mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.

cmsis_dsp/FilteringFunctions/arm_conv_opt_q7.c@5:3762170b6d4d, 2015-11-20 (annotated)

Committer:
mbed_official
Date:
Fri Nov 20 08:45:18 2015 +0000
Revision:
5:3762170b6d4d
Parent:
3:7a284390b0ce 
Synchronized with git revision 2eb940b9a73af188d3004a2575fdfbb05febe62b



Full URL: https://github.com/mbedmicro/mbed/commit/2eb940b9a73af188d3004a2575fdfbb05febe62b/



Added option to build rpc library. closes #1426

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
mbed_official 5:3762170b6d4d 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
mbed_official 5:3762170b6d4d 4 * $Date: 19. March 2015
mbed_official 5:3762170b6d4d 5 * $Revision: V.1.4.5
emilmont 1:fdd22bb7aa52 6 *
emilmont 2:da51fb522205 7 * Project: CMSIS DSP Library
emilmont 2:da51fb522205 8 * Title: arm_conv_opt_q7.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Convolution of Q7 sequences.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emilmont 1:fdd22bb7aa52 13 *
mbed_official 3:7a284390b0ce 14 * Redistribution and use in source and binary forms, with or without
mbed_official 3:7a284390b0ce 15 * modification, are permitted provided that the following conditions
mbed_official 3:7a284390b0ce 16 * are met:
mbed_official 3:7a284390b0ce 17 * - Redistributions of source code must retain the above copyright
mbed_official 3:7a284390b0ce 18 * notice, this list of conditions and the following disclaimer.
mbed_official 3:7a284390b0ce 19 * - Redistributions in binary form must reproduce the above copyright
mbed_official 3:7a284390b0ce 20 * notice, this list of conditions and the following disclaimer in
mbed_official 3:7a284390b0ce 21 * the documentation and/or other materials provided with the
mbed_official 3:7a284390b0ce 22 * distribution.
mbed_official 3:7a284390b0ce 23 * - Neither the name of ARM LIMITED nor the names of its contributors
mbed_official 3:7a284390b0ce 24 * may be used to endorse or promote products derived from this
mbed_official 3:7a284390b0ce 25 * software without specific prior written permission.
mbed_official 3:7a284390b0ce 26 *
mbed_official 3:7a284390b0ce 27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
mbed_official 3:7a284390b0ce 28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
mbed_official 3:7a284390b0ce 29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
mbed_official 3:7a284390b0ce 30 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
mbed_official 3:7a284390b0ce 31 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
mbed_official 3:7a284390b0ce 32 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
mbed_official 3:7a284390b0ce 33 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
mbed_official 3:7a284390b0ce 34 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
mbed_official 3:7a284390b0ce 35 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
mbed_official 3:7a284390b0ce 36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
mbed_official 3:7a284390b0ce 37 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
mbed_official 3:7a284390b0ce 38 * POSSIBILITY OF SUCH DAMAGE.
emilmont 1:fdd22bb7aa52 39 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 40
emilmont 1:fdd22bb7aa52 41 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 42
emilmont 1:fdd22bb7aa52 43 /**
emilmont 1:fdd22bb7aa52 44 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 45 */
emilmont 1:fdd22bb7aa52 46
emilmont 1:fdd22bb7aa52 47 /**
emilmont 1:fdd22bb7aa52 48 * @addtogroup Conv
emilmont 1:fdd22bb7aa52 49 * @{
emilmont 1:fdd22bb7aa52 50 */
emilmont 1:fdd22bb7aa52 51
emilmont 1:fdd22bb7aa52 52 /**
emilmont 1:fdd22bb7aa52 53 * @brief Convolution of Q7 sequences.
emilmont 1:fdd22bb7aa52 54 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 55 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 56 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 57 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 58 * @param[out] *pDst points to the location where the output result is written. Length srcALen+srcBLen-1.
emilmont 1:fdd22bb7aa52 59 * @param[in] *pScratch1 points to scratch buffer(of type q15_t) of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 60 * @param[in] *pScratch2 points to scratch buffer (of type q15_t) of size min(srcALen, srcBLen).
emilmont 1:fdd22bb7aa52 61 * @return none.
emilmont 1:fdd22bb7aa52 62 *
emilmont 1:fdd22bb7aa52 63 * \par Restrictions
emilmont 1:fdd22bb7aa52 64 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 65 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 66 *
emilmont 1:fdd22bb7aa52 67 * @details
emilmont 1:fdd22bb7aa52 68 * <b>Scaling and Overflow Behavior:</b>
emilmont 1:fdd22bb7aa52 69 *
emilmont 1:fdd22bb7aa52 70 * \par
emilmont 1:fdd22bb7aa52 71 * The function is implemented using a 32-bit internal accumulator.
emilmont 1:fdd22bb7aa52 72 * Both the inputs are represented in 1.7 format and multiplications yield a 2.14 result.
emilmont 1:fdd22bb7aa52 73 * The 2.14 intermediate results are accumulated in a 32-bit accumulator in 18.14 format.
emilmont 1:fdd22bb7aa52 74 * This approach provides 17 guard bits and there is no risk of overflow as long as <code>max(srcALen, srcBLen)<131072</code>.
emilmont 1:fdd22bb7aa52 75 * The 18.14 result is then truncated to 18.7 format by discarding the low 7 bits and then saturated to 1.7 format.
emilmont 1:fdd22bb7aa52 76 *
emilmont 1:fdd22bb7aa52 77 */
emilmont 1:fdd22bb7aa52 78
emilmont 1:fdd22bb7aa52 79 void arm_conv_opt_q7(
emilmont 1:fdd22bb7aa52 80 q7_t * pSrcA,
emilmont 1:fdd22bb7aa52 81 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 82 q7_t * pSrcB,
emilmont 1:fdd22bb7aa52 83 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 84 q7_t * pDst,
emilmont 1:fdd22bb7aa52 85 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 86 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 87 {
emilmont 1:fdd22bb7aa52 88
emilmont 1:fdd22bb7aa52 89 q15_t *pScr2, *pScr1; /* Intermediate pointers for scratch pointers */
emilmont 1:fdd22bb7aa52 90 q15_t x4; /* Temporary input variable */
emilmont 1:fdd22bb7aa52 91 q7_t *pIn1, *pIn2; /* inputA and inputB pointer */
emilmont 1:fdd22bb7aa52 92 uint32_t j, k, blkCnt, tapCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 93 q7_t *px; /* Temporary input1 pointer */
emilmont 1:fdd22bb7aa52 94 q15_t *py; /* Temporary input2 pointer */
emilmont 1:fdd22bb7aa52 95 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 96 q31_t x1, x2, x3, y1; /* Temporary input variables */
emilmont 1:fdd22bb7aa52 97 q7_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 98 q7_t out0, out1, out2, out3; /* temporary variables */
emilmont 1:fdd22bb7aa52 99
emilmont 1:fdd22bb7aa52 100 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 101 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 102 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 103 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 104 {
emilmont 1:fdd22bb7aa52 105 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 106 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 107
emilmont 1:fdd22bb7aa52 108 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 109 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 110 }
emilmont 1:fdd22bb7aa52 111 else
emilmont 1:fdd22bb7aa52 112 {
emilmont 1:fdd22bb7aa52 113 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 114 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 115
emilmont 1:fdd22bb7aa52 116 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 117 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 118
emilmont 1:fdd22bb7aa52 119 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 120 j = srcBLen;
emilmont 1:fdd22bb7aa52 121 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 122 srcALen = j;
emilmont 1:fdd22bb7aa52 123 }
emilmont 1:fdd22bb7aa52 124
emilmont 1:fdd22bb7aa52 125 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 126 pScr2 = pScratch2;
emilmont 1:fdd22bb7aa52 127
emilmont 1:fdd22bb7aa52 128 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 129 px = pIn2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 130
emilmont 1:fdd22bb7aa52 131 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 132 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 133
emilmont 1:fdd22bb7aa52 134 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 135 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 136 while(k > 0u)
emilmont 1:fdd22bb7aa52 137 {
emilmont 1:fdd22bb7aa52 138 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 139 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 140 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 141 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 142 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 143 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 144 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 145 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 146 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 149 k--;
emilmont 1:fdd22bb7aa52 150 }
emilmont 1:fdd22bb7aa52 151
emilmont 1:fdd22bb7aa52 152 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 153 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 154 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 155
emilmont 1:fdd22bb7aa52 156 while(k > 0u)
emilmont 1:fdd22bb7aa52 157 {
emilmont 1:fdd22bb7aa52 158 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 159 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 160 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 161
emilmont 1:fdd22bb7aa52 162 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 163 k--;
emilmont 1:fdd22bb7aa52 164 }
emilmont 1:fdd22bb7aa52 165
emilmont 1:fdd22bb7aa52 166 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 167 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 168
emilmont 1:fdd22bb7aa52 169 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 170 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 171
emilmont 1:fdd22bb7aa52 172 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 173 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 176 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 177 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 178
emilmont 1:fdd22bb7aa52 179 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 180 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 181 while(k > 0u)
emilmont 1:fdd22bb7aa52 182 {
emilmont 1:fdd22bb7aa52 183 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 184 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 185 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 186 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 187 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 188 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 189 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 190 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 191 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 192
emilmont 1:fdd22bb7aa52 193 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 194 k--;
emilmont 1:fdd22bb7aa52 195 }
emilmont 1:fdd22bb7aa52 196
emilmont 1:fdd22bb7aa52 197 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 198 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 199 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 while(k > 0u)
emilmont 1:fdd22bb7aa52 202 {
emilmont 1:fdd22bb7aa52 203 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 204 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 205 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 206
emilmont 1:fdd22bb7aa52 207 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 208 k--;
emilmont 1:fdd22bb7aa52 209 }
emilmont 1:fdd22bb7aa52 210
emilmont 1:fdd22bb7aa52 211 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 212
emilmont 1:fdd22bb7aa52 213 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 214 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 215
emilmont 1:fdd22bb7aa52 216 /* Update pointer */
emilmont 1:fdd22bb7aa52 217 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 218
emilmont 1:fdd22bb7aa52 219 #else
emilmont 1:fdd22bb7aa52 220
emilmont 1:fdd22bb7aa52 221 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 222 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 223
emilmont 1:fdd22bb7aa52 224 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 225 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 226 while(k > 0u)
emilmont 1:fdd22bb7aa52 227 {
emilmont 1:fdd22bb7aa52 228 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 229 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 230 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 231 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 232 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 235 k--;
emilmont 1:fdd22bb7aa52 236 }
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 239 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 240 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 while(k > 0u)
emilmont 1:fdd22bb7aa52 243 {
emilmont 1:fdd22bb7aa52 244 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 245 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 246
emilmont 1:fdd22bb7aa52 247 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 248 k--;
emilmont 1:fdd22bb7aa52 249 }
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 #endif
emilmont 1:fdd22bb7aa52 252
emilmont 1:fdd22bb7aa52 253 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 254 py = pScratch2;
emilmont 1:fdd22bb7aa52 255
emilmont 1:fdd22bb7aa52 256 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 257 pIn2 = (q7_t *) py;
emilmont 1:fdd22bb7aa52 258
emilmont 1:fdd22bb7aa52 259 pScr2 = py;
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 262 blkCnt = (srcALen + srcBLen - 1u) >> 2;
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 265 {
emilmont 1:fdd22bb7aa52 266 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 267 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 268
emilmont 1:fdd22bb7aa52 269 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 270 acc0 = 0;
emilmont 1:fdd22bb7aa52 271 acc1 = 0;
emilmont 1:fdd22bb7aa52 272 acc2 = 0;
emilmont 1:fdd22bb7aa52 273 acc3 = 0;
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 276 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 279 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 282
emilmont 1:fdd22bb7aa52 283 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 284 {
emilmont 1:fdd22bb7aa52 285
emilmont 1:fdd22bb7aa52 286 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 287 y1 = _SIMD32_OFFSET(pScr2);
emilmont 1:fdd22bb7aa52 288
emilmont 1:fdd22bb7aa52 289 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 290 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 291 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 292
emilmont 1:fdd22bb7aa52 293 /* pack input data */
emilmont 1:fdd22bb7aa52 294 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 295 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 296 #else
emilmont 1:fdd22bb7aa52 297 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 298 #endif
emilmont 1:fdd22bb7aa52 299
emilmont 1:fdd22bb7aa52 300 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 301 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 302
emilmont 1:fdd22bb7aa52 303 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 304 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 /* pack input data */
emilmont 1:fdd22bb7aa52 307 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 308 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 309 #else
emilmont 1:fdd22bb7aa52 310 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 311 #endif
emilmont 1:fdd22bb7aa52 312
emilmont 1:fdd22bb7aa52 313 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 314
emilmont 1:fdd22bb7aa52 315 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 316 y1 = _SIMD32_OFFSET(pScr2 + 2u);
emilmont 1:fdd22bb7aa52 317
emilmont 1:fdd22bb7aa52 318 acc0 = __SMLAD(x2, y1, acc0);
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320 acc2 = __SMLAD(x1, y1, acc2);
emilmont 1:fdd22bb7aa52 321
emilmont 1:fdd22bb7aa52 322 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 323
emilmont 1:fdd22bb7aa52 324 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 325
emilmont 1:fdd22bb7aa52 326 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 327 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 328 #else
emilmont 1:fdd22bb7aa52 329 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 330 #endif
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 333
emilmont 1:fdd22bb7aa52 334 pScr2 += 4u;
emilmont 1:fdd22bb7aa52 335
emilmont 1:fdd22bb7aa52 336
emilmont 1:fdd22bb7aa52 337 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 338 tapCnt--;
emilmont 1:fdd22bb7aa52 339 }
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 344 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346
emilmont 1:fdd22bb7aa52 347 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 348 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 349
emilmont 1:fdd22bb7aa52 350 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 351 {
emilmont 1:fdd22bb7aa52 352
emilmont 1:fdd22bb7aa52 353 /* accumlate the results */
emilmont 1:fdd22bb7aa52 354 acc0 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 355 acc1 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 356 acc2 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 357 acc3 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 358
emilmont 1:fdd22bb7aa52 359 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 360
emilmont 1:fdd22bb7aa52 361 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 362 tapCnt--;
emilmont 1:fdd22bb7aa52 363 }
emilmont 1:fdd22bb7aa52 364
emilmont 1:fdd22bb7aa52 365 blkCnt--;
emilmont 1:fdd22bb7aa52 366
emilmont 1:fdd22bb7aa52 367 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 368 out0 = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 369 out1 = (q7_t) (__SSAT(acc1 >> 7u, 8));
emilmont 1:fdd22bb7aa52 370 out2 = (q7_t) (__SSAT(acc2 >> 7u, 8));
emilmont 1:fdd22bb7aa52 371 out3 = (q7_t) (__SSAT(acc3 >> 7u, 8));
emilmont 1:fdd22bb7aa52 372
emilmont 1:fdd22bb7aa52 373 *__SIMD32(pOut)++ = __PACKq7(out0, out1, out2, out3);
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 376 pScr2 = py;
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 379
emilmont 1:fdd22bb7aa52 380 }
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382
emilmont 1:fdd22bb7aa52 383 blkCnt = (srcALen + srcBLen - 1u) & 0x3;
emilmont 1:fdd22bb7aa52 384
emilmont 1:fdd22bb7aa52 385 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 386 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 387 {
emilmont 1:fdd22bb7aa52 388 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 389 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 392 acc0 = 0;
emilmont 1:fdd22bb7aa52 393
emilmont 1:fdd22bb7aa52 394 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 395
emilmont 1:fdd22bb7aa52 396 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 397 {
emilmont 1:fdd22bb7aa52 398 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 399 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 400
emilmont 1:fdd22bb7aa52 401 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 402 tapCnt--;
emilmont 1:fdd22bb7aa52 403 }
emilmont 1:fdd22bb7aa52 404
emilmont 1:fdd22bb7aa52 405 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 406
emilmont 1:fdd22bb7aa52 407 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 408 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 409 {
emilmont 1:fdd22bb7aa52 410
emilmont 1:fdd22bb7aa52 411 /* accumlate the results */
emilmont 1:fdd22bb7aa52 412 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 413
emilmont 1:fdd22bb7aa52 414 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 415 tapCnt--;
emilmont 1:fdd22bb7aa52 416 }
emilmont 1:fdd22bb7aa52 417
emilmont 1:fdd22bb7aa52 418 blkCnt--;
emilmont 1:fdd22bb7aa52 419
emilmont 1:fdd22bb7aa52 420 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 421 *pOut++ = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 422
emilmont 1:fdd22bb7aa52 423 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 424 pScr2 = py;
emilmont 1:fdd22bb7aa52 425
emilmont 1:fdd22bb7aa52 426 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 427
emilmont 1:fdd22bb7aa52 428 }
emilmont 1:fdd22bb7aa52 429
emilmont 1:fdd22bb7aa52 430 }
emilmont 1:fdd22bb7aa52 431
emilmont 1:fdd22bb7aa52 432
emilmont 1:fdd22bb7aa52 433 /**
emilmont 1:fdd22bb7aa52 434 * @} end of Conv group
emilmont 1:fdd22bb7aa52 435 */