aa52 642 /* Initialize sum with zero to carry on MAC operations */
emilmont 1:fdd22bb7aa52 643 sum = 0.0f;
emilmont 1:fdd22bb7aa52 644
emilmont 1:fdd22bb7aa52 645 /* Loop to perform MAC operations according to convolution equation */
emilmont 1:fdd22bb7aa52 646 for (j = 0u; j <= i; j++)
emilmont 1:fdd22bb7aa52 647 {
emilmont 1:fdd22bb7aa52 648 /* Check the array limitations for inputs */
emilmont 1:fdd22bb7aa52 649 if((((i - j) < srcBLen) && (j < srcALen)))
emilmont 1:fdd22bb7aa52 650 {
emilmont 1:fdd22bb7aa52 651 /* z[i] += x[i-j] * y[j] */
emilmont 1:fdd22bb7aa52 652 sum += pIn1[j] * pIn2[i - j];
emilmont 1:fdd22bb7aa52 653 }
emilmont 1:fdd22bb7aa52 654 }
emilmont 1:fdd22bb7aa52 655 /* Store the output in the destination buffer */
emilmont 1:fdd22bb7aa52 656 pDst[i] = sum;
emilmont 1:fdd22bb7aa52 657 }
emilmont 1:fdd22bb7aa52 658 /* set status as ARM_SUCCESS as there are no argument errors */
emilmont 1:fdd22bb7aa52 659 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 660 }
emilmont 1:fdd22bb7aa52 661 return (status);
emilmont 1:fdd22bb7aa52 662
mbed_official 3:7a284390b0ce 663 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emilmont 1:fdd22bb7aa52 664
emilmont 1:fdd22bb7aa52 665 }
emilmont 1:fdd22bb7aa52 666
emilmont 1:fdd22bb7aa52 667 /**
emilmont 1:fdd22bb7aa52 668 * @} end of PartialConv group
emilmont 1:fdd22bb7aa52 669 */