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_fast_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_fast_opt_q15.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Fast Q15 Partial convolution.
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 (fast version) for Cortex-M3 and Cortex-M4.
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 * See <code>arm_conv_partial_q15()</code> for a slower implementation of this function which uses a 64-bit accumulator to avoid wrap around distortion.
emilmont 1:fdd22bb7aa52 66 *
emilmont 1:fdd22bb7aa52 67 * \par Restrictions
emilmont 1:fdd22bb7aa52 68 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 69 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 70 *
emilmont 1:fdd22bb7aa52 71 */
emilmont 1:fdd22bb7aa52 72
emilmont 1:fdd22bb7aa52 73 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 74
emilmont 1:fdd22bb7aa52 75 arm_status arm_conv_partial_fast_opt_q15(
emilmont 1:fdd22bb7aa52 76 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 77 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 78 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 79 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 80 q15_t * pDst,
emilmont 1:fdd22bb7aa52 81 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 82 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 83 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 84 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 85 {
emilmont 1:fdd22bb7aa52 86
emilmont 1:fdd22bb7aa52 87 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 88 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 89 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 90 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 91 q31_t x1, x2, x3; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 92 q31_t y1, y2; /* State variables */
emilmont 1:fdd22bb7aa52 93 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 94 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 95 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 96 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 97 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 98 arm_status status;
emilmont 1:fdd22bb7aa52 99
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
emilmont 1:fdd22bb7aa52 151 /* Copy smaller length input sequence in reverse order into second scratch buffer */
emilmont 1:fdd22bb7aa52 152 while(k > 0u)
emilmont 1:fdd22bb7aa52 153 {
emilmont 1:fdd22bb7aa52 154 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 155 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 156 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 157 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 158 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 159
emilmont 1:fdd22bb7aa52 160 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 161 k--;
emilmont 1:fdd22bb7aa52 162 }
emilmont 1:fdd22bb7aa52 163
emilmont 1:fdd22bb7aa52 164 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 165 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 166 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 167
emilmont 1:fdd22bb7aa52 168 while(k > 0u)
emilmont 1:fdd22bb7aa52 169 {
emilmont 1:fdd22bb7aa52 170 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 171 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 172
emilmont 1:fdd22bb7aa52 173 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 174 k--;
emilmont 1:fdd22bb7aa52 175 }
emilmont 1:fdd22bb7aa52 176
emilmont 1:fdd22bb7aa52 177 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 178 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Assuming scratch1 buffer is aligned by 32-bit */
emilmont 1:fdd22bb7aa52 181 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 182 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 183
emilmont 1:fdd22bb7aa52 184 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 185 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 186
emilmont 1:fdd22bb7aa52 187 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 188
emilmont 1:fdd22bb7aa52 189 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 190 arm_copy_q15(pIn1, pScr1, srcALen);
emilmont 1:fdd22bb7aa52 191
emilmont 1:fdd22bb7aa52 192 /* Update pointers */
emilmont 1:fdd22bb7aa52 193 pScr1 += srcALen;
emilmont 1:fdd22bb7aa52 194
emilmont 1:fdd22bb7aa52 195 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 196 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198 /* Update pointer */
emilmont 1:fdd22bb7aa52 199 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 202 pIn2 = py;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 205
emilmont 1:fdd22bb7aa52 206 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 207
emilmont 1:fdd22bb7aa52 208 /* First part of the processing with loop unrolling process 4 data points at a time.
emilmont 1:fdd22bb7aa52 209 ** a second loop below process for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 210
emilmont 1:fdd22bb7aa52 211 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 212 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 213
emilmont 1:fdd22bb7aa52 214 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 215 {
emilmont 1:fdd22bb7aa52 216 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 217 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 218
emilmont 1:fdd22bb7aa52 219 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 220 acc0 = 0;
emilmont 1:fdd22bb7aa52 221 acc1 = 0;
emilmont 1:fdd22bb7aa52 222 acc2 = 0;
emilmont 1:fdd22bb7aa52 223 acc3 = 0;
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 226 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 229 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 230
emilmont 1:fdd22bb7aa52 231 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 232
emilmont 1:fdd22bb7aa52 233 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 234 {
emilmont 1:fdd22bb7aa52 235
emilmont 1:fdd22bb7aa52 236 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 237 y1 = _SIMD32_OFFSET(pIn2);
emilmont 1:fdd22bb7aa52 238 y2 = _SIMD32_OFFSET(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 239
emilmont 1:fdd22bb7aa52 240 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 241 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 242 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 243
emilmont 1:fdd22bb7aa52 244 /* pack input data */
emilmont 1:fdd22bb7aa52 245 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 246 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 247 #else
emilmont 1:fdd22bb7aa52 248 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 249 #endif
emilmont 1:fdd22bb7aa52 250
emilmont 1:fdd22bb7aa52 251 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 252 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 253
emilmont 1:fdd22bb7aa52 254 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 255 x1 = _SIMD32_OFFSET(pScr1);
emilmont 1:fdd22bb7aa52 256
emilmont 1:fdd22bb7aa52 257 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 258 acc0 = __SMLAD(x2, y2, acc0);
emilmont 1:fdd22bb7aa52 259
emilmont 1:fdd22bb7aa52 260 acc2 = __SMLAD(x1, y2, acc2);
emilmont 1:fdd22bb7aa52 261
emilmont 1:fdd22bb7aa52 262 /* pack input data */
emilmont 1:fdd22bb7aa52 263 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 264 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 265 #else
emilmont 1:fdd22bb7aa52 266 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 267 #endif
emilmont 1:fdd22bb7aa52 268
emilmont 1:fdd22bb7aa52 269 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 270 acc1 = __SMLADX(x3, y2, acc1);
emilmont 1:fdd22bb7aa52 271
emilmont 1:fdd22bb7aa52 272 x2 = _SIMD32_OFFSET(pScr1 + 2u);
emilmont 1:fdd22bb7aa52 273
emilmont 1:fdd22bb7aa52 274 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 275 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 276 #else
emilmont 1:fdd22bb7aa52 277 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 278 #endif
emilmont 1:fdd22bb7aa52 279
emilmont 1:fdd22bb7aa52 280 acc3 = __SMLADX(x3, y2, acc3);
emilmont 1:fdd22bb7aa52 281
emilmont 1:fdd22bb7aa52 282 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 283 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 284 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 285
emilmont 1:fdd22bb7aa52 286
emilmont 1:fdd22bb7aa52 287 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 288 tapCnt--;
emilmont 1:fdd22bb7aa52 289 }
emilmont 1:fdd22bb7aa52 290
emilmont 1:fdd22bb7aa52 291 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 292 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 293
emilmont 1:fdd22bb7aa52 294 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 295 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 296
emilmont 1:fdd22bb7aa52 297 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 298 {
emilmont 1:fdd22bb7aa52 299
emilmont 1:fdd22bb7aa52 300 /* accumlate the results */
emilmont 1:fdd22bb7aa52 301 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 302 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 303 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 304 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 307
emilmont 1:fdd22bb7aa52 308 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 309 tapCnt--;
emilmont 1:fdd22bb7aa52 310 }
emilmont 1:fdd22bb7aa52 311
emilmont 1:fdd22bb7aa52 312 blkCnt--;
emilmont 1:fdd22bb7aa52 313
emilmont 1:fdd22bb7aa52 314
emilmont 1:fdd22bb7aa52 315 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 316
emilmont 1:fdd22bb7aa52 317 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 318
emilmont 1:fdd22bb7aa52 319 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 320 __PKHBT(__SSAT((acc0 >> 15), 16), __SSAT((acc1 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 321 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 322 __PKHBT(__SSAT((acc2 >> 15), 16), __SSAT((acc3 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 323
emilmont 1:fdd22bb7aa52 324 #else
emilmont 1:fdd22bb7aa52 325
emilmont 1:fdd22bb7aa52 326 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 327 __PKHBT(__SSAT((acc1 >> 15), 16), __SSAT((acc0 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 328 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 329 __PKHBT(__SSAT((acc3 >> 15), 16), __SSAT((acc2 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 330
emilmont 1:fdd22bb7aa52 331 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emilmont 1:fdd22bb7aa52 332
emilmont 1:fdd22bb7aa52 333 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 334 pIn2 = py;
emilmont 1:fdd22bb7aa52 335
emilmont 1:fdd22bb7aa52 336 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 337
emilmont 1:fdd22bb7aa52 338 }
emilmont 1:fdd22bb7aa52 339
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 344 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 345 {
emilmont 1:fdd22bb7aa52 346 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 347 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 348
emilmont 1:fdd22bb7aa52 349 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 350 acc0 = 0;
emilmont 1:fdd22bb7aa52 351
emilmont 1:fdd22bb7aa52 352 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 353
emilmont 1:fdd22bb7aa52 354 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 355 {
emilmont 1:fdd22bb7aa52 356
emilmont 1:fdd22bb7aa52 357 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 358 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 361 y1 = *__SIMD32(pIn2)++;
emilmont 1:fdd22bb7aa52 362
emilmont 1:fdd22bb7aa52 363 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 364
emilmont 1:fdd22bb7aa52 365 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 366 tapCnt--;
emilmont 1:fdd22bb7aa52 367 }
emilmont 1:fdd22bb7aa52 368
emilmont 1:fdd22bb7aa52 369 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 370
emilmont 1:fdd22bb7aa52 371 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 372 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 373 {
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 /* accumlate the results */
emilmont 1:fdd22bb7aa52 376 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 379 tapCnt--;
emilmont 1:fdd22bb7aa52 380 }
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382 blkCnt--;
emilmont 1:fdd22bb7aa52 383
emilmont 1:fdd22bb7aa52 384 /* The result is in 2.30 format. Convert to 1.15 with saturation.
emilmont 1:fdd22bb7aa52 385 ** Then store the output in the destination buffer. */
emilmont 1:fdd22bb7aa52 386 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 387
emilmont 1:fdd22bb7aa52 388 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 389 pIn2 = py;
emilmont 1:fdd22bb7aa52 390
emilmont 1:fdd22bb7aa52 391 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 392
emilmont 1:fdd22bb7aa52 393 }
emilmont 1:fdd22bb7aa52 394 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 395 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 396 }
emilmont 1:fdd22bb7aa52 397 /* Return to application */
emilmont 1:fdd22bb7aa52 398 return (status);
emilmont 1:fdd22bb7aa52 399 }
emilmont 1:fdd22bb7aa52 400
emilmont 1:fdd22bb7aa52 401 #else
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 arm_status arm_conv_partial_fast_opt_q15(
emilmont 1:fdd22bb7aa52 404 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 405 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 406 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 407 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 408 q15_t * pDst,
emilmont 1:fdd22bb7aa52 409 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 410 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 411 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 412 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 413 {
emilmont 1:fdd22bb7aa52 414
emilmont 1:fdd22bb7aa52 415 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 416 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 417 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 418 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 419 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 420 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 421 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 422 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 423 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 424 arm_status status; /* Status variable */
emilmont 1:fdd22bb7aa52 425 uint32_t tapCnt; /* loop count */
emilmont 1:fdd22bb7aa52 426 q15_t x10, x11, x20, x21; /* Temporary variables to hold srcA buffer */
emilmont 1:fdd22bb7aa52 427 q15_t y10, y11; /* Temporary variables to hold srcB buffer */
emilmont 1:fdd22bb7aa52 428
emilmont 1:fdd22bb7aa52 429
emilmont 1:fdd22bb7aa52 430 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 431 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 432 {
emilmont 1:fdd22bb7aa52 433 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 434 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 435 }
emilmont 1:fdd22bb7aa52 436 else
emilmont 1:fdd22bb7aa52 437 {
emilmont 1:fdd22bb7aa52 438
emilmont 1:fdd22bb7aa52 439 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 440 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 441 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 442 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 443 {
emilmont 1:fdd22bb7aa52 444 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 445 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 446
emilmont 1:fdd22bb7aa52 447 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 448 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 449 }
emilmont 1:fdd22bb7aa52 450 else
emilmont 1:fdd22bb7aa52 451 {
emilmont 1:fdd22bb7aa52 452 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 453 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 454
emilmont 1:fdd22bb7aa52 455 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 456 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 457
emilmont 1:fdd22bb7aa52 458 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 459 j = srcBLen;
emilmont 1:fdd22bb7aa52 460 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 461 srcALen = j;
emilmont 1:fdd22bb7aa52 462 }
emilmont 1:fdd22bb7aa52 463
emilmont 1:fdd22bb7aa52 464 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 465 py = pScratch2;
emilmont 1:fdd22bb7aa52 466
emilmont 1:fdd22bb7aa52 467 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 468 pScr2 = pScratch2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 469
emilmont 1:fdd22bb7aa52 470 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 471 px = pIn2;
emilmont 1:fdd22bb7aa52 472
emilmont 1:fdd22bb7aa52 473 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 474 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 475
emilmont 1:fdd22bb7aa52 476 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 477 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 478 while(k > 0u)
emilmont 1:fdd22bb7aa52 479 {
emilmont 1:fdd22bb7aa52 480 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 481 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 482 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 483 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 484 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 485
emilmont 1:fdd22bb7aa52 486 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 487 k--;
emilmont 1:fdd22bb7aa52 488 }
emilmont 1:fdd22bb7aa52 489
emilmont 1:fdd22bb7aa52 490 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 491 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 492 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 493
emilmont 1:fdd22bb7aa52 494 while(k > 0u)
emilmont 1:fdd22bb7aa52 495 {
emilmont 1:fdd22bb7aa52 496 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 497 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 498
emilmont 1:fdd22bb7aa52 499 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 500 k--;
emilmont 1:fdd22bb7aa52 501 }
emilmont 1:fdd22bb7aa52 502
emilmont 1:fdd22bb7aa52 503 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 504 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 505
emilmont 1:fdd22bb7aa52 506 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 507 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 508
emilmont 1:fdd22bb7aa52 509 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 510 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 511
emilmont 1:fdd22bb7aa52 512 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 513
emilmont 1:fdd22bb7aa52 514
emilmont 1:fdd22bb7aa52 515 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 516 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 517
emilmont 1:fdd22bb7aa52 518 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 519 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 520 while(k > 0u)
emilmont 1:fdd22bb7aa52 521 {
emilmont 1:fdd22bb7aa52 522 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 523 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 524 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 525 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 526 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 527
emilmont 1:fdd22bb7aa52 528 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 529 k--;
emilmont 1:fdd22bb7aa52 530 }
emilmont 1:fdd22bb7aa52 531
emilmont 1:fdd22bb7aa52 532 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 533 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 534 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 535
emilmont 1:fdd22bb7aa52 536 while(k > 0u)
emilmont 1:fdd22bb7aa52 537 {
emilmont 1:fdd22bb7aa52 538 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 539 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 540
emilmont 1:fdd22bb7aa52 541 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 542 k--;
emilmont 1:fdd22bb7aa52 543 }
emilmont 1:fdd22bb7aa52 544
emilmont 1:fdd22bb7aa52 545
emilmont 1:fdd22bb7aa52 546 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 547 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 548
emilmont 1:fdd22bb7aa52 549 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 550 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 551 while(k > 0u)
emilmont 1:fdd22bb7aa52 552 {
emilmont 1:fdd22bb7aa52 553 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 554 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 555 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 556 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 557 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 558
emilmont 1:fdd22bb7aa52 559 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 560 k--;
emilmont 1:fdd22bb7aa52 561 }
emilmont 1:fdd22bb7aa52 562
emilmont 1:fdd22bb7aa52 563 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 564 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 565 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 566
emilmont 1:fdd22bb7aa52 567 while(k > 0u)
emilmont 1:fdd22bb7aa52 568 {
emilmont 1:fdd22bb7aa52 569 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 570 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 571
emilmont 1:fdd22bb7aa52 572 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 573 k--;
emilmont 1:fdd22bb7aa52 574 }
emilmont 1:fdd22bb7aa52 575
emilmont 1:fdd22bb7aa52 576
emilmont 1:fdd22bb7aa52 577 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 578 pIn2 = py;
emilmont 1:fdd22bb7aa52 579
emilmont 1:fdd22bb7aa52 580 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 581
emilmont 1:fdd22bb7aa52 582 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 583
emilmont 1:fdd22bb7aa52 584 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 585 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 586
emilmont 1:fdd22bb7aa52 587 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 588 {
emilmont 1:fdd22bb7aa52 589 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 590 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 591
emilmont 1:fdd22bb7aa52 592 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 593 acc0 = 0;
emilmont 1:fdd22bb7aa52 594 acc1 = 0;
emilmont 1:fdd22bb7aa52 595 acc2 = 0;
emilmont 1:fdd22bb7aa52 596 acc3 = 0;
emilmont 1:fdd22bb7aa52 597
emilmont 1:fdd22bb7aa52 598 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 599 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 600 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 601
emilmont 1:fdd22bb7aa52 602 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 603 x20 = *pScr1++;
emilmont 1:fdd22bb7aa52 604 x21 = *pScr1++;
emilmont 1:fdd22bb7aa52 605
emilmont 1:fdd22bb7aa52 606 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 607
emilmont 1:fdd22bb7aa52 608 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 609 {
emilmont 1:fdd22bb7aa52 610
emilmont 1:fdd22bb7aa52 611 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 612 y10 = *pIn2;
emilmont 1:fdd22bb7aa52 613 y11 = *(pIn2 + 1u);
emilmont 1:fdd22bb7aa52 614
emilmont 1:fdd22bb7aa52 615 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 616 acc0 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 617 acc0 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 618 acc2 += (q31_t) x20 *y10;
emilmont 1:fdd22bb7aa52 619 acc2 += (q31_t) x21 *y11;
emilmont 1:fdd22bb7aa52 620
emilmont 1:fdd22bb7aa52 621 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 622 acc1 += (q31_t) x11 *y10;
emilmont 1:fdd22bb7aa52 623 acc1 += (q31_t) x20 *y11;
emilmont 1:fdd22bb7aa52 624
emilmont 1:fdd22bb7aa52 625 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 626 x10 = *pScr1;
emilmont 1:fdd22bb7aa52 627 x11 = *(pScr1 + 1u);
emilmont 1:fdd22bb7aa52 628
emilmont 1:fdd22bb7aa52 629 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 630 acc3 += (q31_t) x21 *y10;
emilmont 1:fdd22bb7aa52 631 acc3 += (q31_t) x10 *y11;
emilmont 1:fdd22bb7aa52 632
emilmont 1:fdd22bb7aa52 633 /* Read next two samples from scratch2 buffer */
emilmont 1:fdd22bb7aa52 634 y10 = *(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 635 y11 = *(pIn2 + 3u);
emilmont 1:fdd22bb7aa52 636
emilmont 1:fdd22bb7aa52 637 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 638 acc0 += (q31_t) x20 *y10;
emilmont 1:fdd22bb7aa52 639 acc0 += (q31_t) x21 *y11;
emilmont 1:fdd22bb7aa52 640 acc2 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 641 acc2 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 642 acc1 += (q31_t) x21 *y10;
emilmont 1:fdd22bb7aa52 643 acc1 += (q31_t) x10 *y11;
emilmont 1:fdd22bb7aa52 644
emilmont 1:fdd22bb7aa52 645 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 646 x20 = *(pScr1 + 2);
emilmont 1:fdd22bb7aa52 647 x21 = *(pScr1 + 3);
emilmont 1:fdd22bb7aa52 648
emilmont 1:fdd22bb7aa52 649 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 650 acc3 += (q31_t) x11 *y10;
emilmont 1:fdd22bb7aa52 651 acc3 += (q31_t) x20 *y11;
emilmont 1:fdd22bb7aa52 652
emilmont 1:fdd22bb7aa52 653 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 654 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 655 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 656
emilmont 1:fdd22bb7aa52 657 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 658 tapCnt--;
emilmont 1:fdd22bb7aa52 659 }
emilmont 1:fdd22bb7aa52 660
emilmont 1:fdd22bb7aa52 661 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 662 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 663
emilmont 1:fdd22bb7aa52 664 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 665 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 666
emilmont 1:fdd22bb7aa52 667 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 668 {
emilmont 1:fdd22bb7aa52 669 /* accumlate the results */
emilmont 1:fdd22bb7aa52 670 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 671 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 672 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 673 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 674
emilmont 1:fdd22bb7aa52 675 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 676
emilmont 1:fdd22bb7aa52 677 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 678 tapCnt--;
emilmont 1:fdd22bb7aa52 679 }
emilmont 1:fdd22bb7aa52 680
emilmont 1:fdd22bb7aa52 681 blkCnt--;
emilmont 1:fdd22bb7aa52 682
emilmont 1:fdd22bb7aa52 683
emilmont 1:fdd22bb7aa52 684 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 685 *pOut++ = __SSAT((acc0 >> 15), 16);
emilmont 1:fdd22bb7aa52 686 *pOut++ = __SSAT((acc1 >> 15), 16);
emilmont 1:fdd22bb7aa52 687 *pOut++ = __SSAT((acc2 >> 15), 16);
emilmont 1:fdd22bb7aa52 688 *pOut++ = __SSAT((acc3 >> 15), 16);
emilmont 1:fdd22bb7aa52 689
emilmont 1:fdd22bb7aa52 690 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 691 pIn2 = py;
emilmont 1:fdd22bb7aa52 692
emilmont 1:fdd22bb7aa52 693 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 694
emilmont 1:fdd22bb7aa52 695 }
emilmont 1:fdd22bb7aa52 696
emilmont 1:fdd22bb7aa52 697
emilmont 1:fdd22bb7aa52 698 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 699
emilmont 1:fdd22bb7aa52 700 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 701 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 702 {
emilmont 1:fdd22bb7aa52 703 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 704 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 705
emilmont 1:fdd22bb7aa52 706 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 707 acc0 = 0;
emilmont 1:fdd22bb7aa52 708
emilmont 1:fdd22bb7aa52 709 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 710
emilmont 1:fdd22bb7aa52 711 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 712 {
emilmont 1:fdd22bb7aa52 713
emilmont 1:fdd22bb7aa52 714 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 715 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 716 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 717
emilmont 1:fdd22bb7aa52 718 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 719 y10 = *pIn2++;
emilmont 1:fdd22bb7aa52 720 y11 = *pIn2++;
emilmont 1:fdd22bb7aa52 721
emilmont 1:fdd22bb7aa52 722 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 723 acc0 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 724 acc0 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 725
emilmont 1:fdd22bb7aa52 726 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 727 tapCnt--;
emilmont 1:fdd22bb7aa52 728 }
emilmont 1:fdd22bb7aa52 729
emilmont 1:fdd22bb7aa52 730 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 731
emilmont 1:fdd22bb7aa52 732 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 733 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 734 {
emilmont 1:fdd22bb7aa52 735
emilmont 1:fdd22bb7aa52 736 /* accumlate the results */
emilmont 1:fdd22bb7aa52 737 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 738
emilmont 1:fdd22bb7aa52 739 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 740 tapCnt--;
emilmont 1:fdd22bb7aa52 741 }
emilmont 1:fdd22bb7aa52 742
emilmont 1:fdd22bb7aa52 743 blkCnt--;
emilmont 1:fdd22bb7aa52 744
emilmont 1:fdd22bb7aa52 745 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 746 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 747
emilmont 1:fdd22bb7aa52 748 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 749 pIn2 = py;
emilmont 1:fdd22bb7aa52 750
emilmont 1:fdd22bb7aa52 751 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 752
emilmont 1:fdd22bb7aa52 753 }
emilmont 1:fdd22bb7aa52 754
emilmont 1:fdd22bb7aa52 755 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 756 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 757
emilmont 1:fdd22bb7aa52 758 }
emilmont 1:fdd22bb7aa52 759
emilmont 1:fdd22bb7aa52 760 /* Return to application */
emilmont 1:fdd22bb7aa52 761 return (status);
emilmont 1:fdd22bb7aa52 762 }
emilmont 1:fdd22bb7aa52 763
emilmont 2:da51fb522205 764 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 765
emilmont 1:fdd22bb7aa52 766 /**
emilmont 1:fdd22bb7aa52 767 * @} end of PartialConv group
emilmont 1:fdd22bb7aa52 768 */