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_conv_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_conv_q31.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Convolution 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 Conv
emh203 0:3d9c67d97d6f 49 * @{
emh203 0:3d9c67d97d6f 50 */
emh203 0:3d9c67d97d6f 51
emh203 0:3d9c67d97d6f 52 /**
emh203 0:3d9c67d97d6f 53 * @brief Convolution 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 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 the inputs by log2(min(srcALen, srcBLen)) (log2 is read as log to the base 2) times to avoid overflows,
emh203 0:3d9c67d97d6f 71 * as maximum of min(srcALen, srcBLen) number of additions are 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_conv_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_conv_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
emh203 0:3d9c67d97d6f 87 #ifndef ARM_MATH_CM0_FAMILY
emh203 0:3d9c67d97d6f 88
emh203 0:3d9c67d97d6f 89 /* Run the below code for Cortex-M4 and Cortex-M3 */
emh203 0:3d9c67d97d6f 90
emh203 0:3d9c67d97d6f 91 q31_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 92 q31_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 93 q31_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 94 q31_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 95 q31_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 96 q31_t *pSrc1, *pSrc2; /* Intermediate pointers */
emh203 0:3d9c67d97d6f 97 q63_t sum; /* Accumulator */
emh203 0:3d9c67d97d6f 98 q63_t acc0, acc1, acc2; /* Accumulator */
emh203 0:3d9c67d97d6f 99 q31_t x0, x1, x2, c0; /* Temporary variables to hold state and coefficient values */
emh203 0:3d9c67d97d6f 100 uint32_t j, k, count, blkCnt, blockSize1, blockSize2, blockSize3; /* loop counter */
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 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 106 {
emh203 0:3d9c67d97d6f 107 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 108 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 109
emh203 0:3d9c67d97d6f 110 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 111 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 112 }
emh203 0:3d9c67d97d6f 113 else
emh203 0:3d9c67d97d6f 114 {
emh203 0:3d9c67d97d6f 115 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 116 pIn1 = (q31_t *) pSrcB;
emh203 0:3d9c67d97d6f 117
emh203 0:3d9c67d97d6f 118 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 119 pIn2 = (q31_t *) pSrcA;
emh203 0:3d9c67d97d6f 120
emh203 0:3d9c67d97d6f 121 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 122 j = srcBLen;
emh203 0:3d9c67d97d6f 123 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 124 srcALen = j;
emh203 0:3d9c67d97d6f 125 }
emh203 0:3d9c67d97d6f 126
emh203 0:3d9c67d97d6f 127 /* 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] */
emh203 0:3d9c67d97d6f 128 /* The function is internally
emh203 0:3d9c67d97d6f 129 * divided into three stages according to the number of multiplications that has to be
emh203 0:3d9c67d97d6f 130 * taken place between inputA samples and inputB samples. In the first stage of the
emh203 0:3d9c67d97d6f 131 * algorithm, the multiplications increase by one for every iteration.
emh203 0:3d9c67d97d6f 132 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emh203 0:3d9c67d97d6f 133 * In the third stage of the algorithm, the multiplications decrease by one
emh203 0:3d9c67d97d6f 134 * for every iteration. */
emh203 0:3d9c67d97d6f 135
emh203 0:3d9c67d97d6f 136 /* The algorithm is implemented in three stages.
emh203 0:3d9c67d97d6f 137 The loop counters of each stage is initiated here. */
emh203 0:3d9c67d97d6f 138 blockSize1 = srcBLen - 1u;
emh203 0:3d9c67d97d6f 139 blockSize2 = srcALen - (srcBLen - 1u);
emh203 0:3d9c67d97d6f 140 blockSize3 = blockSize1;
emh203 0:3d9c67d97d6f 141
emh203 0:3d9c67d97d6f 142 /* --------------------------
emh203 0:3d9c67d97d6f 143 * Initializations of stage1
emh203 0:3d9c67d97d6f 144 * -------------------------*/
emh203 0:3d9c67d97d6f 145
emh203 0:3d9c67d97d6f 146 /* sum = x[0] * y[0]
emh203 0:3d9c67d97d6f 147 * sum = x[0] * y[1] + x[1] * y[0]
emh203 0:3d9c67d97d6f 148 * ....
emh203 0:3d9c67d97d6f 149 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emh203 0:3d9c67d97d6f 150 */
emh203 0:3d9c67d97d6f 151
emh203 0:3d9c67d97d6f 152 /* In this stage the MAC operations are increased by 1 for every iteration.
emh203 0:3d9c67d97d6f 153 The count variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 154 count = 1u;
emh203 0:3d9c67d97d6f 155
emh203 0:3d9c67d97d6f 156 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 157 px = pIn1;
emh203 0:3d9c67d97d6f 158
emh203 0:3d9c67d97d6f 159 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 160 py = pIn2;
emh203 0:3d9c67d97d6f 161
emh203 0:3d9c67d97d6f 162
emh203 0:3d9c67d97d6f 163 /* ------------------------
emh203 0:3d9c67d97d6f 164 * Stage1 process
emh203 0:3d9c67d97d6f 165 * ----------------------*/
emh203 0:3d9c67d97d6f 166
emh203 0:3d9c67d97d6f 167 /* The first stage starts here */
emh203 0:3d9c67d97d6f 168 while(blockSize1 > 0u)
emh203 0:3d9c67d97d6f 169 {
emh203 0:3d9c67d97d6f 170 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 171 sum = 0;
emh203 0:3d9c67d97d6f 172
emh203 0:3d9c67d97d6f 173 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 174 k = count >> 2u;
emh203 0:3d9c67d97d6f 175
emh203 0:3d9c67d97d6f 176 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 177 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 178 while(k > 0u)
emh203 0:3d9c67d97d6f 179 {
emh203 0:3d9c67d97d6f 180 /* x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 181 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 182 /* x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 183 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 184 /* x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 185 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 186 /* x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 187 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 188
emh203 0:3d9c67d97d6f 189 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 190 k--;
emh203 0:3d9c67d97d6f 191 }
emh203 0:3d9c67d97d6f 192
emh203 0:3d9c67d97d6f 193 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 194 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 195 k = count % 0x4u;
emh203 0:3d9c67d97d6f 196
emh203 0:3d9c67d97d6f 197 while(k > 0u)
emh203 0:3d9c67d97d6f 198 {
emh203 0:3d9c67d97d6f 199 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 200 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 201
emh203 0:3d9c67d97d6f 202 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 203 k--;
emh203 0:3d9c67d97d6f 204 }
emh203 0:3d9c67d97d6f 205
emh203 0:3d9c67d97d6f 206 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 207 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 208
emh203 0:3d9c67d97d6f 209 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 210 py = pIn2 + count;
emh203 0:3d9c67d97d6f 211 px = pIn1;
emh203 0:3d9c67d97d6f 212
emh203 0:3d9c67d97d6f 213 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 214 count++;
emh203 0:3d9c67d97d6f 215
emh203 0:3d9c67d97d6f 216 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 217 blockSize1--;
emh203 0:3d9c67d97d6f 218 }
emh203 0:3d9c67d97d6f 219
emh203 0:3d9c67d97d6f 220 /* --------------------------
emh203 0:3d9c67d97d6f 221 * Initializations of stage2
emh203 0:3d9c67d97d6f 222 * ------------------------*/
emh203 0:3d9c67d97d6f 223
emh203 0:3d9c67d97d6f 224 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emh203 0:3d9c67d97d6f 225 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emh203 0:3d9c67d97d6f 226 * ....
emh203 0:3d9c67d97d6f 227 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emh203 0:3d9c67d97d6f 228 */
emh203 0:3d9c67d97d6f 229
emh203 0:3d9c67d97d6f 230 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 231 px = pIn1;
emh203 0:3d9c67d97d6f 232
emh203 0:3d9c67d97d6f 233 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 234 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 235 py = pSrc2;
emh203 0:3d9c67d97d6f 236
emh203 0:3d9c67d97d6f 237 /* count is index by which the pointer pIn1 to be incremented */
emh203 0:3d9c67d97d6f 238 count = 0u;
emh203 0:3d9c67d97d6f 239
emh203 0:3d9c67d97d6f 240 /* -------------------
emh203 0:3d9c67d97d6f 241 * Stage2 process
emh203 0:3d9c67d97d6f 242 * ------------------*/
emh203 0:3d9c67d97d6f 243
emh203 0:3d9c67d97d6f 244 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emh203 0:3d9c67d97d6f 245 * So, to loop unroll over blockSize2,
emh203 0:3d9c67d97d6f 246 * srcBLen should be greater than or equal to 4 */
emh203 0:3d9c67d97d6f 247 if(srcBLen >= 4u)
emh203 0:3d9c67d97d6f 248 {
emh203 0:3d9c67d97d6f 249 /* Loop unroll by 3 */
emh203 0:3d9c67d97d6f 250 blkCnt = blockSize2 / 3;
emh203 0:3d9c67d97d6f 251
emh203 0:3d9c67d97d6f 252 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 253 {
emh203 0:3d9c67d97d6f 254 /* Set all accumulators to zero */
emh203 0:3d9c67d97d6f 255 acc0 = 0;
emh203 0:3d9c67d97d6f 256 acc1 = 0;
emh203 0:3d9c67d97d6f 257 acc2 = 0;
emh203 0:3d9c67d97d6f 258
emh203 0:3d9c67d97d6f 259 /* read x[0], x[1], x[2] samples */
emh203 0:3d9c67d97d6f 260 x0 = *(px++);
emh203 0:3d9c67d97d6f 261 x1 = *(px++);
emh203 0:3d9c67d97d6f 262
emh203 0:3d9c67d97d6f 263 /* Apply loop unrolling and compute 3 MACs simultaneously. */
emh203 0:3d9c67d97d6f 264 k = srcBLen / 3;
emh203 0:3d9c67d97d6f 265
emh203 0:3d9c67d97d6f 266 /* First part of the processing with loop unrolling. Compute 3 MACs at a time.
emh203 0:3d9c67d97d6f 267 ** a second loop below computes MACs for the remaining 1 to 2 samples. */
emh203 0:3d9c67d97d6f 268 do
emh203 0:3d9c67d97d6f 269 {
emh203 0:3d9c67d97d6f 270 /* Read y[srcBLen - 1] sample */
emh203 0:3d9c67d97d6f 271 c0 = *(py);
emh203 0:3d9c67d97d6f 272
emh203 0:3d9c67d97d6f 273 /* Read x[3] sample */
emh203 0:3d9c67d97d6f 274 x2 = *(px);
emh203 0:3d9c67d97d6f 275
emh203 0:3d9c67d97d6f 276 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 277 /* acc0 += x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 278 acc0 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 279 /* acc1 += x[1] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 280 acc1 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 281 /* acc2 += x[2] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 282 acc2 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 283
emh203 0:3d9c67d97d6f 284 /* Read y[srcBLen - 2] sample */
emh203 0:3d9c67d97d6f 285 c0 = *(py - 1u);
emh203 0:3d9c67d97d6f 286
emh203 0:3d9c67d97d6f 287 /* Read x[4] sample */
emh203 0:3d9c67d97d6f 288 x0 = *(px + 1u);
emh203 0:3d9c67d97d6f 289
emh203 0:3d9c67d97d6f 290 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 291 /* acc0 += x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 292 acc0 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 293 /* acc1 += x[2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 294 acc1 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 295 /* acc2 += x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 296 acc2 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 297
emh203 0:3d9c67d97d6f 298 /* Read y[srcBLen - 3] sample */
emh203 0:3d9c67d97d6f 299 c0 = *(py - 2u);
emh203 0:3d9c67d97d6f 300
emh203 0:3d9c67d97d6f 301 /* Read x[5] sample */
emh203 0:3d9c67d97d6f 302 x1 = *(px + 2u);
emh203 0:3d9c67d97d6f 303
emh203 0:3d9c67d97d6f 304 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 305 /* acc0 += x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 306 acc0 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 307 /* acc1 += x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 308 acc1 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 309 /* acc2 += x[4] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 310 acc2 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 311
emh203 0:3d9c67d97d6f 312 /* update scratch pointers */
emh203 0:3d9c67d97d6f 313 px += 3u;
emh203 0:3d9c67d97d6f 314 py -= 3u;
emh203 0:3d9c67d97d6f 315
emh203 0:3d9c67d97d6f 316 } while(--k);
emh203 0:3d9c67d97d6f 317
emh203 0:3d9c67d97d6f 318 /* If the srcBLen is not a multiple of 3, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 319 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 320 k = srcBLen - (3 * (srcBLen / 3));
emh203 0:3d9c67d97d6f 321
emh203 0:3d9c67d97d6f 322 while(k > 0u)
emh203 0:3d9c67d97d6f 323 {
emh203 0:3d9c67d97d6f 324 /* Read y[srcBLen - 5] sample */
emh203 0:3d9c67d97d6f 325 c0 = *(py--);
emh203 0:3d9c67d97d6f 326
emh203 0:3d9c67d97d6f 327 /* Read x[7] sample */
emh203 0:3d9c67d97d6f 328 x2 = *(px++);
emh203 0:3d9c67d97d6f 329
emh203 0:3d9c67d97d6f 330 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 331 /* acc0 += x[4] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 332 acc0 += ((q63_t) x0 * c0);
emh203 0:3d9c67d97d6f 333 /* acc1 += x[5] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 334 acc1 += ((q63_t) x1 * c0);
emh203 0:3d9c67d97d6f 335 /* acc2 += x[6] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 336 acc2 += ((q63_t) x2 * c0);
emh203 0:3d9c67d97d6f 337
emh203 0:3d9c67d97d6f 338 /* Reuse the present samples for the next MAC */
emh203 0:3d9c67d97d6f 339 x0 = x1;
emh203 0:3d9c67d97d6f 340 x1 = x2;
emh203 0:3d9c67d97d6f 341
emh203 0:3d9c67d97d6f 342 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 343 k--;
emh203 0:3d9c67d97d6f 344 }
emh203 0:3d9c67d97d6f 345
emh203 0:3d9c67d97d6f 346 /* Store the results in the accumulators in the destination buffer. */
emh203 0:3d9c67d97d6f 347 *pOut++ = (q31_t) (acc0 >> 31);
emh203 0:3d9c67d97d6f 348 *pOut++ = (q31_t) (acc1 >> 31);
emh203 0:3d9c67d97d6f 349 *pOut++ = (q31_t) (acc2 >> 31);
emh203 0:3d9c67d97d6f 350
emh203 0:3d9c67d97d6f 351 /* Increment the pointer pIn1 index, count by 3 */
emh203 0:3d9c67d97d6f 352 count += 3u;
emh203 0:3d9c67d97d6f 353
emh203 0:3d9c67d97d6f 354 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 355 px = pIn1 + count;
emh203 0:3d9c67d97d6f 356 py = pSrc2;
emh203 0:3d9c67d97d6f 357
emh203 0:3d9c67d97d6f 358 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 359 blkCnt--;
emh203 0:3d9c67d97d6f 360 }
emh203 0:3d9c67d97d6f 361
emh203 0:3d9c67d97d6f 362 /* If the blockSize2 is not a multiple of 3, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 363 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 364 blkCnt = blockSize2 - 3 * (blockSize2 / 3);
emh203 0:3d9c67d97d6f 365
emh203 0:3d9c67d97d6f 366 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 367 {
emh203 0:3d9c67d97d6f 368 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 369 sum = 0;
emh203 0:3d9c67d97d6f 370
emh203 0:3d9c67d97d6f 371 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 372 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 373
emh203 0:3d9c67d97d6f 374 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 375 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 376 while(k > 0u)
emh203 0:3d9c67d97d6f 377 {
emh203 0:3d9c67d97d6f 378 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 379 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 380 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 381 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 382 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 383
emh203 0:3d9c67d97d6f 384 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 385 k--;
emh203 0:3d9c67d97d6f 386 }
emh203 0:3d9c67d97d6f 387
emh203 0:3d9c67d97d6f 388 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 389 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 390 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 391
emh203 0:3d9c67d97d6f 392 while(k > 0u)
emh203 0:3d9c67d97d6f 393 {
emh203 0:3d9c67d97d6f 394 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 395 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 396
emh203 0:3d9c67d97d6f 397 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 398 k--;
emh203 0:3d9c67d97d6f 399 }
emh203 0:3d9c67d97d6f 400
emh203 0:3d9c67d97d6f 401 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 402 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 403
emh203 0:3d9c67d97d6f 404 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 405 count++;
emh203 0:3d9c67d97d6f 406
emh203 0:3d9c67d97d6f 407 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 408 px = pIn1 + count;
emh203 0:3d9c67d97d6f 409 py = pSrc2;
emh203 0:3d9c67d97d6f 410
emh203 0:3d9c67d97d6f 411 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 412 blkCnt--;
emh203 0:3d9c67d97d6f 413 }
emh203 0:3d9c67d97d6f 414 }
emh203 0:3d9c67d97d6f 415 else
emh203 0:3d9c67d97d6f 416 {
emh203 0:3d9c67d97d6f 417 /* If the srcBLen is not a multiple of 4,
emh203 0:3d9c67d97d6f 418 * the blockSize2 loop cannot be unrolled by 4 */
emh203 0:3d9c67d97d6f 419 blkCnt = blockSize2;
emh203 0:3d9c67d97d6f 420
emh203 0:3d9c67d97d6f 421 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 422 {
emh203 0:3d9c67d97d6f 423 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 424 sum = 0;
emh203 0:3d9c67d97d6f 425
emh203 0:3d9c67d97d6f 426 /* srcBLen number of MACS should be performed */
emh203 0:3d9c67d97d6f 427 k = srcBLen;
emh203 0:3d9c67d97d6f 428
emh203 0:3d9c67d97d6f 429 while(k > 0u)
emh203 0:3d9c67d97d6f 430 {
emh203 0:3d9c67d97d6f 431 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 432 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 433
emh203 0:3d9c67d97d6f 434 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 435 k--;
emh203 0:3d9c67d97d6f 436 }
emh203 0:3d9c67d97d6f 437
emh203 0:3d9c67d97d6f 438 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 439 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 440
emh203 0:3d9c67d97d6f 441 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 442 count++;
emh203 0:3d9c67d97d6f 443
emh203 0:3d9c67d97d6f 444 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 445 px = pIn1 + count;
emh203 0:3d9c67d97d6f 446 py = pSrc2;
emh203 0:3d9c67d97d6f 447
emh203 0:3d9c67d97d6f 448 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 449 blkCnt--;
emh203 0:3d9c67d97d6f 450 }
emh203 0:3d9c67d97d6f 451 }
emh203 0:3d9c67d97d6f 452
emh203 0:3d9c67d97d6f 453
emh203 0:3d9c67d97d6f 454 /* --------------------------
emh203 0:3d9c67d97d6f 455 * Initializations of stage3
emh203 0:3d9c67d97d6f 456 * -------------------------*/
emh203 0:3d9c67d97d6f 457
emh203 0:3d9c67d97d6f 458 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emh203 0:3d9c67d97d6f 459 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emh203 0:3d9c67d97d6f 460 * ....
emh203 0:3d9c67d97d6f 461 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emh203 0:3d9c67d97d6f 462 * sum += x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 463 */
emh203 0:3d9c67d97d6f 464
emh203 0:3d9c67d97d6f 465 /* In this stage the MAC operations are decreased by 1 for every iteration.
emh203 0:3d9c67d97d6f 466 The blockSize3 variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 467
emh203 0:3d9c67d97d6f 468 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 469 pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
emh203 0:3d9c67d97d6f 470 px = pSrc1;
emh203 0:3d9c67d97d6f 471
emh203 0:3d9c67d97d6f 472 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 473 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 474 py = pSrc2;
emh203 0:3d9c67d97d6f 475
emh203 0:3d9c67d97d6f 476 /* -------------------
emh203 0:3d9c67d97d6f 477 * Stage3 process
emh203 0:3d9c67d97d6f 478 * ------------------*/
emh203 0:3d9c67d97d6f 479
emh203 0:3d9c67d97d6f 480 while(blockSize3 > 0u)
emh203 0:3d9c67d97d6f 481 {
emh203 0:3d9c67d97d6f 482 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 483 sum = 0;
emh203 0:3d9c67d97d6f 484
emh203 0:3d9c67d97d6f 485 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 486 k = blockSize3 >> 2u;
emh203 0:3d9c67d97d6f 487
emh203 0:3d9c67d97d6f 488 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 489 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 490 while(k > 0u)
emh203 0:3d9c67d97d6f 491 {
emh203 0:3d9c67d97d6f 492 /* sum += x[srcALen - srcBLen + 1] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 493 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 494 /* sum += x[srcALen - srcBLen + 2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 495 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 496 /* sum += x[srcALen - srcBLen + 3] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 497 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 498 /* sum += x[srcALen - srcBLen + 4] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 499 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 500
emh203 0:3d9c67d97d6f 501 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 502 k--;
emh203 0:3d9c67d97d6f 503 }
emh203 0:3d9c67d97d6f 504
emh203 0:3d9c67d97d6f 505 /* If the blockSize3 is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 506 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 507 k = blockSize3 % 0x4u;
emh203 0:3d9c67d97d6f 508
emh203 0:3d9c67d97d6f 509 while(k > 0u)
emh203 0:3d9c67d97d6f 510 {
emh203 0:3d9c67d97d6f 511 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 512 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 513
emh203 0:3d9c67d97d6f 514 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 515 k--;
emh203 0:3d9c67d97d6f 516 }
emh203 0:3d9c67d97d6f 517
emh203 0:3d9c67d97d6f 518 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 519 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 520
emh203 0:3d9c67d97d6f 521 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 522 px = ++pSrc1;
emh203 0:3d9c67d97d6f 523 py = pSrc2;
emh203 0:3d9c67d97d6f 524
emh203 0:3d9c67d97d6f 525 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 526 blockSize3--;
emh203 0:3d9c67d97d6f 527 }
emh203 0:3d9c67d97d6f 528
emh203 0:3d9c67d97d6f 529 #else
emh203 0:3d9c67d97d6f 530
emh203 0:3d9c67d97d6f 531 /* Run the below code for Cortex-M0 */
emh203 0:3d9c67d97d6f 532
emh203 0:3d9c67d97d6f 533 q31_t *pIn1 = pSrcA; /* input pointer */
emh203 0:3d9c67d97d6f 534 q31_t *pIn2 = pSrcB; /* coefficient pointer */
emh203 0:3d9c67d97d6f 535 q63_t sum; /* Accumulator */
emh203 0:3d9c67d97d6f 536 uint32_t i, j; /* loop counter */
emh203 0:3d9c67d97d6f 537
emh203 0:3d9c67d97d6f 538 /* Loop to calculate output of convolution for output length number of times */
emh203 0:3d9c67d97d6f 539 for (i = 0; i < (srcALen + srcBLen - 1); i++)
emh203 0:3d9c67d97d6f 540 {
emh203 0:3d9c67d97d6f 541 /* Initialize sum with zero to carry on MAC operations */
emh203 0:3d9c67d97d6f 542 sum = 0;
emh203 0:3d9c67d97d6f 543
emh203 0:3d9c67d97d6f 544 /* Loop to perform MAC operations according to convolution equation */
emh203 0:3d9c67d97d6f 545 for (j = 0; j <= i; j++)
emh203 0:3d9c67d97d6f 546 {
emh203 0:3d9c67d97d6f 547 /* Check the array limitations */
emh203 0:3d9c67d97d6f 548 if(((i - j) < srcBLen) && (j < srcALen))
emh203 0:3d9c67d97d6f 549 {
emh203 0:3d9c67d97d6f 550 /* z[i] += x[i-j] * y[j] */
emh203 0:3d9c67d97d6f 551 sum += ((q63_t) pIn1[j] * (pIn2[i - j]));
emh203 0:3d9c67d97d6f 552 }
emh203 0:3d9c67d97d6f 553 }
emh203 0:3d9c67d97d6f 554
emh203 0:3d9c67d97d6f 555 /* Store the output in the destination buffer */
emh203 0:3d9c67d97d6f 556 pDst[i] = (q31_t) (sum >> 31u);
emh203 0:3d9c67d97d6f 557 }
emh203 0:3d9c67d97d6f 558
emh203 0:3d9c67d97d6f 559 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emh203 0:3d9c67d97d6f 560
emh203 0:3d9c67d97d6f 561 }
emh203 0:3d9c67d97d6f 562
emh203 0:3d9c67d97d6f 563 /**
emh203 0:3d9c67d97d6f 564 * @} end of Conv group
emh203 0:3d9c67d97d6f 565 */