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_q7.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_q7.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Convolution of Q7 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 Q7 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 a 32-bit internal accumulator.
emh203 0:3d9c67d97d6f 66 * Both the inputs are represented in 1.7 format and multiplications yield a 2.14 result.
emh203 0:3d9c67d97d6f 67 * The 2.14 intermediate results are accumulated in a 32-bit accumulator in 18.14 format.
emh203 0:3d9c67d97d6f 68 * This approach provides 17 guard bits and there is no risk of overflow as long as <code>max(srcALen, srcBLen)<131072</code>.
emh203 0:3d9c67d97d6f 69 * The 18.14 result is then truncated to 18.7 format by discarding the low 7 bits and then saturated to 1.7 format.
emh203 0:3d9c67d97d6f 70 *
emh203 0:3d9c67d97d6f 71 * \par
emh203 0:3d9c67d97d6f 72 * Refer the function <code>arm_conv_opt_q7()</code> for a faster implementation of this function.
emh203 0:3d9c67d97d6f 73 *
emh203 0:3d9c67d97d6f 74 */
emh203 0:3d9c67d97d6f 75
emh203 0:3d9c67d97d6f 76 void arm_conv_q7(
emh203 0:3d9c67d97d6f 77 q7_t * pSrcA,
emh203 0:3d9c67d97d6f 78 uint32_t srcALen,
emh203 0:3d9c67d97d6f 79 q7_t * pSrcB,
emh203 0:3d9c67d97d6f 80 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 81 q7_t * pDst)
emh203 0:3d9c67d97d6f 82 {
emh203 0:3d9c67d97d6f 83
emh203 0:3d9c67d97d6f 84
emh203 0:3d9c67d97d6f 85 #ifndef ARM_MATH_CM0_FAMILY
emh203 0:3d9c67d97d6f 86
emh203 0:3d9c67d97d6f 87 /* Run the below code for Cortex-M4 and Cortex-M3 */
emh203 0:3d9c67d97d6f 88
emh203 0:3d9c67d97d6f 89 q7_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 90 q7_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 91 q7_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 92 q7_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 93 q7_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 94 q7_t *pSrc1, *pSrc2; /* Intermediate pointers */
emh203 0:3d9c67d97d6f 95 q7_t x0, x1, x2, x3, c0, c1; /* Temporary variables to hold state and coefficient values */
emh203 0:3d9c67d97d6f 96 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulator */
emh203 0:3d9c67d97d6f 97 q31_t input1, input2; /* Temporary input variables */
emh203 0:3d9c67d97d6f 98 q15_t in1, in2; /* Temporary input variables */
emh203 0:3d9c67d97d6f 99 uint32_t j, k, count, blkCnt, blockSize1, blockSize2, blockSize3; /* loop counter */
emh203 0:3d9c67d97d6f 100
emh203 0:3d9c67d97d6f 101 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 102 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 103 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 104 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 105 {
emh203 0:3d9c67d97d6f 106 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 107 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 108
emh203 0:3d9c67d97d6f 109 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 110 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 111 }
emh203 0:3d9c67d97d6f 112 else
emh203 0:3d9c67d97d6f 113 {
emh203 0:3d9c67d97d6f 114 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 115 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 116
emh203 0:3d9c67d97d6f 117 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 118 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 119
emh203 0:3d9c67d97d6f 120 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 121 j = srcBLen;
emh203 0:3d9c67d97d6f 122 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 123 srcALen = j;
emh203 0:3d9c67d97d6f 124 }
emh203 0:3d9c67d97d6f 125
emh203 0:3d9c67d97d6f 126 /* 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 127 /* The function is internally
emh203 0:3d9c67d97d6f 128 * divided into three stages according to the number of multiplications that has to be
emh203 0:3d9c67d97d6f 129 * taken place between inputA samples and inputB samples. In the first stage of the
emh203 0:3d9c67d97d6f 130 * algorithm, the multiplications increase by one for every iteration.
emh203 0:3d9c67d97d6f 131 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emh203 0:3d9c67d97d6f 132 * In the third stage of the algorithm, the multiplications decrease by one
emh203 0:3d9c67d97d6f 133 * for every iteration. */
emh203 0:3d9c67d97d6f 134
emh203 0:3d9c67d97d6f 135 /* The algorithm is implemented in three stages.
emh203 0:3d9c67d97d6f 136 The loop counters of each stage is initiated here. */
emh203 0:3d9c67d97d6f 137 blockSize1 = srcBLen - 1u;
emh203 0:3d9c67d97d6f 138 blockSize2 = (srcALen - srcBLen) + 1u;
emh203 0:3d9c67d97d6f 139 blockSize3 = blockSize1;
emh203 0:3d9c67d97d6f 140
emh203 0:3d9c67d97d6f 141 /* --------------------------
emh203 0:3d9c67d97d6f 142 * Initializations of stage1
emh203 0:3d9c67d97d6f 143 * -------------------------*/
emh203 0:3d9c67d97d6f 144
emh203 0:3d9c67d97d6f 145 /* sum = x[0] * y[0]
emh203 0:3d9c67d97d6f 146 * sum = x[0] * y[1] + x[1] * y[0]
emh203 0:3d9c67d97d6f 147 * ....
emh203 0:3d9c67d97d6f 148 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emh203 0:3d9c67d97d6f 149 */
emh203 0:3d9c67d97d6f 150
emh203 0:3d9c67d97d6f 151 /* In this stage the MAC operations are increased by 1 for every iteration.
emh203 0:3d9c67d97d6f 152 The count variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 153 count = 1u;
emh203 0:3d9c67d97d6f 154
emh203 0:3d9c67d97d6f 155 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 156 px = pIn1;
emh203 0:3d9c67d97d6f 157
emh203 0:3d9c67d97d6f 158 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 159 py = pIn2;
emh203 0:3d9c67d97d6f 160
emh203 0:3d9c67d97d6f 161
emh203 0:3d9c67d97d6f 162 /* ------------------------
emh203 0:3d9c67d97d6f 163 * Stage1 process
emh203 0:3d9c67d97d6f 164 * ----------------------*/
emh203 0:3d9c67d97d6f 165
emh203 0:3d9c67d97d6f 166 /* The first stage starts here */
emh203 0:3d9c67d97d6f 167 while(blockSize1 > 0u)
emh203 0:3d9c67d97d6f 168 {
emh203 0:3d9c67d97d6f 169 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 170 sum = 0;
emh203 0:3d9c67d97d6f 171
emh203 0:3d9c67d97d6f 172 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 173 k = count >> 2u;
emh203 0:3d9c67d97d6f 174
emh203 0:3d9c67d97d6f 175 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 176 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 177 while(k > 0u)
emh203 0:3d9c67d97d6f 178 {
emh203 0:3d9c67d97d6f 179 /* x[0] , x[1] */
emh203 0:3d9c67d97d6f 180 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 181 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 182 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 183
emh203 0:3d9c67d97d6f 184 /* y[srcBLen - 1] , y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 185 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 186 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 187 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 188
emh203 0:3d9c67d97d6f 189 /* x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 190 /* x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 191 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 192
emh203 0:3d9c67d97d6f 193 /* x[2] , x[3] */
emh203 0:3d9c67d97d6f 194 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 195 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 196 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 197
emh203 0:3d9c67d97d6f 198 /* y[srcBLen - 3] , y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 199 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 200 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 201 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 202
emh203 0:3d9c67d97d6f 203 /* x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 204 /* x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 205 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 206
emh203 0:3d9c67d97d6f 207 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 208 k--;
emh203 0:3d9c67d97d6f 209 }
emh203 0:3d9c67d97d6f 210
emh203 0:3d9c67d97d6f 211 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 212 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 213 k = count % 0x4u;
emh203 0:3d9c67d97d6f 214
emh203 0:3d9c67d97d6f 215 while(k > 0u)
emh203 0:3d9c67d97d6f 216 {
emh203 0:3d9c67d97d6f 217 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 218 sum += ((q15_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 219
emh203 0:3d9c67d97d6f 220 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 221 k--;
emh203 0:3d9c67d97d6f 222 }
emh203 0:3d9c67d97d6f 223
emh203 0:3d9c67d97d6f 224 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 225 *pOut++ = (q7_t) (__SSAT(sum >> 7u, 8));
emh203 0:3d9c67d97d6f 226
emh203 0:3d9c67d97d6f 227 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 228 py = pIn2 + count;
emh203 0:3d9c67d97d6f 229 px = pIn1;
emh203 0:3d9c67d97d6f 230
emh203 0:3d9c67d97d6f 231 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 232 count++;
emh203 0:3d9c67d97d6f 233
emh203 0:3d9c67d97d6f 234 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 235 blockSize1--;
emh203 0:3d9c67d97d6f 236 }
emh203 0:3d9c67d97d6f 237
emh203 0:3d9c67d97d6f 238 /* --------------------------
emh203 0:3d9c67d97d6f 239 * Initializations of stage2
emh203 0:3d9c67d97d6f 240 * ------------------------*/
emh203 0:3d9c67d97d6f 241
emh203 0:3d9c67d97d6f 242 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emh203 0:3d9c67d97d6f 243 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emh203 0:3d9c67d97d6f 244 * ....
emh203 0:3d9c67d97d6f 245 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emh203 0:3d9c67d97d6f 246 */
emh203 0:3d9c67d97d6f 247
emh203 0:3d9c67d97d6f 248 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 249 px = pIn1;
emh203 0:3d9c67d97d6f 250
emh203 0:3d9c67d97d6f 251 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 252 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 253 py = pSrc2;
emh203 0:3d9c67d97d6f 254
emh203 0:3d9c67d97d6f 255 /* count is index by which the pointer pIn1 to be incremented */
emh203 0:3d9c67d97d6f 256 count = 0u;
emh203 0:3d9c67d97d6f 257
emh203 0:3d9c67d97d6f 258 /* -------------------
emh203 0:3d9c67d97d6f 259 * Stage2 process
emh203 0:3d9c67d97d6f 260 * ------------------*/
emh203 0:3d9c67d97d6f 261
emh203 0:3d9c67d97d6f 262 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emh203 0:3d9c67d97d6f 263 * So, to loop unroll over blockSize2,
emh203 0:3d9c67d97d6f 264 * srcBLen should be greater than or equal to 4 */
emh203 0:3d9c67d97d6f 265 if(srcBLen >= 4u)
emh203 0:3d9c67d97d6f 266 {
emh203 0:3d9c67d97d6f 267 /* Loop unroll over blockSize2, by 4 */
emh203 0:3d9c67d97d6f 268 blkCnt = blockSize2 >> 2u;
emh203 0:3d9c67d97d6f 269
emh203 0:3d9c67d97d6f 270 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 271 {
emh203 0:3d9c67d97d6f 272 /* Set all accumulators to zero */
emh203 0:3d9c67d97d6f 273 acc0 = 0;
emh203 0:3d9c67d97d6f 274 acc1 = 0;
emh203 0:3d9c67d97d6f 275 acc2 = 0;
emh203 0:3d9c67d97d6f 276 acc3 = 0;
emh203 0:3d9c67d97d6f 277
emh203 0:3d9c67d97d6f 278 /* read x[0], x[1], x[2] samples */
emh203 0:3d9c67d97d6f 279 x0 = *(px++);
emh203 0:3d9c67d97d6f 280 x1 = *(px++);
emh203 0:3d9c67d97d6f 281 x2 = *(px++);
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 284 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 285
emh203 0:3d9c67d97d6f 286 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 287 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 288 do
emh203 0:3d9c67d97d6f 289 {
emh203 0:3d9c67d97d6f 290 /* Read y[srcBLen - 1] sample */
emh203 0:3d9c67d97d6f 291 c0 = *(py--);
emh203 0:3d9c67d97d6f 292 /* Read y[srcBLen - 2] sample */
emh203 0:3d9c67d97d6f 293 c1 = *(py--);
emh203 0:3d9c67d97d6f 294
emh203 0:3d9c67d97d6f 295 /* Read x[3] sample */
emh203 0:3d9c67d97d6f 296 x3 = *(px++);
emh203 0:3d9c67d97d6f 297
emh203 0:3d9c67d97d6f 298 /* x[0] and x[1] are packed */
emh203 0:3d9c67d97d6f 299 in1 = (q15_t) x0;
emh203 0:3d9c67d97d6f 300 in2 = (q15_t) x1;
emh203 0:3d9c67d97d6f 301
emh203 0:3d9c67d97d6f 302 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 303
emh203 0:3d9c67d97d6f 304 /* y[srcBLen - 1] and y[srcBLen - 2] are packed */
emh203 0:3d9c67d97d6f 305 in1 = (q15_t) c0;
emh203 0:3d9c67d97d6f 306 in2 = (q15_t) c1;
emh203 0:3d9c67d97d6f 307
emh203 0:3d9c67d97d6f 308 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 309
emh203 0:3d9c67d97d6f 310 /* acc0 += x[0] * y[srcBLen - 1] + x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 311 acc0 = __SMLAD(input1, input2, acc0);
emh203 0:3d9c67d97d6f 312
emh203 0:3d9c67d97d6f 313 /* x[1] and x[2] are packed */
emh203 0:3d9c67d97d6f 314 in1 = (q15_t) x1;
emh203 0:3d9c67d97d6f 315 in2 = (q15_t) x2;
emh203 0:3d9c67d97d6f 316
emh203 0:3d9c67d97d6f 317 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 318
emh203 0:3d9c67d97d6f 319 /* acc1 += x[1] * y[srcBLen - 1] + x[2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 320 acc1 = __SMLAD(input1, input2, acc1);
emh203 0:3d9c67d97d6f 321
emh203 0:3d9c67d97d6f 322 /* x[2] and x[3] are packed */
emh203 0:3d9c67d97d6f 323 in1 = (q15_t) x2;
emh203 0:3d9c67d97d6f 324 in2 = (q15_t) x3;
emh203 0:3d9c67d97d6f 325
emh203 0:3d9c67d97d6f 326 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 327
emh203 0:3d9c67d97d6f 328 /* acc2 += x[2] * y[srcBLen - 1] + x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 329 acc2 = __SMLAD(input1, input2, acc2);
emh203 0:3d9c67d97d6f 330
emh203 0:3d9c67d97d6f 331 /* Read x[4] sample */
emh203 0:3d9c67d97d6f 332 x0 = *(px++);
emh203 0:3d9c67d97d6f 333
emh203 0:3d9c67d97d6f 334 /* x[3] and x[4] are packed */
emh203 0:3d9c67d97d6f 335 in1 = (q15_t) x3;
emh203 0:3d9c67d97d6f 336 in2 = (q15_t) x0;
emh203 0:3d9c67d97d6f 337
emh203 0:3d9c67d97d6f 338 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 339
emh203 0:3d9c67d97d6f 340 /* acc3 += x[3] * y[srcBLen - 1] + x[4] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 341 acc3 = __SMLAD(input1, input2, acc3);
emh203 0:3d9c67d97d6f 342
emh203 0:3d9c67d97d6f 343 /* Read y[srcBLen - 3] sample */
emh203 0:3d9c67d97d6f 344 c0 = *(py--);
emh203 0:3d9c67d97d6f 345 /* Read y[srcBLen - 4] sample */
emh203 0:3d9c67d97d6f 346 c1 = *(py--);
emh203 0:3d9c67d97d6f 347
emh203 0:3d9c67d97d6f 348 /* Read x[5] sample */
emh203 0:3d9c67d97d6f 349 x1 = *(px++);
emh203 0:3d9c67d97d6f 350
emh203 0:3d9c67d97d6f 351 /* x[2] and x[3] are packed */
emh203 0:3d9c67d97d6f 352 in1 = (q15_t) x2;
emh203 0:3d9c67d97d6f 353 in2 = (q15_t) x3;
emh203 0:3d9c67d97d6f 354
emh203 0:3d9c67d97d6f 355 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 356
emh203 0:3d9c67d97d6f 357 /* y[srcBLen - 3] and y[srcBLen - 4] are packed */
emh203 0:3d9c67d97d6f 358 in1 = (q15_t) c0;
emh203 0:3d9c67d97d6f 359 in2 = (q15_t) c1;
emh203 0:3d9c67d97d6f 360
emh203 0:3d9c67d97d6f 361 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 362
emh203 0:3d9c67d97d6f 363 /* acc0 += x[2] * y[srcBLen - 3] + x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 364 acc0 = __SMLAD(input1, input2, acc0);
emh203 0:3d9c67d97d6f 365
emh203 0:3d9c67d97d6f 366 /* x[3] and x[4] are packed */
emh203 0:3d9c67d97d6f 367 in1 = (q15_t) x3;
emh203 0:3d9c67d97d6f 368 in2 = (q15_t) x0;
emh203 0:3d9c67d97d6f 369
emh203 0:3d9c67d97d6f 370 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 371
emh203 0:3d9c67d97d6f 372 /* acc1 += x[3] * y[srcBLen - 3] + x[4] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 373 acc1 = __SMLAD(input1, input2, acc1);
emh203 0:3d9c67d97d6f 374
emh203 0:3d9c67d97d6f 375 /* x[4] and x[5] are packed */
emh203 0:3d9c67d97d6f 376 in1 = (q15_t) x0;
emh203 0:3d9c67d97d6f 377 in2 = (q15_t) x1;
emh203 0:3d9c67d97d6f 378
emh203 0:3d9c67d97d6f 379 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 380
emh203 0:3d9c67d97d6f 381 /* acc2 += x[4] * y[srcBLen - 3] + x[5] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 382 acc2 = __SMLAD(input1, input2, acc2);
emh203 0:3d9c67d97d6f 383
emh203 0:3d9c67d97d6f 384 /* Read x[6] sample */
emh203 0:3d9c67d97d6f 385 x2 = *(px++);
emh203 0:3d9c67d97d6f 386
emh203 0:3d9c67d97d6f 387 /* x[5] and x[6] are packed */
emh203 0:3d9c67d97d6f 388 in1 = (q15_t) x1;
emh203 0:3d9c67d97d6f 389 in2 = (q15_t) x2;
emh203 0:3d9c67d97d6f 390
emh203 0:3d9c67d97d6f 391 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 392
emh203 0:3d9c67d97d6f 393 /* acc3 += x[5] * y[srcBLen - 3] + x[6] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 394 acc3 = __SMLAD(input1, input2, acc3);
emh203 0:3d9c67d97d6f 395
emh203 0:3d9c67d97d6f 396 } while(--k);
emh203 0:3d9c67d97d6f 397
emh203 0:3d9c67d97d6f 398 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 399 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 400 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 401
emh203 0:3d9c67d97d6f 402 while(k > 0u)
emh203 0:3d9c67d97d6f 403 {
emh203 0:3d9c67d97d6f 404 /* Read y[srcBLen - 5] sample */
emh203 0:3d9c67d97d6f 405 c0 = *(py--);
emh203 0:3d9c67d97d6f 406
emh203 0:3d9c67d97d6f 407 /* Read x[7] sample */
emh203 0:3d9c67d97d6f 408 x3 = *(px++);
emh203 0:3d9c67d97d6f 409
emh203 0:3d9c67d97d6f 410 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 411 /* acc0 += x[4] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 412 acc0 += ((q15_t) x0 * c0);
emh203 0:3d9c67d97d6f 413 /* acc1 += x[5] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 414 acc1 += ((q15_t) x1 * c0);
emh203 0:3d9c67d97d6f 415 /* acc2 += x[6] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 416 acc2 += ((q15_t) x2 * c0);
emh203 0:3d9c67d97d6f 417 /* acc3 += x[7] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 418 acc3 += ((q15_t) x3 * c0);
emh203 0:3d9c67d97d6f 419
emh203 0:3d9c67d97d6f 420 /* Reuse the present samples for the next MAC */
emh203 0:3d9c67d97d6f 421 x0 = x1;
emh203 0:3d9c67d97d6f 422 x1 = x2;
emh203 0:3d9c67d97d6f 423 x2 = x3;
emh203 0:3d9c67d97d6f 424
emh203 0:3d9c67d97d6f 425 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 426 k--;
emh203 0:3d9c67d97d6f 427 }
emh203 0:3d9c67d97d6f 428
emh203 0:3d9c67d97d6f 429
emh203 0:3d9c67d97d6f 430 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 431 *pOut++ = (q7_t) (__SSAT(acc0 >> 7u, 8));
emh203 0:3d9c67d97d6f 432 *pOut++ = (q7_t) (__SSAT(acc1 >> 7u, 8));
emh203 0:3d9c67d97d6f 433 *pOut++ = (q7_t) (__SSAT(acc2 >> 7u, 8));
emh203 0:3d9c67d97d6f 434 *pOut++ = (q7_t) (__SSAT(acc3 >> 7u, 8));
emh203 0:3d9c67d97d6f 435
emh203 0:3d9c67d97d6f 436 /* Increment the pointer pIn1 index, count by 4 */
emh203 0:3d9c67d97d6f 437 count += 4u;
emh203 0:3d9c67d97d6f 438
emh203 0:3d9c67d97d6f 439 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 440 px = pIn1 + count;
emh203 0:3d9c67d97d6f 441 py = pSrc2;
emh203 0:3d9c67d97d6f 442
emh203 0:3d9c67d97d6f 443 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 444 blkCnt--;
emh203 0:3d9c67d97d6f 445 }
emh203 0:3d9c67d97d6f 446
emh203 0:3d9c67d97d6f 447 /* If the blockSize2 is not a multiple of 4, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 448 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 449 blkCnt = blockSize2 % 0x4u;
emh203 0:3d9c67d97d6f 450
emh203 0:3d9c67d97d6f 451 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 452 {
emh203 0:3d9c67d97d6f 453 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 454 sum = 0;
emh203 0:3d9c67d97d6f 455
emh203 0:3d9c67d97d6f 456 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 457 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 458
emh203 0:3d9c67d97d6f 459 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 460 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 461 while(k > 0u)
emh203 0:3d9c67d97d6f 462 {
emh203 0:3d9c67d97d6f 463
emh203 0:3d9c67d97d6f 464 /* Reading two inputs of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 465 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 466 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 467 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 468
emh203 0:3d9c67d97d6f 469 /* Reading two inputs of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 470 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 471 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 472 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 473
emh203 0:3d9c67d97d6f 474 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 475 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 476
emh203 0:3d9c67d97d6f 477 /* Reading two inputs of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 478 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 479 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 480 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 481
emh203 0:3d9c67d97d6f 482 /* Reading two inputs of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 483 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 484 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 485 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 486
emh203 0:3d9c67d97d6f 487 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 488 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 489
emh203 0:3d9c67d97d6f 490 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 491 k--;
emh203 0:3d9c67d97d6f 492 }
emh203 0:3d9c67d97d6f 493
emh203 0:3d9c67d97d6f 494 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 495 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 496 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 497
emh203 0:3d9c67d97d6f 498 while(k > 0u)
emh203 0:3d9c67d97d6f 499 {
emh203 0:3d9c67d97d6f 500 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 501 sum += ((q15_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 502
emh203 0:3d9c67d97d6f 503 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 504 k--;
emh203 0:3d9c67d97d6f 505 }
emh203 0:3d9c67d97d6f 506
emh203 0:3d9c67d97d6f 507 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 508 *pOut++ = (q7_t) (__SSAT(sum >> 7u, 8));
emh203 0:3d9c67d97d6f 509
emh203 0:3d9c67d97d6f 510 /* Increment the pointer pIn1 index, count by 1 */
emh203 0:3d9c67d97d6f 511 count++;
emh203 0:3d9c67d97d6f 512
emh203 0:3d9c67d97d6f 513 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 514 px = pIn1 + count;
emh203 0:3d9c67d97d6f 515 py = pSrc2;
emh203 0:3d9c67d97d6f 516
emh203 0:3d9c67d97d6f 517 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 518 blkCnt--;
emh203 0:3d9c67d97d6f 519 }
emh203 0:3d9c67d97d6f 520 }
emh203 0:3d9c67d97d6f 521 else
emh203 0:3d9c67d97d6f 522 {
emh203 0:3d9c67d97d6f 523 /* If the srcBLen is not a multiple of 4,
emh203 0:3d9c67d97d6f 524 * the blockSize2 loop cannot be unrolled by 4 */
emh203 0:3d9c67d97d6f 525 blkCnt = blockSize2;
emh203 0:3d9c67d97d6f 526
emh203 0:3d9c67d97d6f 527 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 528 {
emh203 0:3d9c67d97d6f 529 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 530 sum = 0;
emh203 0:3d9c67d97d6f 531
emh203 0:3d9c67d97d6f 532 /* srcBLen number of MACS should be performed */
emh203 0:3d9c67d97d6f 533 k = srcBLen;
emh203 0:3d9c67d97d6f 534
emh203 0:3d9c67d97d6f 535 while(k > 0u)
emh203 0:3d9c67d97d6f 536 {
emh203 0:3d9c67d97d6f 537 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 538 sum += ((q15_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 539
emh203 0:3d9c67d97d6f 540 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 541 k--;
emh203 0:3d9c67d97d6f 542 }
emh203 0:3d9c67d97d6f 543
emh203 0:3d9c67d97d6f 544 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 545 *pOut++ = (q7_t) (__SSAT(sum >> 7u, 8));
emh203 0:3d9c67d97d6f 546
emh203 0:3d9c67d97d6f 547 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 548 count++;
emh203 0:3d9c67d97d6f 549
emh203 0:3d9c67d97d6f 550 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 551 px = pIn1 + count;
emh203 0:3d9c67d97d6f 552 py = pSrc2;
emh203 0:3d9c67d97d6f 553
emh203 0:3d9c67d97d6f 554 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 555 blkCnt--;
emh203 0:3d9c67d97d6f 556 }
emh203 0:3d9c67d97d6f 557 }
emh203 0:3d9c67d97d6f 558
emh203 0:3d9c67d97d6f 559
emh203 0:3d9c67d97d6f 560 /* --------------------------
emh203 0:3d9c67d97d6f 561 * Initializations of stage3
emh203 0:3d9c67d97d6f 562 * -------------------------*/
emh203 0:3d9c67d97d6f 563
emh203 0:3d9c67d97d6f 564 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emh203 0:3d9c67d97d6f 565 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emh203 0:3d9c67d97d6f 566 * ....
emh203 0:3d9c67d97d6f 567 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emh203 0:3d9c67d97d6f 568 * sum += x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 569 */
emh203 0:3d9c67d97d6f 570
emh203 0:3d9c67d97d6f 571 /* In this stage the MAC operations are decreased by 1 for every iteration.
emh203 0:3d9c67d97d6f 572 The blockSize3 variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 573
emh203 0:3d9c67d97d6f 574 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 575 pSrc1 = pIn1 + (srcALen - (srcBLen - 1u));
emh203 0:3d9c67d97d6f 576 px = pSrc1;
emh203 0:3d9c67d97d6f 577
emh203 0:3d9c67d97d6f 578 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 579 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 580 py = pSrc2;
emh203 0:3d9c67d97d6f 581
emh203 0:3d9c67d97d6f 582 /* -------------------
emh203 0:3d9c67d97d6f 583 * Stage3 process
emh203 0:3d9c67d97d6f 584 * ------------------*/
emh203 0:3d9c67d97d6f 585
emh203 0:3d9c67d97d6f 586 while(blockSize3 > 0u)
emh203 0:3d9c67d97d6f 587 {
emh203 0:3d9c67d97d6f 588 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 589 sum = 0;
emh203 0:3d9c67d97d6f 590
emh203 0:3d9c67d97d6f 591 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 592 k = blockSize3 >> 2u;
emh203 0:3d9c67d97d6f 593
emh203 0:3d9c67d97d6f 594 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 595 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 596 while(k > 0u)
emh203 0:3d9c67d97d6f 597 {
emh203 0:3d9c67d97d6f 598 /* Reading two inputs, x[srcALen - srcBLen + 1] and x[srcALen - srcBLen + 2] of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 599 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 600 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 601 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 602
emh203 0:3d9c67d97d6f 603 /* Reading two inputs, y[srcBLen - 1] and y[srcBLen - 2] of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 604 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 605 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 606 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 607
emh203 0:3d9c67d97d6f 608 /* sum += x[srcALen - srcBLen + 1] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 609 /* sum += x[srcALen - srcBLen + 2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 610 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 611
emh203 0:3d9c67d97d6f 612 /* Reading two inputs, x[srcALen - srcBLen + 3] and x[srcALen - srcBLen + 4] of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 613 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 614 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 615 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 616
emh203 0:3d9c67d97d6f 617 /* Reading two inputs, y[srcBLen - 3] and y[srcBLen - 4] of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 618 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 619 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 620 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16u);
emh203 0:3d9c67d97d6f 621
emh203 0:3d9c67d97d6f 622 /* sum += x[srcALen - srcBLen + 3] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 623 /* sum += x[srcALen - srcBLen + 4] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 624 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 625
emh203 0:3d9c67d97d6f 626 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 627 k--;
emh203 0:3d9c67d97d6f 628 }
emh203 0:3d9c67d97d6f 629
emh203 0:3d9c67d97d6f 630 /* If the blockSize3 is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 631 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 632 k = blockSize3 % 0x4u;
emh203 0:3d9c67d97d6f 633
emh203 0:3d9c67d97d6f 634 while(k > 0u)
emh203 0:3d9c67d97d6f 635 {
emh203 0:3d9c67d97d6f 636 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 637 sum += ((q15_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 638
emh203 0:3d9c67d97d6f 639 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 640 k--;
emh203 0:3d9c67d97d6f 641 }
emh203 0:3d9c67d97d6f 642
emh203 0:3d9c67d97d6f 643 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 644 *pOut++ = (q7_t) (__SSAT(sum >> 7u, 8));
emh203 0:3d9c67d97d6f 645
emh203 0:3d9c67d97d6f 646 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 647 px = ++pSrc1;
emh203 0:3d9c67d97d6f 648 py = pSrc2;
emh203 0:3d9c67d97d6f 649
emh203 0:3d9c67d97d6f 650 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 651 blockSize3--;
emh203 0:3d9c67d97d6f 652 }
emh203 0:3d9c67d97d6f 653
emh203 0:3d9c67d97d6f 654 #else
emh203 0:3d9c67d97d6f 655
emh203 0:3d9c67d97d6f 656 /* Run the below code for Cortex-M0 */
emh203 0:3d9c67d97d6f 657
emh203 0:3d9c67d97d6f 658 q7_t *pIn1 = pSrcA; /* input pointer */
emh203 0:3d9c67d97d6f 659 q7_t *pIn2 = pSrcB; /* coefficient pointer */
emh203 0:3d9c67d97d6f 660 q31_t sum; /* Accumulator */
emh203 0:3d9c67d97d6f 661 uint32_t i, j; /* loop counter */
emh203 0:3d9c67d97d6f 662
emh203 0:3d9c67d97d6f 663 /* Loop to calculate output of convolution for output length number of times */
emh203 0:3d9c67d97d6f 664 for (i = 0; i < (srcALen + srcBLen - 1); i++)
emh203 0:3d9c67d97d6f 665 {
emh203 0:3d9c67d97d6f 666 /* Initialize sum with zero to carry on MAC operations */
emh203 0:3d9c67d97d6f 667 sum = 0;
emh203 0:3d9c67d97d6f 668
emh203 0:3d9c67d97d6f 669 /* Loop to perform MAC operations according to convolution equation */
emh203 0:3d9c67d97d6f 670 for (j = 0; j <= i; j++)
emh203 0:3d9c67d97d6f 671 {
emh203 0:3d9c67d97d6f 672 /* Check the array limitations */
emh203 0:3d9c67d97d6f 673 if(((i - j) < srcBLen) && (j < srcALen))
emh203 0:3d9c67d97d6f 674 {
emh203 0:3d9c67d97d6f 675 /* z[i] += x[i-j] * y[j] */
emh203 0:3d9c67d97d6f 676 sum += (q15_t) pIn1[j] * (pIn2[i - j]);
emh203 0:3d9c67d97d6f 677 }
emh203 0:3d9c67d97d6f 678 }
emh203 0:3d9c67d97d6f 679
emh203 0:3d9c67d97d6f 680 /* Store the output in the destination buffer */
emh203 0:3d9c67d97d6f 681 pDst[i] = (q7_t) __SSAT((sum >> 7u), 8u);
emh203 0:3d9c67d97d6f 682 }
emh203 0:3d9c67d97d6f 683
emh203 0:3d9c67d97d6f 684 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emh203 0:3d9c67d97d6f 685
emh203 0:3d9c67d97d6f 686 }
emh203 0:3d9c67d97d6f 687
emh203 0:3d9c67d97d6f 688 /**
emh203 0:3d9c67d97d6f 689 * @} end of Conv group
emh203 0:3d9c67d97d6f 690 */