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_partial_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_partial_q7.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Partial 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 PartialConv
emh203 0:3d9c67d97d6f 49 * @{
emh203 0:3d9c67d97d6f 50 */
emh203 0:3d9c67d97d6f 51
emh203 0:3d9c67d97d6f 52 /**
emh203 0:3d9c67d97d6f 53 * @brief Partial 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.
emh203 0:3d9c67d97d6f 59 * @param[in] firstIndex is the first output sample to start with.
emh203 0:3d9c67d97d6f 60 * @param[in] numPoints is the number of output points to be computed.
emh203 0:3d9c67d97d6f 61 * @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].
emh203 0:3d9c67d97d6f 62 *
emh203 0:3d9c67d97d6f 63 * \par
emh203 0:3d9c67d97d6f 64 * Refer the function <code>arm_conv_partial_opt_q7()</code> for a faster implementation of this function.
emh203 0:3d9c67d97d6f 65 *
emh203 0:3d9c67d97d6f 66 */
emh203 0:3d9c67d97d6f 67
emh203 0:3d9c67d97d6f 68 arm_status arm_conv_partial_q7(
emh203 0:3d9c67d97d6f 69 q7_t * pSrcA,
emh203 0:3d9c67d97d6f 70 uint32_t srcALen,
emh203 0:3d9c67d97d6f 71 q7_t * pSrcB,
emh203 0:3d9c67d97d6f 72 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 73 q7_t * pDst,
emh203 0:3d9c67d97d6f 74 uint32_t firstIndex,
emh203 0:3d9c67d97d6f 75 uint32_t numPoints)
emh203 0:3d9c67d97d6f 76 {
emh203 0:3d9c67d97d6f 77
emh203 0:3d9c67d97d6f 78
emh203 0:3d9c67d97d6f 79 #ifndef ARM_MATH_CM0_FAMILY
emh203 0:3d9c67d97d6f 80
emh203 0:3d9c67d97d6f 81 /* Run the below code for Cortex-M4 and Cortex-M3 */
emh203 0:3d9c67d97d6f 82
emh203 0:3d9c67d97d6f 83 q7_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 84 q7_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 85 q7_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 86 q7_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 87 q7_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 88 q7_t *pSrc1, *pSrc2; /* Intermediate pointers */
emh203 0:3d9c67d97d6f 89 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulator */
emh203 0:3d9c67d97d6f 90 q31_t input1, input2;
emh203 0:3d9c67d97d6f 91 q15_t in1, in2;
emh203 0:3d9c67d97d6f 92 q7_t x0, x1, x2, x3, c0, c1;
emh203 0:3d9c67d97d6f 93 uint32_t j, k, count, check, blkCnt;
emh203 0:3d9c67d97d6f 94 int32_t blockSize1, blockSize2, blockSize3; /* loop counter */
emh203 0:3d9c67d97d6f 95 arm_status status;
emh203 0:3d9c67d97d6f 96
emh203 0:3d9c67d97d6f 97
emh203 0:3d9c67d97d6f 98 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 99 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 100 {
emh203 0:3d9c67d97d6f 101 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 102 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 103 }
emh203 0:3d9c67d97d6f 104 else
emh203 0:3d9c67d97d6f 105 {
emh203 0:3d9c67d97d6f 106
emh203 0:3d9c67d97d6f 107 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 108 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 109 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 110 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 111 {
emh203 0:3d9c67d97d6f 112 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 113 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 114
emh203 0:3d9c67d97d6f 115 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 116 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 117 }
emh203 0:3d9c67d97d6f 118 else
emh203 0:3d9c67d97d6f 119 {
emh203 0:3d9c67d97d6f 120 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 121 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 122
emh203 0:3d9c67d97d6f 123 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 124 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 125
emh203 0:3d9c67d97d6f 126 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 127 j = srcBLen;
emh203 0:3d9c67d97d6f 128 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 129 srcALen = j;
emh203 0:3d9c67d97d6f 130 }
emh203 0:3d9c67d97d6f 131
emh203 0:3d9c67d97d6f 132 /* Conditions to check which loopCounter holds
emh203 0:3d9c67d97d6f 133 * the first and last indices of the output samples to be calculated. */
emh203 0:3d9c67d97d6f 134 check = firstIndex + numPoints;
emh203 0:3d9c67d97d6f 135 blockSize3 = ((int32_t)check > (int32_t)srcALen) ? (int32_t)check - (int32_t)srcALen : 0;
emh203 0:3d9c67d97d6f 136 blockSize3 = ((int32_t)firstIndex > (int32_t)srcALen - 1) ? blockSize3 - (int32_t)firstIndex + (int32_t)srcALen : blockSize3;
emh203 0:3d9c67d97d6f 137 blockSize1 = (((int32_t) srcBLen - 1) - (int32_t) firstIndex);
emh203 0:3d9c67d97d6f 138 blockSize1 = (blockSize1 > 0) ? ((check > (srcBLen - 1u)) ? blockSize1 :
emh203 0:3d9c67d97d6f 139 (int32_t) numPoints) : 0;
emh203 0:3d9c67d97d6f 140 blockSize2 = (int32_t) check - ((blockSize3 + blockSize1) +
emh203 0:3d9c67d97d6f 141 (int32_t) firstIndex);
emh203 0:3d9c67d97d6f 142 blockSize2 = (blockSize2 > 0) ? blockSize2 : 0;
emh203 0:3d9c67d97d6f 143
emh203 0:3d9c67d97d6f 144 /* 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 145 /* The function is internally
emh203 0:3d9c67d97d6f 146 * divided into three stages according to the number of multiplications that has to be
emh203 0:3d9c67d97d6f 147 * taken place between inputA samples and inputB samples. In the first stage of the
emh203 0:3d9c67d97d6f 148 * algorithm, the multiplications increase by one for every iteration.
emh203 0:3d9c67d97d6f 149 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emh203 0:3d9c67d97d6f 150 * In the third stage of the algorithm, the multiplications decrease by one
emh203 0:3d9c67d97d6f 151 * for every iteration. */
emh203 0:3d9c67d97d6f 152
emh203 0:3d9c67d97d6f 153 /* Set the output pointer to point to the firstIndex
emh203 0:3d9c67d97d6f 154 * of the output sample to be calculated. */
emh203 0:3d9c67d97d6f 155 pOut = pDst + firstIndex;
emh203 0:3d9c67d97d6f 156
emh203 0:3d9c67d97d6f 157 /* --------------------------
emh203 0:3d9c67d97d6f 158 * Initializations of stage1
emh203 0:3d9c67d97d6f 159 * -------------------------*/
emh203 0:3d9c67d97d6f 160
emh203 0:3d9c67d97d6f 161 /* sum = x[0] * y[0]
emh203 0:3d9c67d97d6f 162 * sum = x[0] * y[1] + x[1] * y[0]
emh203 0:3d9c67d97d6f 163 * ....
emh203 0:3d9c67d97d6f 164 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emh203 0:3d9c67d97d6f 165 */
emh203 0:3d9c67d97d6f 166
emh203 0:3d9c67d97d6f 167 /* In this stage the MAC operations are increased by 1 for every iteration.
emh203 0:3d9c67d97d6f 168 The count variable holds the number of MAC operations performed.
emh203 0:3d9c67d97d6f 169 Since the partial convolution starts from from firstIndex
emh203 0:3d9c67d97d6f 170 Number of Macs to be performed is firstIndex + 1 */
emh203 0:3d9c67d97d6f 171 count = 1u + firstIndex;
emh203 0:3d9c67d97d6f 172
emh203 0:3d9c67d97d6f 173 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 174 px = pIn1;
emh203 0:3d9c67d97d6f 175
emh203 0:3d9c67d97d6f 176 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 177 pSrc2 = pIn2 + firstIndex;
emh203 0:3d9c67d97d6f 178 py = pSrc2;
emh203 0:3d9c67d97d6f 179
emh203 0:3d9c67d97d6f 180 /* ------------------------
emh203 0:3d9c67d97d6f 181 * Stage1 process
emh203 0:3d9c67d97d6f 182 * ----------------------*/
emh203 0:3d9c67d97d6f 183
emh203 0:3d9c67d97d6f 184 /* The first stage starts here */
emh203 0:3d9c67d97d6f 185 while(blockSize1 > 0)
emh203 0:3d9c67d97d6f 186 {
emh203 0:3d9c67d97d6f 187 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 188 sum = 0;
emh203 0:3d9c67d97d6f 189
emh203 0:3d9c67d97d6f 190 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 191 k = count >> 2u;
emh203 0:3d9c67d97d6f 192
emh203 0:3d9c67d97d6f 193 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 194 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 195 while(k > 0u)
emh203 0:3d9c67d97d6f 196 {
emh203 0:3d9c67d97d6f 197 /* x[0] , x[1] */
emh203 0:3d9c67d97d6f 198 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 199 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 200 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 201
emh203 0:3d9c67d97d6f 202 /* y[srcBLen - 1] , y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 203 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 204 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 205 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 206
emh203 0:3d9c67d97d6f 207 /* x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 208 /* x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 209 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 210
emh203 0:3d9c67d97d6f 211 /* x[2] , x[3] */
emh203 0:3d9c67d97d6f 212 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 213 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 214 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 215
emh203 0:3d9c67d97d6f 216 /* y[srcBLen - 3] , y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 217 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 218 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 219 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 220
emh203 0:3d9c67d97d6f 221 /* x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 222 /* x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 223 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 224
emh203 0:3d9c67d97d6f 225 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 226 k--;
emh203 0:3d9c67d97d6f 227 }
emh203 0:3d9c67d97d6f 228
emh203 0:3d9c67d97d6f 229 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 230 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 231 k = count % 0x4u;
emh203 0:3d9c67d97d6f 232
emh203 0:3d9c67d97d6f 233 while(k > 0u)
emh203 0:3d9c67d97d6f 234 {
emh203 0:3d9c67d97d6f 235 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 236 sum += ((q31_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 237
emh203 0:3d9c67d97d6f 238 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 239 k--;
emh203 0:3d9c67d97d6f 240 }
emh203 0:3d9c67d97d6f 241
emh203 0:3d9c67d97d6f 242 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 243 *pOut++ = (q7_t) (__SSAT(sum >> 7, 8));
emh203 0:3d9c67d97d6f 244
emh203 0:3d9c67d97d6f 245 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 246 py = ++pSrc2;
emh203 0:3d9c67d97d6f 247 px = pIn1;
emh203 0:3d9c67d97d6f 248
emh203 0:3d9c67d97d6f 249 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 250 count++;
emh203 0:3d9c67d97d6f 251
emh203 0:3d9c67d97d6f 252 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 253 blockSize1--;
emh203 0:3d9c67d97d6f 254 }
emh203 0:3d9c67d97d6f 255
emh203 0:3d9c67d97d6f 256 /* --------------------------
emh203 0:3d9c67d97d6f 257 * Initializations of stage2
emh203 0:3d9c67d97d6f 258 * ------------------------*/
emh203 0:3d9c67d97d6f 259
emh203 0:3d9c67d97d6f 260 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emh203 0:3d9c67d97d6f 261 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emh203 0:3d9c67d97d6f 262 * ....
emh203 0:3d9c67d97d6f 263 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emh203 0:3d9c67d97d6f 264 */
emh203 0:3d9c67d97d6f 265
emh203 0:3d9c67d97d6f 266 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 267 if((int32_t)firstIndex - (int32_t)srcBLen + 1 > 0)
emh203 0:3d9c67d97d6f 268 {
emh203 0:3d9c67d97d6f 269 px = pIn1 + firstIndex - srcBLen + 1;
emh203 0:3d9c67d97d6f 270 }
emh203 0:3d9c67d97d6f 271 else
emh203 0:3d9c67d97d6f 272 {
emh203 0:3d9c67d97d6f 273 px = pIn1;
emh203 0:3d9c67d97d6f 274 }
emh203 0:3d9c67d97d6f 275
emh203 0:3d9c67d97d6f 276 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 277 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 278 py = pSrc2;
emh203 0:3d9c67d97d6f 279
emh203 0:3d9c67d97d6f 280 /* count is index by which the pointer pIn1 to be incremented */
emh203 0:3d9c67d97d6f 281 count = 0u;
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283 /* -------------------
emh203 0:3d9c67d97d6f 284 * Stage2 process
emh203 0:3d9c67d97d6f 285 * ------------------*/
emh203 0:3d9c67d97d6f 286
emh203 0:3d9c67d97d6f 287 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emh203 0:3d9c67d97d6f 288 * So, to loop unroll over blockSize2,
emh203 0:3d9c67d97d6f 289 * srcBLen should be greater than or equal to 4 */
emh203 0:3d9c67d97d6f 290 if(srcBLen >= 4u)
emh203 0:3d9c67d97d6f 291 {
emh203 0:3d9c67d97d6f 292 /* Loop unroll over blockSize2, by 4 */
emh203 0:3d9c67d97d6f 293 blkCnt = ((uint32_t) blockSize2 >> 2u);
emh203 0:3d9c67d97d6f 294
emh203 0:3d9c67d97d6f 295 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 296 {
emh203 0:3d9c67d97d6f 297 /* Set all accumulators to zero */
emh203 0:3d9c67d97d6f 298 acc0 = 0;
emh203 0:3d9c67d97d6f 299 acc1 = 0;
emh203 0:3d9c67d97d6f 300 acc2 = 0;
emh203 0:3d9c67d97d6f 301 acc3 = 0;
emh203 0:3d9c67d97d6f 302
emh203 0:3d9c67d97d6f 303 /* read x[0], x[1], x[2] samples */
emh203 0:3d9c67d97d6f 304 x0 = *(px++);
emh203 0:3d9c67d97d6f 305 x1 = *(px++);
emh203 0:3d9c67d97d6f 306 x2 = *(px++);
emh203 0:3d9c67d97d6f 307
emh203 0:3d9c67d97d6f 308 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 309 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 310
emh203 0:3d9c67d97d6f 311 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 312 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 313 do
emh203 0:3d9c67d97d6f 314 {
emh203 0:3d9c67d97d6f 315 /* Read y[srcBLen - 1] sample */
emh203 0:3d9c67d97d6f 316 c0 = *(py--);
emh203 0:3d9c67d97d6f 317 /* Read y[srcBLen - 2] sample */
emh203 0:3d9c67d97d6f 318 c1 = *(py--);
emh203 0:3d9c67d97d6f 319
emh203 0:3d9c67d97d6f 320 /* Read x[3] sample */
emh203 0:3d9c67d97d6f 321 x3 = *(px++);
emh203 0:3d9c67d97d6f 322
emh203 0:3d9c67d97d6f 323 /* x[0] and x[1] are packed */
emh203 0:3d9c67d97d6f 324 in1 = (q15_t) x0;
emh203 0:3d9c67d97d6f 325 in2 = (q15_t) x1;
emh203 0:3d9c67d97d6f 326
emh203 0:3d9c67d97d6f 327 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 328
emh203 0:3d9c67d97d6f 329 /* y[srcBLen - 1] and y[srcBLen - 2] are packed */
emh203 0:3d9c67d97d6f 330 in1 = (q15_t) c0;
emh203 0:3d9c67d97d6f 331 in2 = (q15_t) c1;
emh203 0:3d9c67d97d6f 332
emh203 0:3d9c67d97d6f 333 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 334
emh203 0:3d9c67d97d6f 335 /* acc0 += x[0] * y[srcBLen - 1] + x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 336 acc0 = __SMLAD(input1, input2, acc0);
emh203 0:3d9c67d97d6f 337
emh203 0:3d9c67d97d6f 338 /* x[1] and x[2] are packed */
emh203 0:3d9c67d97d6f 339 in1 = (q15_t) x1;
emh203 0:3d9c67d97d6f 340 in2 = (q15_t) x2;
emh203 0:3d9c67d97d6f 341
emh203 0:3d9c67d97d6f 342 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 343
emh203 0:3d9c67d97d6f 344 /* acc1 += x[1] * y[srcBLen - 1] + x[2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 345 acc1 = __SMLAD(input1, input2, acc1);
emh203 0:3d9c67d97d6f 346
emh203 0:3d9c67d97d6f 347 /* x[2] and x[3] are packed */
emh203 0:3d9c67d97d6f 348 in1 = (q15_t) x2;
emh203 0:3d9c67d97d6f 349 in2 = (q15_t) x3;
emh203 0:3d9c67d97d6f 350
emh203 0:3d9c67d97d6f 351 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 352
emh203 0:3d9c67d97d6f 353 /* acc2 += x[2] * y[srcBLen - 1] + x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 354 acc2 = __SMLAD(input1, input2, acc2);
emh203 0:3d9c67d97d6f 355
emh203 0:3d9c67d97d6f 356 /* Read x[4] sample */
emh203 0:3d9c67d97d6f 357 x0 = *(px++);
emh203 0:3d9c67d97d6f 358
emh203 0:3d9c67d97d6f 359 /* x[3] and x[4] are packed */
emh203 0:3d9c67d97d6f 360 in1 = (q15_t) x3;
emh203 0:3d9c67d97d6f 361 in2 = (q15_t) x0;
emh203 0:3d9c67d97d6f 362
emh203 0:3d9c67d97d6f 363 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 364
emh203 0:3d9c67d97d6f 365 /* acc3 += x[3] * y[srcBLen - 1] + x[4] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 366 acc3 = __SMLAD(input1, input2, acc3);
emh203 0:3d9c67d97d6f 367
emh203 0:3d9c67d97d6f 368 /* Read y[srcBLen - 3] sample */
emh203 0:3d9c67d97d6f 369 c0 = *(py--);
emh203 0:3d9c67d97d6f 370 /* Read y[srcBLen - 4] sample */
emh203 0:3d9c67d97d6f 371 c1 = *(py--);
emh203 0:3d9c67d97d6f 372
emh203 0:3d9c67d97d6f 373 /* Read x[5] sample */
emh203 0:3d9c67d97d6f 374 x1 = *(px++);
emh203 0:3d9c67d97d6f 375
emh203 0:3d9c67d97d6f 376 /* x[2] and x[3] are packed */
emh203 0:3d9c67d97d6f 377 in1 = (q15_t) x2;
emh203 0:3d9c67d97d6f 378 in2 = (q15_t) x3;
emh203 0:3d9c67d97d6f 379
emh203 0:3d9c67d97d6f 380 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 381
emh203 0:3d9c67d97d6f 382 /* y[srcBLen - 3] and y[srcBLen - 4] are packed */
emh203 0:3d9c67d97d6f 383 in1 = (q15_t) c0;
emh203 0:3d9c67d97d6f 384 in2 = (q15_t) c1;
emh203 0:3d9c67d97d6f 385
emh203 0:3d9c67d97d6f 386 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 387
emh203 0:3d9c67d97d6f 388 /* acc0 += x[2] * y[srcBLen - 3] + x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 389 acc0 = __SMLAD(input1, input2, acc0);
emh203 0:3d9c67d97d6f 390
emh203 0:3d9c67d97d6f 391 /* x[3] and x[4] are packed */
emh203 0:3d9c67d97d6f 392 in1 = (q15_t) x3;
emh203 0:3d9c67d97d6f 393 in2 = (q15_t) x0;
emh203 0:3d9c67d97d6f 394
emh203 0:3d9c67d97d6f 395 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 396
emh203 0:3d9c67d97d6f 397 /* acc1 += x[3] * y[srcBLen - 3] + x[4] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 398 acc1 = __SMLAD(input1, input2, acc1);
emh203 0:3d9c67d97d6f 399
emh203 0:3d9c67d97d6f 400 /* x[4] and x[5] are packed */
emh203 0:3d9c67d97d6f 401 in1 = (q15_t) x0;
emh203 0:3d9c67d97d6f 402 in2 = (q15_t) x1;
emh203 0:3d9c67d97d6f 403
emh203 0:3d9c67d97d6f 404 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 405
emh203 0:3d9c67d97d6f 406 /* acc2 += x[4] * y[srcBLen - 3] + x[5] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 407 acc2 = __SMLAD(input1, input2, acc2);
emh203 0:3d9c67d97d6f 408
emh203 0:3d9c67d97d6f 409 /* Read x[6] sample */
emh203 0:3d9c67d97d6f 410 x2 = *(px++);
emh203 0:3d9c67d97d6f 411
emh203 0:3d9c67d97d6f 412 /* x[5] and x[6] are packed */
emh203 0:3d9c67d97d6f 413 in1 = (q15_t) x1;
emh203 0:3d9c67d97d6f 414 in2 = (q15_t) x2;
emh203 0:3d9c67d97d6f 415
emh203 0:3d9c67d97d6f 416 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 417
emh203 0:3d9c67d97d6f 418 /* acc3 += x[5] * y[srcBLen - 3] + x[6] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 419 acc3 = __SMLAD(input1, input2, acc3);
emh203 0:3d9c67d97d6f 420
emh203 0:3d9c67d97d6f 421 } while(--k);
emh203 0:3d9c67d97d6f 422
emh203 0:3d9c67d97d6f 423 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 424 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 425 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 426
emh203 0:3d9c67d97d6f 427 while(k > 0u)
emh203 0:3d9c67d97d6f 428 {
emh203 0:3d9c67d97d6f 429 /* Read y[srcBLen - 5] sample */
emh203 0:3d9c67d97d6f 430 c0 = *(py--);
emh203 0:3d9c67d97d6f 431
emh203 0:3d9c67d97d6f 432 /* Read x[7] sample */
emh203 0:3d9c67d97d6f 433 x3 = *(px++);
emh203 0:3d9c67d97d6f 434
emh203 0:3d9c67d97d6f 435 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 436 /* acc0 += x[4] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 437 acc0 += ((q31_t) x0 * c0);
emh203 0:3d9c67d97d6f 438 /* acc1 += x[5] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 439 acc1 += ((q31_t) x1 * c0);
emh203 0:3d9c67d97d6f 440 /* acc2 += x[6] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 441 acc2 += ((q31_t) x2 * c0);
emh203 0:3d9c67d97d6f 442 /* acc3 += x[7] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 443 acc3 += ((q31_t) x3 * c0);
emh203 0:3d9c67d97d6f 444
emh203 0:3d9c67d97d6f 445 /* Reuse the present samples for the next MAC */
emh203 0:3d9c67d97d6f 446 x0 = x1;
emh203 0:3d9c67d97d6f 447 x1 = x2;
emh203 0:3d9c67d97d6f 448 x2 = x3;
emh203 0:3d9c67d97d6f 449
emh203 0:3d9c67d97d6f 450 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 451 k--;
emh203 0:3d9c67d97d6f 452 }
emh203 0:3d9c67d97d6f 453
emh203 0:3d9c67d97d6f 454 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 455 *pOut++ = (q7_t) (__SSAT(acc0 >> 7, 8));
emh203 0:3d9c67d97d6f 456 *pOut++ = (q7_t) (__SSAT(acc1 >> 7, 8));
emh203 0:3d9c67d97d6f 457 *pOut++ = (q7_t) (__SSAT(acc2 >> 7, 8));
emh203 0:3d9c67d97d6f 458 *pOut++ = (q7_t) (__SSAT(acc3 >> 7, 8));
emh203 0:3d9c67d97d6f 459
emh203 0:3d9c67d97d6f 460 /* Increment the pointer pIn1 index, count by 4 */
emh203 0:3d9c67d97d6f 461 count += 4u;
emh203 0:3d9c67d97d6f 462
emh203 0:3d9c67d97d6f 463 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 464 px = pIn1 + count;
emh203 0:3d9c67d97d6f 465 py = pSrc2;
emh203 0:3d9c67d97d6f 466
emh203 0:3d9c67d97d6f 467
emh203 0:3d9c67d97d6f 468 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 469 blkCnt--;
emh203 0:3d9c67d97d6f 470 }
emh203 0:3d9c67d97d6f 471
emh203 0:3d9c67d97d6f 472 /* If the blockSize2 is not a multiple of 4, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 473 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 474 blkCnt = (uint32_t) blockSize2 % 0x4u;
emh203 0:3d9c67d97d6f 475
emh203 0:3d9c67d97d6f 476 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 477 {
emh203 0:3d9c67d97d6f 478 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 479 sum = 0;
emh203 0:3d9c67d97d6f 480
emh203 0:3d9c67d97d6f 481 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 482 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 483
emh203 0:3d9c67d97d6f 484 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 485 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 486 while(k > 0u)
emh203 0:3d9c67d97d6f 487 {
emh203 0:3d9c67d97d6f 488
emh203 0:3d9c67d97d6f 489 /* Reading two inputs of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 490 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 491 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 492 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 493
emh203 0:3d9c67d97d6f 494 /* Reading two inputs of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 495 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 496 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 497 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 498
emh203 0:3d9c67d97d6f 499 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 500 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 501
emh203 0:3d9c67d97d6f 502 /* Reading two inputs of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 503 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 504 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 505 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 506
emh203 0:3d9c67d97d6f 507 /* Reading two inputs of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 508 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 509 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 510 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 511
emh203 0:3d9c67d97d6f 512 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 513 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 514
emh203 0:3d9c67d97d6f 515 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 516 k--;
emh203 0:3d9c67d97d6f 517 }
emh203 0:3d9c67d97d6f 518
emh203 0:3d9c67d97d6f 519 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 520 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 521 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 522
emh203 0:3d9c67d97d6f 523 while(k > 0u)
emh203 0:3d9c67d97d6f 524 {
emh203 0:3d9c67d97d6f 525 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 526 sum += ((q31_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 527
emh203 0:3d9c67d97d6f 528 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 529 k--;
emh203 0:3d9c67d97d6f 530 }
emh203 0:3d9c67d97d6f 531
emh203 0:3d9c67d97d6f 532 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 533 *pOut++ = (q7_t) (__SSAT(sum >> 7, 8));
emh203 0:3d9c67d97d6f 534
emh203 0:3d9c67d97d6f 535 /* Increment the pointer pIn1 index, count by 1 */
emh203 0:3d9c67d97d6f 536 count++;
emh203 0:3d9c67d97d6f 537
emh203 0:3d9c67d97d6f 538 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 539 px = pIn1 + count;
emh203 0:3d9c67d97d6f 540 py = pSrc2;
emh203 0:3d9c67d97d6f 541
emh203 0:3d9c67d97d6f 542 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 543 blkCnt--;
emh203 0:3d9c67d97d6f 544 }
emh203 0:3d9c67d97d6f 545 }
emh203 0:3d9c67d97d6f 546 else
emh203 0:3d9c67d97d6f 547 {
emh203 0:3d9c67d97d6f 548 /* If the srcBLen is not a multiple of 4,
emh203 0:3d9c67d97d6f 549 * the blockSize2 loop cannot be unrolled by 4 */
emh203 0:3d9c67d97d6f 550 blkCnt = (uint32_t) blockSize2;
emh203 0:3d9c67d97d6f 551
emh203 0:3d9c67d97d6f 552 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 553 {
emh203 0:3d9c67d97d6f 554 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 555 sum = 0;
emh203 0:3d9c67d97d6f 556
emh203 0:3d9c67d97d6f 557 /* srcBLen number of MACS should be performed */
emh203 0:3d9c67d97d6f 558 k = srcBLen;
emh203 0:3d9c67d97d6f 559
emh203 0:3d9c67d97d6f 560 while(k > 0u)
emh203 0:3d9c67d97d6f 561 {
emh203 0:3d9c67d97d6f 562 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 563 sum += ((q31_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 564
emh203 0:3d9c67d97d6f 565 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 566 k--;
emh203 0:3d9c67d97d6f 567 }
emh203 0:3d9c67d97d6f 568
emh203 0:3d9c67d97d6f 569 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 570 *pOut++ = (q7_t) (__SSAT(sum >> 7, 8));
emh203 0:3d9c67d97d6f 571
emh203 0:3d9c67d97d6f 572 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 573 count++;
emh203 0:3d9c67d97d6f 574
emh203 0:3d9c67d97d6f 575 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 576 px = pIn1 + count;
emh203 0:3d9c67d97d6f 577 py = pSrc2;
emh203 0:3d9c67d97d6f 578
emh203 0:3d9c67d97d6f 579 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 580 blkCnt--;
emh203 0:3d9c67d97d6f 581 }
emh203 0:3d9c67d97d6f 582 }
emh203 0:3d9c67d97d6f 583
emh203 0:3d9c67d97d6f 584
emh203 0:3d9c67d97d6f 585 /* --------------------------
emh203 0:3d9c67d97d6f 586 * Initializations of stage3
emh203 0:3d9c67d97d6f 587 * -------------------------*/
emh203 0:3d9c67d97d6f 588
emh203 0:3d9c67d97d6f 589 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emh203 0:3d9c67d97d6f 590 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emh203 0:3d9c67d97d6f 591 * ....
emh203 0:3d9c67d97d6f 592 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emh203 0:3d9c67d97d6f 593 * sum += x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 594 */
emh203 0:3d9c67d97d6f 595
emh203 0:3d9c67d97d6f 596 /* In this stage the MAC operations are decreased by 1 for every iteration.
emh203 0:3d9c67d97d6f 597 The count variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 598 count = srcBLen - 1u;
emh203 0:3d9c67d97d6f 599
emh203 0:3d9c67d97d6f 600 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 601 pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
emh203 0:3d9c67d97d6f 602 px = pSrc1;
emh203 0:3d9c67d97d6f 603
emh203 0:3d9c67d97d6f 604 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 605 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 606 py = pSrc2;
emh203 0:3d9c67d97d6f 607
emh203 0:3d9c67d97d6f 608 /* -------------------
emh203 0:3d9c67d97d6f 609 * Stage3 process
emh203 0:3d9c67d97d6f 610 * ------------------*/
emh203 0:3d9c67d97d6f 611
emh203 0:3d9c67d97d6f 612 while(blockSize3 > 0)
emh203 0:3d9c67d97d6f 613 {
emh203 0:3d9c67d97d6f 614 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 615 sum = 0;
emh203 0:3d9c67d97d6f 616
emh203 0:3d9c67d97d6f 617 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 618 k = count >> 2u;
emh203 0:3d9c67d97d6f 619
emh203 0:3d9c67d97d6f 620 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 621 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 622 while(k > 0u)
emh203 0:3d9c67d97d6f 623 {
emh203 0:3d9c67d97d6f 624 /* Reading two inputs, x[srcALen - srcBLen + 1] and x[srcALen - srcBLen + 2] of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 625 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 626 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 627 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 628
emh203 0:3d9c67d97d6f 629 /* Reading two inputs, y[srcBLen - 1] and y[srcBLen - 2] of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 630 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 631 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 632 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 633
emh203 0:3d9c67d97d6f 634 /* sum += x[srcALen - srcBLen + 1] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 635 /* sum += x[srcALen - srcBLen + 2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 636 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 637
emh203 0:3d9c67d97d6f 638 /* Reading two inputs, x[srcALen - srcBLen + 3] and x[srcALen - srcBLen + 4] of SrcA buffer and packing */
emh203 0:3d9c67d97d6f 639 in1 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 640 in2 = (q15_t) * px++;
emh203 0:3d9c67d97d6f 641 input1 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 642
emh203 0:3d9c67d97d6f 643 /* Reading two inputs, y[srcBLen - 3] and y[srcBLen - 4] of SrcB buffer and packing */
emh203 0:3d9c67d97d6f 644 in1 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 645 in2 = (q15_t) * py--;
emh203 0:3d9c67d97d6f 646 input2 = ((q31_t) in1 & 0x0000FFFF) | ((q31_t) in2 << 16);
emh203 0:3d9c67d97d6f 647
emh203 0:3d9c67d97d6f 648 /* sum += x[srcALen - srcBLen + 3] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 649 /* sum += x[srcALen - srcBLen + 4] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 650 sum = __SMLAD(input1, input2, sum);
emh203 0:3d9c67d97d6f 651
emh203 0:3d9c67d97d6f 652 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 653 k--;
emh203 0:3d9c67d97d6f 654 }
emh203 0:3d9c67d97d6f 655
emh203 0:3d9c67d97d6f 656 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 657 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 658 k = count % 0x4u;
emh203 0:3d9c67d97d6f 659
emh203 0:3d9c67d97d6f 660 while(k > 0u)
emh203 0:3d9c67d97d6f 661 {
emh203 0:3d9c67d97d6f 662 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 663 /* sum += x[srcALen-1] * y[srcBLen-1] */
emh203 0:3d9c67d97d6f 664 sum += ((q31_t) * px++ * *py--);
emh203 0:3d9c67d97d6f 665
emh203 0:3d9c67d97d6f 666 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 667 k--;
emh203 0:3d9c67d97d6f 668 }
emh203 0:3d9c67d97d6f 669
emh203 0:3d9c67d97d6f 670 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 671 *pOut++ = (q7_t) (__SSAT(sum >> 7, 8));
emh203 0:3d9c67d97d6f 672
emh203 0:3d9c67d97d6f 673 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 674 px = ++pSrc1;
emh203 0:3d9c67d97d6f 675 py = pSrc2;
emh203 0:3d9c67d97d6f 676
emh203 0:3d9c67d97d6f 677 /* Decrement the MAC count */
emh203 0:3d9c67d97d6f 678 count--;
emh203 0:3d9c67d97d6f 679
emh203 0:3d9c67d97d6f 680 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 681 blockSize3--;
emh203 0:3d9c67d97d6f 682
emh203 0:3d9c67d97d6f 683 }
emh203 0:3d9c67d97d6f 684
emh203 0:3d9c67d97d6f 685 /* set status as ARM_MATH_SUCCESS */
emh203 0:3d9c67d97d6f 686 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 687 }
emh203 0:3d9c67d97d6f 688
emh203 0:3d9c67d97d6f 689 /* Return to application */
emh203 0:3d9c67d97d6f 690 return (status);
emh203 0:3d9c67d97d6f 691
emh203 0:3d9c67d97d6f 692 #else
emh203 0:3d9c67d97d6f 693
emh203 0:3d9c67d97d6f 694 /* Run the below code for Cortex-M0 */
emh203 0:3d9c67d97d6f 695
emh203 0:3d9c67d97d6f 696 q7_t *pIn1 = pSrcA; /* inputA pointer */
emh203 0:3d9c67d97d6f 697 q7_t *pIn2 = pSrcB; /* inputB pointer */
emh203 0:3d9c67d97d6f 698 q31_t sum; /* Accumulator */
emh203 0:3d9c67d97d6f 699 uint32_t i, j; /* loop counters */
emh203 0:3d9c67d97d6f 700 arm_status status; /* status of Partial convolution */
emh203 0:3d9c67d97d6f 701
emh203 0:3d9c67d97d6f 702 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 703 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 704 {
emh203 0:3d9c67d97d6f 705 /* Set status as ARM_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 706 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 707 }
emh203 0:3d9c67d97d6f 708 else
emh203 0:3d9c67d97d6f 709 {
emh203 0:3d9c67d97d6f 710 /* Loop to calculate convolution for output length number of values */
emh203 0:3d9c67d97d6f 711 for (i = firstIndex; i <= (firstIndex + numPoints - 1); i++)
emh203 0:3d9c67d97d6f 712 {
emh203 0:3d9c67d97d6f 713 /* Initialize sum with zero to carry on MAC operations */
emh203 0:3d9c67d97d6f 714 sum = 0;
emh203 0:3d9c67d97d6f 715
emh203 0:3d9c67d97d6f 716 /* Loop to perform MAC operations according to convolution equation */
emh203 0:3d9c67d97d6f 717 for (j = 0; j <= i; j++)
emh203 0:3d9c67d97d6f 718 {
emh203 0:3d9c67d97d6f 719 /* Check the array limitations */
emh203 0:3d9c67d97d6f 720 if(((i - j) < srcBLen) && (j < srcALen))
emh203 0:3d9c67d97d6f 721 {
emh203 0:3d9c67d97d6f 722 /* z[i] += x[i-j] * y[j] */
emh203 0:3d9c67d97d6f 723 sum += ((q15_t) pIn1[j] * (pIn2[i - j]));
emh203 0:3d9c67d97d6f 724 }
emh203 0:3d9c67d97d6f 725 }
emh203 0:3d9c67d97d6f 726
emh203 0:3d9c67d97d6f 727 /* Store the output in the destination buffer */
emh203 0:3d9c67d97d6f 728 pDst[i] = (q7_t) __SSAT((sum >> 7u), 8u);
emh203 0:3d9c67d97d6f 729 }
emh203 0:3d9c67d97d6f 730 /* set status as ARM_SUCCESS as there are no argument errors */
emh203 0:3d9c67d97d6f 731 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 732 }
emh203 0:3d9c67d97d6f 733 return (status);
emh203 0:3d9c67d97d6f 734
emh203 0:3d9c67d97d6f 735 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emh203 0:3d9c67d97d6f 736
emh203 0:3d9c67d97d6f 737 }
emh203 0:3d9c67d97d6f 738
emh203 0:3d9c67d97d6f 739 /**
emh203 0:3d9c67d97d6f 740 * @} end of PartialConv group
emh203 0:3d9c67d97d6f 741 */