mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

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

cmsis_dsp/FilteringFunctions/arm_conv_partial_opt_q15.c@5:3762170b6d4d, 2015-11-20 (annotated)

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



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



Added option to build rpc library. closes #1426

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
mbed_official 5:3762170b6d4d 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
mbed_official 5:3762170b6d4d 4 * $Date: 19. March 2015
mbed_official 5:3762170b6d4d 5 * $Revision: V.1.4.5
emilmont 1:fdd22bb7aa52 6 *
emilmont 2:da51fb522205 7 * Project: CMSIS DSP Library
emilmont 2:da51fb522205 8 * Title: arm_conv_partial_opt_q15.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Partial convolution of Q15 sequences.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3
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 PartialConv
emilmont 1:fdd22bb7aa52 49 * @{
emilmont 1:fdd22bb7aa52 50 */
emilmont 1:fdd22bb7aa52 51
emilmont 1:fdd22bb7aa52 52 /**
emilmont 1:fdd22bb7aa52 53 * @brief Partial convolution of Q15 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.
emilmont 1:fdd22bb7aa52 59 * @param[in] firstIndex is the first output sample to start with.
emilmont 1:fdd22bb7aa52 60 * @param[in] numPoints is the number of output points to be computed.
emilmont 1:fdd22bb7aa52 61 * @param[in] *pScratch1 points to scratch buffer of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 62 * @param[in] *pScratch2 points to scratch buffer of size min(srcALen, srcBLen).
emilmont 1:fdd22bb7aa52 63 * @return Returns either ARM_MATH_SUCCESS if the function completed correctly or ARM_MATH_ARGUMENT_ERROR if the requested subset is not in the range [0 srcALen+srcBLen-2].
emilmont 1:fdd22bb7aa52 64 *
emilmont 1:fdd22bb7aa52 65 * \par Restrictions
emilmont 1:fdd22bb7aa52 66 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 67 * In this case input, output, state buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 68 *
emilmont 1:fdd22bb7aa52 69 * Refer to <code>arm_conv_partial_fast_q15()</code> for a faster but less precise version of this function for Cortex-M3 and Cortex-M4.
emilmont 1:fdd22bb7aa52 70 *
emilmont 1:fdd22bb7aa52 71 *
emilmont 1:fdd22bb7aa52 72 */
emilmont 1:fdd22bb7aa52 73
emilmont 1:fdd22bb7aa52 74 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 75
emilmont 1:fdd22bb7aa52 76 arm_status arm_conv_partial_opt_q15(
emilmont 1:fdd22bb7aa52 77 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 78 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 79 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 80 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 81 q15_t * pDst,
emilmont 1:fdd22bb7aa52 82 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 83 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 84 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 85 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 86 {
emilmont 1:fdd22bb7aa52 87
emilmont 1:fdd22bb7aa52 88 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 89 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 90 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 91 q63_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 92 q31_t x1, x2, x3; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 93 q31_t y1, y2; /* State variables */
emilmont 1:fdd22bb7aa52 94 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 95 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 96 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 97 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 98 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 99 arm_status status; /* Status variable */
emilmont 1:fdd22bb7aa52 100 uint32_t tapCnt; /* loop count */
emilmont 1:fdd22bb7aa52 101
emilmont 1:fdd22bb7aa52 102 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 103 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 104 {
emilmont 1:fdd22bb7aa52 105 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 106 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 107 }
emilmont 1:fdd22bb7aa52 108 else
emilmont 1:fdd22bb7aa52 109 {
emilmont 1:fdd22bb7aa52 110
emilmont 1:fdd22bb7aa52 111 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 112 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 113 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 114 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 115 {
emilmont 1:fdd22bb7aa52 116 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 117 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 118
emilmont 1:fdd22bb7aa52 119 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 120 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 121 }
emilmont 1:fdd22bb7aa52 122 else
emilmont 1:fdd22bb7aa52 123 {
emilmont 1:fdd22bb7aa52 124 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 125 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 126
emilmont 1:fdd22bb7aa52 127 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 128 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 129
emilmont 1:fdd22bb7aa52 130 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 131 j = srcBLen;
emilmont 1:fdd22bb7aa52 132 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 133 srcALen = j;
emilmont 1:fdd22bb7aa52 134 }
emilmont 1:fdd22bb7aa52 135
emilmont 1:fdd22bb7aa52 136 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 137 py = pScratch2;
emilmont 1:fdd22bb7aa52 138
emilmont 1:fdd22bb7aa52 139 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 140 pScr2 = pScratch2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 141
emilmont 1:fdd22bb7aa52 142 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 143 px = pIn2;
emilmont 1:fdd22bb7aa52 144
emilmont 1:fdd22bb7aa52 145 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 146 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 149 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 150 while(k > 0u)
emilmont 1:fdd22bb7aa52 151 {
emilmont 1:fdd22bb7aa52 152 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 153 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 154 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 155 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 156 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 157
emilmont 1:fdd22bb7aa52 158 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 159 k--;
emilmont 1:fdd22bb7aa52 160 }
emilmont 1:fdd22bb7aa52 161
emilmont 1:fdd22bb7aa52 162 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 163 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 164 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 165
emilmont 1:fdd22bb7aa52 166 while(k > 0u)
emilmont 1:fdd22bb7aa52 167 {
emilmont 1:fdd22bb7aa52 168 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 169 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 170
emilmont 1:fdd22bb7aa52 171 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 172 k--;
emilmont 1:fdd22bb7aa52 173 }
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 176 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 177
emilmont 1:fdd22bb7aa52 178 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 179 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 180
emilmont 1:fdd22bb7aa52 181 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 182 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 183
emilmont 1:fdd22bb7aa52 184 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 185
emilmont 1:fdd22bb7aa52 186 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 187 arm_copy_q15(pIn1, pScr1, srcALen);
emilmont 1:fdd22bb7aa52 188
emilmont 1:fdd22bb7aa52 189 /* Update pointers */
emilmont 1:fdd22bb7aa52 190 pScr1 += srcALen;
emilmont 1:fdd22bb7aa52 191
emilmont 1:fdd22bb7aa52 192 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 193 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 194
emilmont 1:fdd22bb7aa52 195 /* Update pointer */
emilmont 1:fdd22bb7aa52 196 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 199 pIn2 = py;
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 202
emilmont 1:fdd22bb7aa52 203 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 204
emilmont 1:fdd22bb7aa52 205 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 206 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 207
emilmont 1:fdd22bb7aa52 208 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 209 {
emilmont 1:fdd22bb7aa52 210 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 211 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 212
emilmont 1:fdd22bb7aa52 213 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 214 acc0 = 0;
emilmont 1:fdd22bb7aa52 215 acc1 = 0;
emilmont 1:fdd22bb7aa52 216 acc2 = 0;
emilmont 1:fdd22bb7aa52 217 acc3 = 0;
emilmont 1:fdd22bb7aa52 218
emilmont 1:fdd22bb7aa52 219 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 220 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 221
emilmont 1:fdd22bb7aa52 222 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 223 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 226
emilmont 1:fdd22bb7aa52 227 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 228 {
emilmont 1:fdd22bb7aa52 229
emilmont 1:fdd22bb7aa52 230 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 231 y1 = _SIMD32_OFFSET(pIn2);
emilmont 1:fdd22bb7aa52 232 y2 = _SIMD32_OFFSET(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 235 acc0 = __SMLALD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 236 acc2 = __SMLALD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* pack input data */
emilmont 1:fdd22bb7aa52 239 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 240 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 241 #else
emilmont 1:fdd22bb7aa52 242 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 243 #endif
emilmont 1:fdd22bb7aa52 244
emilmont 1:fdd22bb7aa52 245 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 246 acc1 = __SMLALDX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 247
emilmont 1:fdd22bb7aa52 248 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 249 x1 = _SIMD32_OFFSET(pScr1);
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 252 acc0 = __SMLALD(x2, y2, acc0);
emilmont 1:fdd22bb7aa52 253 acc2 = __SMLALD(x1, y2, acc2);
emilmont 1:fdd22bb7aa52 254
emilmont 1:fdd22bb7aa52 255 /* pack input data */
emilmont 1:fdd22bb7aa52 256 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 257 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 258 #else
emilmont 1:fdd22bb7aa52 259 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 260 #endif
emilmont 1:fdd22bb7aa52 261
emilmont 1:fdd22bb7aa52 262 acc3 = __SMLALDX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 263 acc1 = __SMLALDX(x3, y2, acc1);
emilmont 1:fdd22bb7aa52 264
emilmont 1:fdd22bb7aa52 265 x2 = _SIMD32_OFFSET(pScr1 + 2u);
emilmont 1:fdd22bb7aa52 266
emilmont 1:fdd22bb7aa52 267 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 268 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 269 #else
emilmont 1:fdd22bb7aa52 270 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 271 #endif
emilmont 1:fdd22bb7aa52 272
emilmont 1:fdd22bb7aa52 273 acc3 = __SMLALDX(x3, y2, acc3);
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 276 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 277 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 278
emilmont 1:fdd22bb7aa52 279
emilmont 1:fdd22bb7aa52 280 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 281 tapCnt--;
emilmont 1:fdd22bb7aa52 282 }
emilmont 1:fdd22bb7aa52 283
emilmont 1:fdd22bb7aa52 284 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 285 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 286
emilmont 1:fdd22bb7aa52 287 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 288 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 289
emilmont 1:fdd22bb7aa52 290 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 291 {
emilmont 1:fdd22bb7aa52 292 /* accumlate the results */
emilmont 1:fdd22bb7aa52 293 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 294 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 295 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 296 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 297
emilmont 1:fdd22bb7aa52 298 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 299
emilmont 1:fdd22bb7aa52 300 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 301 tapCnt--;
emilmont 1:fdd22bb7aa52 302 }
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 blkCnt--;
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306
emilmont 1:fdd22bb7aa52 307 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 308
emilmont 1:fdd22bb7aa52 309 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 310
emilmont 1:fdd22bb7aa52 311 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 312 __PKHBT(__SSAT((acc0 >> 15), 16), __SSAT((acc1 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 313 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 314 __PKHBT(__SSAT((acc2 >> 15), 16), __SSAT((acc3 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 315
emilmont 1:fdd22bb7aa52 316 #else
emilmont 1:fdd22bb7aa52 317
emilmont 1:fdd22bb7aa52 318 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 319 __PKHBT(__SSAT((acc1 >> 15), 16), __SSAT((acc0 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 320 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 321 __PKHBT(__SSAT((acc3 >> 15), 16), __SSAT((acc2 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 322
emilmont 1:fdd22bb7aa52 323 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emilmont 1:fdd22bb7aa52 324
emilmont 1:fdd22bb7aa52 325 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 326 pIn2 = py;
emilmont 1:fdd22bb7aa52 327
emilmont 1:fdd22bb7aa52 328 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 329
emilmont 1:fdd22bb7aa52 330 }
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332
emilmont 1:fdd22bb7aa52 333 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 334
emilmont 1:fdd22bb7aa52 335 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 336 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 337 {
emilmont 1:fdd22bb7aa52 338 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 339 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 342 acc0 = 0;
emilmont 1:fdd22bb7aa52 343
emilmont 1:fdd22bb7aa52 344 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 347 {
emilmont 1:fdd22bb7aa52 348
emilmont 1:fdd22bb7aa52 349 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 350 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 351
emilmont 1:fdd22bb7aa52 352 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 353 y1 = *__SIMD32(pIn2)++;
emilmont 1:fdd22bb7aa52 354
emilmont 1:fdd22bb7aa52 355 acc0 = __SMLALD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 356
emilmont 1:fdd22bb7aa52 357 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 358 tapCnt--;
emilmont 1:fdd22bb7aa52 359 }
emilmont 1:fdd22bb7aa52 360
emilmont 1:fdd22bb7aa52 361 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 362
emilmont 1:fdd22bb7aa52 363 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 364 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 365 {
emilmont 1:fdd22bb7aa52 366
emilmont 1:fdd22bb7aa52 367 /* accumlate the results */
emilmont 1:fdd22bb7aa52 368 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 369
emilmont 1:fdd22bb7aa52 370 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 371 tapCnt--;
emilmont 1:fdd22bb7aa52 372 }
emilmont 1:fdd22bb7aa52 373
emilmont 1:fdd22bb7aa52 374 blkCnt--;
emilmont 1:fdd22bb7aa52 375
emilmont 1:fdd22bb7aa52 376 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 377 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 378
emilmont 1:fdd22bb7aa52 379 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 380 pIn2 = py;
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 383
emilmont 1:fdd22bb7aa52 384 }
emilmont 1:fdd22bb7aa52 385
emilmont 1:fdd22bb7aa52 386 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 387 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 388
emilmont 1:fdd22bb7aa52 389 }
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391 /* Return to application */
emilmont 1:fdd22bb7aa52 392 return (status);
emilmont 1:fdd22bb7aa52 393 }
emilmont 1:fdd22bb7aa52 394
emilmont 1:fdd22bb7aa52 395 #else
emilmont 1:fdd22bb7aa52 396
emilmont 1:fdd22bb7aa52 397 arm_status arm_conv_partial_opt_q15(
emilmont 1:fdd22bb7aa52 398 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 399 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 400 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 401 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 402 q15_t * pDst,
emilmont 1:fdd22bb7aa52 403 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 404 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 405 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 406 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 407 {
emilmont 1:fdd22bb7aa52 408
emilmont 1:fdd22bb7aa52 409 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 410 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 411 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 412 q63_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 413 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 414 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 415 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 416 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 417 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 418 arm_status status; /* Status variable */
emilmont 1:fdd22bb7aa52 419 uint32_t tapCnt; /* loop count */
emilmont 1:fdd22bb7aa52 420 q15_t x10, x11, x20, x21; /* Temporary variables to hold srcA buffer */
emilmont 1:fdd22bb7aa52 421 q15_t y10, y11; /* Temporary variables to hold srcB buffer */
emilmont 1:fdd22bb7aa52 422
emilmont 1:fdd22bb7aa52 423
emilmont 1:fdd22bb7aa52 424 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 425 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 426 {
emilmont 1:fdd22bb7aa52 427 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 428 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 429 }
emilmont 1:fdd22bb7aa52 430 else
emilmont 1:fdd22bb7aa52 431 {
emilmont 1:fdd22bb7aa52 432
emilmont 1:fdd22bb7aa52 433 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 434 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 435 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 436 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 437 {
emilmont 1:fdd22bb7aa52 438 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 439 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 440
emilmont 1:fdd22bb7aa52 441 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 442 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 443 }
emilmont 1:fdd22bb7aa52 444 else
emilmont 1:fdd22bb7aa52 445 {
emilmont 1:fdd22bb7aa52 446 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 447 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 448
emilmont 1:fdd22bb7aa52 449 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 450 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 451
emilmont 1:fdd22bb7aa52 452 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 453 j = srcBLen;
emilmont 1:fdd22bb7aa52 454 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 455 srcALen = j;
emilmont 1:fdd22bb7aa52 456 }
emilmont 1:fdd22bb7aa52 457
emilmont 1:fdd22bb7aa52 458 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 459 py = pScratch2;
emilmont 1:fdd22bb7aa52 460
emilmont 1:fdd22bb7aa52 461 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 462 pScr2 = pScratch2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 463
emilmont 1:fdd22bb7aa52 464 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 465 px = pIn2;
emilmont 1:fdd22bb7aa52 466
emilmont 1:fdd22bb7aa52 467 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 468 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 469
emilmont 1:fdd22bb7aa52 470 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 471 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 472 while(k > 0u)
emilmont 1:fdd22bb7aa52 473 {
emilmont 1:fdd22bb7aa52 474 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 475 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 476 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 477 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 478 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 479
emilmont 1:fdd22bb7aa52 480 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 481 k--;
emilmont 1:fdd22bb7aa52 482 }
emilmont 1:fdd22bb7aa52 483
emilmont 1:fdd22bb7aa52 484 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 485 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 486 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 487
emilmont 1:fdd22bb7aa52 488 while(k > 0u)
emilmont 1:fdd22bb7aa52 489 {
emilmont 1:fdd22bb7aa52 490 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 491 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 492
emilmont 1:fdd22bb7aa52 493 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 494 k--;
emilmont 1:fdd22bb7aa52 495 }
emilmont 1:fdd22bb7aa52 496
emilmont 1:fdd22bb7aa52 497 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 498 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 501 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 502
emilmont 1:fdd22bb7aa52 503 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 504 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 505
emilmont 1:fdd22bb7aa52 506 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 507
emilmont 1:fdd22bb7aa52 508
emilmont 1:fdd22bb7aa52 509 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 510 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 511
emilmont 1:fdd22bb7aa52 512 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 513 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 514 while(k > 0u)
emilmont 1:fdd22bb7aa52 515 {
emilmont 1:fdd22bb7aa52 516 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 517 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 518 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 519 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 520 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 521
emilmont 1:fdd22bb7aa52 522 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 523 k--;
emilmont 1:fdd22bb7aa52 524 }
emilmont 1:fdd22bb7aa52 525
emilmont 1:fdd22bb7aa52 526 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 527 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 528 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 529
emilmont 1:fdd22bb7aa52 530 while(k > 0u)
emilmont 1:fdd22bb7aa52 531 {
emilmont 1:fdd22bb7aa52 532 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 533 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 534
emilmont 1:fdd22bb7aa52 535 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 536 k--;
emilmont 1:fdd22bb7aa52 537 }
emilmont 1:fdd22bb7aa52 538
emilmont 1:fdd22bb7aa52 539
emilmont 1:fdd22bb7aa52 540 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 541 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 542
emilmont 1:fdd22bb7aa52 543 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 544 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 545 while(k > 0u)
emilmont 1:fdd22bb7aa52 546 {
emilmont 1:fdd22bb7aa52 547 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 548 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 549 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 550 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 551 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 552
emilmont 1:fdd22bb7aa52 553 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 554 k--;
emilmont 1:fdd22bb7aa52 555 }
emilmont 1:fdd22bb7aa52 556
emilmont 1:fdd22bb7aa52 557 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 558 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 559 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 560
emilmont 1:fdd22bb7aa52 561 while(k > 0u)
emilmont 1:fdd22bb7aa52 562 {
emilmont 1:fdd22bb7aa52 563 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 564 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 565
emilmont 1:fdd22bb7aa52 566 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 567 k--;
emilmont 1:fdd22bb7aa52 568 }
emilmont 1:fdd22bb7aa52 569
emilmont 1:fdd22bb7aa52 570
emilmont 1:fdd22bb7aa52 571 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 572 pIn2 = py;
emilmont 1:fdd22bb7aa52 573
emilmont 1:fdd22bb7aa52 574 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 575
emilmont 1:fdd22bb7aa52 576 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 577
emilmont 1:fdd22bb7aa52 578 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 579 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 580
emilmont 1:fdd22bb7aa52 581 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 582 {
emilmont 1:fdd22bb7aa52 583 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 584 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 585
emilmont 1:fdd22bb7aa52 586 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 587 acc0 = 0;
emilmont 1:fdd22bb7aa52 588 acc1 = 0;
emilmont 1:fdd22bb7aa52 589 acc2 = 0;
emilmont 1:fdd22bb7aa52 590 acc3 = 0;
emilmont 1:fdd22bb7aa52 591
emilmont 1:fdd22bb7aa52 592 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 593 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 594 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 595
emilmont 1:fdd22bb7aa52 596 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 597 x20 = *pScr1++;
emilmont 1:fdd22bb7aa52 598 x21 = *pScr1++;
emilmont 1:fdd22bb7aa52 599
emilmont 1:fdd22bb7aa52 600 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 601
emilmont 1:fdd22bb7aa52 602 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 603 {
emilmont 1:fdd22bb7aa52 604
emilmont 1:fdd22bb7aa52 605 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 606 y10 = *pIn2;
emilmont 1:fdd22bb7aa52 607 y11 = *(pIn2 + 1u);
emilmont 1:fdd22bb7aa52 608
emilmont 1:fdd22bb7aa52 609 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 610 acc0 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 611 acc0 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 612 acc2 += (q63_t) x20 *y10;
emilmont 1:fdd22bb7aa52 613 acc2 += (q63_t) x21 *y11;
emilmont 1:fdd22bb7aa52 614
emilmont 1:fdd22bb7aa52 615 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 616 acc1 += (q63_t) x11 *y10;
emilmont 1:fdd22bb7aa52 617 acc1 += (q63_t) x20 *y11;
emilmont 1:fdd22bb7aa52 618
emilmont 1:fdd22bb7aa52 619 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 620 x10 = *pScr1;
emilmont 1:fdd22bb7aa52 621 x11 = *(pScr1 + 1u);
emilmont 1:fdd22bb7aa52 622
emilmont 1:fdd22bb7aa52 623 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 624 acc3 += (q63_t) x21 *y10;
emilmont 1:fdd22bb7aa52 625 acc3 += (q63_t) x10 *y11;
emilmont 1:fdd22bb7aa52 626
emilmont 1:fdd22bb7aa52 627 /* Read next two samples from scratch2 buffer */
emilmont 1:fdd22bb7aa52 628 y10 = *(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 629 y11 = *(pIn2 + 3u);
emilmont 1:fdd22bb7aa52 630
emilmont 1:fdd22bb7aa52 631 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 632 acc0 += (q63_t) x20 *y10;
emilmont 1:fdd22bb7aa52 633 acc0 += (q63_t) x21 *y11;
emilmont 1:fdd22bb7aa52 634 acc2 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 635 acc2 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 636 acc1 += (q63_t) x21 *y10;
emilmont 1:fdd22bb7aa52 637 acc1 += (q63_t) x10 *y11;
emilmont 1:fdd22bb7aa52 638
emilmont 1:fdd22bb7aa52 639 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 640 x20 = *(pScr1 + 2);
emilmont 1:fdd22bb7aa52 641 x21 = *(pScr1 + 3);
emilmont 1:fdd22bb7aa52 642
emilmont 1:fdd22bb7aa52 643 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 644 acc3 += (q63_t) x11 *y10;
emilmont 1:fdd22bb7aa52 645 acc3 += (q63_t) x20 *y11;
emilmont 1:fdd22bb7aa52 646
emilmont 1:fdd22bb7aa52 647 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 648 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 649 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 650
emilmont 1:fdd22bb7aa52 651 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 652 tapCnt--;
emilmont 1:fdd22bb7aa52 653 }
emilmont 1:fdd22bb7aa52 654
emilmont 1:fdd22bb7aa52 655 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 656 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 657
emilmont 1:fdd22bb7aa52 658 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 659 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 660
emilmont 1:fdd22bb7aa52 661 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 662 {
emilmont 1:fdd22bb7aa52 663 /* accumlate the results */
emilmont 1:fdd22bb7aa52 664 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 665 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 666 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 667 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 668
emilmont 1:fdd22bb7aa52 669 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 670
emilmont 1:fdd22bb7aa52 671 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 672 tapCnt--;
emilmont 1:fdd22bb7aa52 673 }
emilmont 1:fdd22bb7aa52 674
emilmont 1:fdd22bb7aa52 675 blkCnt--;
emilmont 1:fdd22bb7aa52 676
emilmont 1:fdd22bb7aa52 677
emilmont 1:fdd22bb7aa52 678 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 679 *pOut++ = __SSAT((acc0 >> 15), 16);
emilmont 1:fdd22bb7aa52 680 *pOut++ = __SSAT((acc1 >> 15), 16);
emilmont 1:fdd22bb7aa52 681 *pOut++ = __SSAT((acc2 >> 15), 16);
emilmont 1:fdd22bb7aa52 682 *pOut++ = __SSAT((acc3 >> 15), 16);
emilmont 1:fdd22bb7aa52 683
emilmont 1:fdd22bb7aa52 684
emilmont 1:fdd22bb7aa52 685 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 686 pIn2 = py;
emilmont 1:fdd22bb7aa52 687
emilmont 1:fdd22bb7aa52 688 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 689
emilmont 1:fdd22bb7aa52 690 }
emilmont 1:fdd22bb7aa52 691
emilmont 1:fdd22bb7aa52 692
emilmont 1:fdd22bb7aa52 693 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 694
emilmont 1:fdd22bb7aa52 695 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 696 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 697 {
emilmont 1:fdd22bb7aa52 698 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 699 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 700
emilmont 1:fdd22bb7aa52 701 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 702 acc0 = 0;
emilmont 1:fdd22bb7aa52 703
emilmont 1:fdd22bb7aa52 704 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 705
emilmont 1:fdd22bb7aa52 706 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 707 {
emilmont 1:fdd22bb7aa52 708
emilmont 1:fdd22bb7aa52 709 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 710 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 711 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 712
emilmont 1:fdd22bb7aa52 713 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 714 y10 = *pIn2++;
emilmont 1:fdd22bb7aa52 715 y11 = *pIn2++;
emilmont 1:fdd22bb7aa52 716
emilmont 1:fdd22bb7aa52 717 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 718 acc0 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 719 acc0 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 720
emilmont 1:fdd22bb7aa52 721 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 722 tapCnt--;
emilmont 1:fdd22bb7aa52 723 }
emilmont 1:fdd22bb7aa52 724
emilmont 1:fdd22bb7aa52 725 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 726
emilmont 1:fdd22bb7aa52 727 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 728 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 729 {
emilmont 1:fdd22bb7aa52 730
emilmont 1:fdd22bb7aa52 731 /* accumlate the results */
emilmont 1:fdd22bb7aa52 732 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 733
emilmont 1:fdd22bb7aa52 734 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 735 tapCnt--;
emilmont 1:fdd22bb7aa52 736 }
emilmont 1:fdd22bb7aa52 737
emilmont 1:fdd22bb7aa52 738 blkCnt--;
emilmont 1:fdd22bb7aa52 739
emilmont 1:fdd22bb7aa52 740 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 741 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 742
emilmont 1:fdd22bb7aa52 743
emilmont 1:fdd22bb7aa52 744 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 745 pIn2 = py;
emilmont 1:fdd22bb7aa52 746
emilmont 1:fdd22bb7aa52 747 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 748
emilmont 1:fdd22bb7aa52 749 }
emilmont 1:fdd22bb7aa52 750
emilmont 1:fdd22bb7aa52 751 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 752 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 753
emilmont 1:fdd22bb7aa52 754 }
emilmont 1:fdd22bb7aa52 755
emilmont 1:fdd22bb7aa52 756 /* Return to application */
emilmont 1:fdd22bb7aa52 757 return (status);
emilmont 1:fdd22bb7aa52 758 }
emilmont 1:fdd22bb7aa52 759
emilmont 2:da51fb522205 760 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 761
emilmont 1:fdd22bb7aa52 762
emilmont 1:fdd22bb7aa52 763 /**
emilmont 1:fdd22bb7aa52 764 * @} end of PartialConv group
emilmont 1:fdd22bb7aa52 765 */