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_q7.c@2:da51fb522205, 2013-05-30 (annotated)

Committer:
emilmont
Date:
Thu May 30 17:10:11 2013 +0100
Revision:
2:da51fb522205
Parent:
1:fdd22bb7aa52
Child:
3:7a284390b0ce
Keep "cmsis-dsp" module in synch with its source

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 2:da51fb522205 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 2:da51fb522205 7 * Project: CMSIS DSP Library
emilmont 2:da51fb522205 8 * Title: arm_conv_partial_opt_q7.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: Partial convolution of Q7 sequences.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3
emilmont 1:fdd22bb7aa52 13 *
emilmont 1:fdd22bb7aa52 14 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 15 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 16 *
emilmont 1:fdd22bb7aa52 17 * Version 1.0.11 2011/10/18
emilmont 1:fdd22bb7aa52 18 * Bug Fix in conv, correlation, partial convolution.
emilmont 1:fdd22bb7aa52 19 *
emilmont 1:fdd22bb7aa52 20 * Version 1.0.10 2011/7/15
emilmont 1:fdd22bb7aa52 21 * Big Endian support added and Merged M0 and M3/M4 Source code.
emilmont 1:fdd22bb7aa52 22 *
emilmont 1:fdd22bb7aa52 23 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 24 * Re-organized the CMSIS folders and updated documentation.
emilmont 1:fdd22bb7aa52 25 *
emilmont 1:fdd22bb7aa52 26 * Version 1.0.2 2010/11/11
emilmont 1:fdd22bb7aa52 27 * Documentation updated.
emilmont 1:fdd22bb7aa52 28 *
emilmont 1:fdd22bb7aa52 29 * Version 1.0.1 2010/10/05
emilmont 1:fdd22bb7aa52 30 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 31 *
emilmont 1:fdd22bb7aa52 32 * Version 1.0.0 2010/09/20
emilmont 1:fdd22bb7aa52 33 * Production release and review comments incorporated
emilmont 1:fdd22bb7aa52 34 *
emilmont 1:fdd22bb7aa52 35 * Version 0.0.7 2010/06/10
emilmont 1:fdd22bb7aa52 36 * Misra-C changes done
emilmont 1:fdd22bb7aa52 37 *
emilmont 1:fdd22bb7aa52 38 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 39
emilmont 1:fdd22bb7aa52 40 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 41
emilmont 1:fdd22bb7aa52 42 /**
emilmont 1:fdd22bb7aa52 43 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 44 */
emilmont 1:fdd22bb7aa52 45
emilmont 1:fdd22bb7aa52 46 /**
emilmont 1:fdd22bb7aa52 47 * @addtogroup PartialConv
emilmont 1:fdd22bb7aa52 48 * @{
emilmont 1:fdd22bb7aa52 49 */
emilmont 1:fdd22bb7aa52 50
emilmont 1:fdd22bb7aa52 51 /**
emilmont 1:fdd22bb7aa52 52 * @brief Partial convolution of Q7 sequences.
emilmont 1:fdd22bb7aa52 53 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 54 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 55 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 56 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 57 * @param[out] *pDst points to the location where the output result is written.
emilmont 1:fdd22bb7aa52 58 * @param[in] firstIndex is the first output sample to start with.
emilmont 1:fdd22bb7aa52 59 * @param[in] numPoints is the number of output points to be computed.
emilmont 1:fdd22bb7aa52 60 * @param[in] *pScratch1 points to scratch buffer(of type q15_t) of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 61 * @param[in] *pScratch2 points to scratch buffer (of type q15_t) of size min(srcALen, srcBLen).
emilmont 1:fdd22bb7aa52 62 * @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 63 *
emilmont 1:fdd22bb7aa52 64 * \par Restrictions
emilmont 1:fdd22bb7aa52 65 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 2:da51fb522205 66 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 67 *
emilmont 1:fdd22bb7aa52 68 *
emilmont 1:fdd22bb7aa52 69 *
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_opt_q7(
emilmont 1:fdd22bb7aa52 76 q7_t * pSrcA,
emilmont 1:fdd22bb7aa52 77 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 78 q7_t * pSrcB,
emilmont 1:fdd22bb7aa52 79 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 80 q7_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 *pScr2, *pScr1; /* Intermediate pointers for scratch pointers */
emilmont 1:fdd22bb7aa52 88 q15_t x4; /* Temporary input variable */
emilmont 1:fdd22bb7aa52 89 q7_t *pIn1, *pIn2; /* inputA and inputB pointer */
emilmont 1:fdd22bb7aa52 90 uint32_t j, k, blkCnt, tapCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 91 q7_t *px; /* Temporary input1 pointer */
emilmont 1:fdd22bb7aa52 92 q15_t *py; /* Temporary input2 pointer */
emilmont 1:fdd22bb7aa52 93 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 94 q31_t x1, x2, x3, y1; /* Temporary input variables */
emilmont 1:fdd22bb7aa52 95 arm_status status;
emilmont 1:fdd22bb7aa52 96 q7_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 97 q7_t out0, out1, out2, out3; /* temporary variables */
emilmont 1:fdd22bb7aa52 98
emilmont 1:fdd22bb7aa52 99 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 100 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 101 {
emilmont 1:fdd22bb7aa52 102 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 103 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 104 }
emilmont 1:fdd22bb7aa52 105 else
emilmont 1:fdd22bb7aa52 106 {
emilmont 1:fdd22bb7aa52 107
emilmont 1:fdd22bb7aa52 108 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 109 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 110 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 111 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 112 {
emilmont 1:fdd22bb7aa52 113 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 114 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 115
emilmont 1:fdd22bb7aa52 116 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 117 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 118 }
emilmont 1:fdd22bb7aa52 119 else
emilmont 1:fdd22bb7aa52 120 {
emilmont 1:fdd22bb7aa52 121 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 122 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 123
emilmont 1:fdd22bb7aa52 124 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 125 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 126
emilmont 1:fdd22bb7aa52 127 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 128 j = srcBLen;
emilmont 1:fdd22bb7aa52 129 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 130 srcALen = j;
emilmont 1:fdd22bb7aa52 131 }
emilmont 1:fdd22bb7aa52 132
emilmont 1:fdd22bb7aa52 133 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 134 pScr2 = pScratch2;
emilmont 1:fdd22bb7aa52 135
emilmont 1:fdd22bb7aa52 136 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 137 px = pIn2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 138
emilmont 1:fdd22bb7aa52 139 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 140 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 141
emilmont 1:fdd22bb7aa52 142 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 143 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 144 while(k > 0u)
emilmont 1:fdd22bb7aa52 145 {
emilmont 1:fdd22bb7aa52 146 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 147 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 148 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 149 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 150 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 151 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 152 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 153 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 154 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 155
emilmont 1:fdd22bb7aa52 156 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 157 k--;
emilmont 1:fdd22bb7aa52 158 }
emilmont 1:fdd22bb7aa52 159
emilmont 1:fdd22bb7aa52 160 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 161 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 162 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 163
emilmont 1:fdd22bb7aa52 164 while(k > 0u)
emilmont 1:fdd22bb7aa52 165 {
emilmont 1:fdd22bb7aa52 166 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 167 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 168 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 169
emilmont 1:fdd22bb7aa52 170 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 171 k--;
emilmont 1:fdd22bb7aa52 172 }
emilmont 1:fdd22bb7aa52 173
emilmont 1:fdd22bb7aa52 174 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 175 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 176
emilmont 1:fdd22bb7aa52 177 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 178 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 181 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 182
emilmont 1:fdd22bb7aa52 183 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 184 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 185 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 186
emilmont 1:fdd22bb7aa52 187 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 188 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 189 while(k > 0u)
emilmont 1:fdd22bb7aa52 190 {
emilmont 1:fdd22bb7aa52 191 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 192 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 193 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 194 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 195 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 196 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 197 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 198 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 199 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 202 k--;
emilmont 1:fdd22bb7aa52 203 }
emilmont 1:fdd22bb7aa52 204
emilmont 1:fdd22bb7aa52 205 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 206 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 207 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 208
emilmont 1:fdd22bb7aa52 209 while(k > 0u)
emilmont 1:fdd22bb7aa52 210 {
emilmont 1:fdd22bb7aa52 211 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 212 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 213 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 214
emilmont 1:fdd22bb7aa52 215 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 216 k--;
emilmont 1:fdd22bb7aa52 217 }
emilmont 1:fdd22bb7aa52 218
emilmont 1:fdd22bb7aa52 219 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 220 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 221
emilmont 1:fdd22bb7aa52 222 /* Update pointer */
emilmont 1:fdd22bb7aa52 223 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 224
emilmont 1:fdd22bb7aa52 225
emilmont 1:fdd22bb7aa52 226 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 227 py = pScratch2;
emilmont 1:fdd22bb7aa52 228
emilmont 1:fdd22bb7aa52 229 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 230 pIn2 = (q7_t *) py;
emilmont 1:fdd22bb7aa52 231
emilmont 1:fdd22bb7aa52 232 pScr2 = py;
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 235
emilmont 1:fdd22bb7aa52 236 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 239 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 240
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 243 {
emilmont 1:fdd22bb7aa52 244 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 245 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 246
emilmont 1:fdd22bb7aa52 247 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 248 acc0 = 0;
emilmont 1:fdd22bb7aa52 249 acc1 = 0;
emilmont 1:fdd22bb7aa52 250 acc2 = 0;
emilmont 1:fdd22bb7aa52 251 acc3 = 0;
emilmont 1:fdd22bb7aa52 252
emilmont 1:fdd22bb7aa52 253 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 254 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 255
emilmont 1:fdd22bb7aa52 256 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 257 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 258
emilmont 1:fdd22bb7aa52 259 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 262 {
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 265 y1 = _SIMD32_OFFSET(pScr2);
emilmont 1:fdd22bb7aa52 266
emilmont 1:fdd22bb7aa52 267 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 268 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 269 acc2 = __SMLAD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 270
emilmont 1:fdd22bb7aa52 271 /* pack input data */
emilmont 1:fdd22bb7aa52 272 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 273 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 274 #else
emilmont 1:fdd22bb7aa52 275 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 276 #endif
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 279 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 282 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 283
emilmont 1:fdd22bb7aa52 284 /* pack input data */
emilmont 1:fdd22bb7aa52 285 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 286 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 287 #else
emilmont 1:fdd22bb7aa52 288 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 289 #endif
emilmont 1:fdd22bb7aa52 290
emilmont 1:fdd22bb7aa52 291 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 292
emilmont 1:fdd22bb7aa52 293 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 294 y1 = _SIMD32_OFFSET(pScr2 + 2u);
emilmont 1:fdd22bb7aa52 295
emilmont 1:fdd22bb7aa52 296 acc0 = __SMLAD(x2, y1, acc0);
emilmont 1:fdd22bb7aa52 297
emilmont 1:fdd22bb7aa52 298 acc2 = __SMLAD(x1, y1, acc2);
emilmont 1:fdd22bb7aa52 299
emilmont 1:fdd22bb7aa52 300 acc1 = __SMLADX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 301
emilmont 1:fdd22bb7aa52 302 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 305 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 306 #else
emilmont 1:fdd22bb7aa52 307 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 308 #endif
emilmont 1:fdd22bb7aa52 309
emilmont 1:fdd22bb7aa52 310 acc3 = __SMLADX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 311
emilmont 1:fdd22bb7aa52 312 pScr2 += 4u;
emilmont 1:fdd22bb7aa52 313
emilmont 1:fdd22bb7aa52 314
emilmont 1:fdd22bb7aa52 315 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 316 tapCnt--;
emilmont 1:fdd22bb7aa52 317 }
emilmont 1:fdd22bb7aa52 318
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320
emilmont 1:fdd22bb7aa52 321 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 322 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 323
emilmont 1:fdd22bb7aa52 324
emilmont 1:fdd22bb7aa52 325 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 326 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 327
emilmont 1:fdd22bb7aa52 328 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 329 {
emilmont 1:fdd22bb7aa52 330
emilmont 1:fdd22bb7aa52 331 /* accumlate the results */
emilmont 1:fdd22bb7aa52 332 acc0 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 333 acc1 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 334 acc2 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 335 acc3 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 336
emilmont 1:fdd22bb7aa52 337 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 338
emilmont 1:fdd22bb7aa52 339 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 340 tapCnt--;
emilmont 1:fdd22bb7aa52 341 }
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 blkCnt--;
emilmont 1:fdd22bb7aa52 344
emilmont 1:fdd22bb7aa52 345 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 346 out0 = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 347 out1 = (q7_t) (__SSAT(acc1 >> 7u, 8));
emilmont 1:fdd22bb7aa52 348 out2 = (q7_t) (__SSAT(acc2 >> 7u, 8));
emilmont 1:fdd22bb7aa52 349 out3 = (q7_t) (__SSAT(acc3 >> 7u, 8));
emilmont 1:fdd22bb7aa52 350
emilmont 1:fdd22bb7aa52 351 *__SIMD32(pOut)++ = __PACKq7(out0, out1, out2, out3);
emilmont 1:fdd22bb7aa52 352
emilmont 1:fdd22bb7aa52 353 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 354 pScr2 = py;
emilmont 1:fdd22bb7aa52 355
emilmont 1:fdd22bb7aa52 356 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 357
emilmont 1:fdd22bb7aa52 358 }
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 blkCnt = (numPoints) & 0x3;
emilmont 1:fdd22bb7aa52 361
emilmont 1:fdd22bb7aa52 362 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 363 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 364 {
emilmont 1:fdd22bb7aa52 365 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 366 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 367
emilmont 1:fdd22bb7aa52 368 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 369 acc0 = 0;
emilmont 1:fdd22bb7aa52 370
emilmont 1:fdd22bb7aa52 371 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 372
emilmont 1:fdd22bb7aa52 373 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 374 {
emilmont 1:fdd22bb7aa52 375
emilmont 1:fdd22bb7aa52 376 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 377 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 378
emilmont 1:fdd22bb7aa52 379 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 380 y1 = *__SIMD32(pScr2)++;
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382 acc0 = __SMLAD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 383
emilmont 1:fdd22bb7aa52 384 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 385 tapCnt--;
emilmont 1:fdd22bb7aa52 386 }
emilmont 1:fdd22bb7aa52 387
emilmont 1:fdd22bb7aa52 388 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 389
emilmont 1:fdd22bb7aa52 390 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 391 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 392 {
emilmont 1:fdd22bb7aa52 393
emilmont 1:fdd22bb7aa52 394 /* accumlate the results */
emilmont 1:fdd22bb7aa52 395 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 396
emilmont 1:fdd22bb7aa52 397 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 398 tapCnt--;
emilmont 1:fdd22bb7aa52 399 }
emilmont 1:fdd22bb7aa52 400
emilmont 1:fdd22bb7aa52 401 blkCnt--;
emilmont 1:fdd22bb7aa52 402
emilmont 1:fdd22bb7aa52 403 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 404 *pOut++ = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 405
emilmont 1:fdd22bb7aa52 406 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 407 pScr2 = py;
emilmont 1:fdd22bb7aa52 408
emilmont 1:fdd22bb7aa52 409 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 410
emilmont 1:fdd22bb7aa52 411 }
emilmont 1:fdd22bb7aa52 412
emilmont 1:fdd22bb7aa52 413 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 414 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 415
emilmont 1:fdd22bb7aa52 416
emilmont 1:fdd22bb7aa52 417 }
emilmont 1:fdd22bb7aa52 418
emilmont 1:fdd22bb7aa52 419 return (status);
emilmont 1:fdd22bb7aa52 420
emilmont 1:fdd22bb7aa52 421 }
emilmont 1:fdd22bb7aa52 422
emilmont 1:fdd22bb7aa52 423 #else
emilmont 1:fdd22bb7aa52 424
emilmont 1:fdd22bb7aa52 425 arm_status arm_conv_partial_opt_q7(
emilmont 1:fdd22bb7aa52 426 q7_t * pSrcA,
emilmont 1:fdd22bb7aa52 427 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 428 q7_t * pSrcB,
emilmont 1:fdd22bb7aa52 429 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 430 q7_t * pDst,
emilmont 1:fdd22bb7aa52 431 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 432 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 433 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 434 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 435 {
emilmont 1:fdd22bb7aa52 436
emilmont 1:fdd22bb7aa52 437 q15_t *pScr2, *pScr1; /* Intermediate pointers for scratch pointers */
emilmont 1:fdd22bb7aa52 438 q15_t x4; /* Temporary input variable */
emilmont 1:fdd22bb7aa52 439 q7_t *pIn1, *pIn2; /* inputA and inputB pointer */
emilmont 1:fdd22bb7aa52 440 uint32_t j, k, blkCnt, tapCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 441 q7_t *px; /* Temporary input1 pointer */
emilmont 1:fdd22bb7aa52 442 q15_t *py; /* Temporary input2 pointer */
emilmont 1:fdd22bb7aa52 443 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 444 arm_status status;
emilmont 1:fdd22bb7aa52 445 q7_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 446 q15_t x10, x11, x20, x21; /* Temporary input variables */
emilmont 1:fdd22bb7aa52 447 q15_t y10, y11; /* Temporary input variables */
emilmont 1:fdd22bb7aa52 448 q7_t out0, out1, out2, out3; /* temporary variables */
emilmont 1:fdd22bb7aa52 449
emilmont 1:fdd22bb7aa52 450 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 451 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 452 {
emilmont 1:fdd22bb7aa52 453 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 454 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 455 }
emilmont 1:fdd22bb7aa52 456 else
emilmont 1:fdd22bb7aa52 457 {
emilmont 1:fdd22bb7aa52 458
emilmont 1:fdd22bb7aa52 459 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 460 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 461 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 462 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 463 {
emilmont 1:fdd22bb7aa52 464 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 465 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 466
emilmont 1:fdd22bb7aa52 467 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 468 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 469 }
emilmont 1:fdd22bb7aa52 470 else
emilmont 1:fdd22bb7aa52 471 {
emilmont 1:fdd22bb7aa52 472 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 473 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 474
emilmont 1:fdd22bb7aa52 475 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 476 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 477
emilmont 1:fdd22bb7aa52 478 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 479 j = srcBLen;
emilmont 1:fdd22bb7aa52 480 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 481 srcALen = j;
emilmont 1:fdd22bb7aa52 482 }
emilmont 1:fdd22bb7aa52 483
emilmont 1:fdd22bb7aa52 484 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 485 pScr2 = pScratch2;
emilmont 1:fdd22bb7aa52 486
emilmont 1:fdd22bb7aa52 487 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 488 px = pIn2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 489
emilmont 1:fdd22bb7aa52 490 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 491 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 492
emilmont 1:fdd22bb7aa52 493 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 494 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 495 while(k > 0u)
emilmont 1:fdd22bb7aa52 496 {
emilmont 1:fdd22bb7aa52 497 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 498 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 499 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 500 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 501 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 502 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 503 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 504 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 505 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 506
emilmont 1:fdd22bb7aa52 507 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 508 k--;
emilmont 1:fdd22bb7aa52 509 }
emilmont 1:fdd22bb7aa52 510
emilmont 1:fdd22bb7aa52 511 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 512 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 513 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 514
emilmont 1:fdd22bb7aa52 515 while(k > 0u)
emilmont 1:fdd22bb7aa52 516 {
emilmont 1:fdd22bb7aa52 517 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 518 x4 = (q15_t) * px--;
emilmont 1:fdd22bb7aa52 519 *pScr2++ = x4;
emilmont 1:fdd22bb7aa52 520
emilmont 1:fdd22bb7aa52 521 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 522 k--;
emilmont 1:fdd22bb7aa52 523 }
emilmont 1:fdd22bb7aa52 524
emilmont 1:fdd22bb7aa52 525 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 526 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 527
emilmont 1:fdd22bb7aa52 528 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 529 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 530
emilmont 1:fdd22bb7aa52 531 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 532 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 533
emilmont 1:fdd22bb7aa52 534 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 535 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 536 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 537
emilmont 1:fdd22bb7aa52 538 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 539 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 540 while(k > 0u)
emilmont 1:fdd22bb7aa52 541 {
emilmont 1:fdd22bb7aa52 542 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 543 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 544 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 545 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 546 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 547 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 548 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 549 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 550 *pScr1++ = x4;
emilmont 1:fdd22bb7aa52 551
emilmont 1:fdd22bb7aa52 552 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 553 k--;
emilmont 1:fdd22bb7aa52 554 }
emilmont 1:fdd22bb7aa52 555
emilmont 1:fdd22bb7aa52 556 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 557 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 558 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 559
emilmont 1:fdd22bb7aa52 560 while(k > 0u)
emilmont 1:fdd22bb7aa52 561 {
emilmont 1:fdd22bb7aa52 562 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 563 x4 = (q15_t) * pIn1++;
emilmont 1:fdd22bb7aa52 564 *pScr1++ = x4;
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 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 571 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 572
emilmont 1:fdd22bb7aa52 573 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 574 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 575 while(k > 0u)
emilmont 1:fdd22bb7aa52 576 {
emilmont 1:fdd22bb7aa52 577 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 578 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 579 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 580 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 581 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 582
emilmont 1:fdd22bb7aa52 583 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 584 k--;
emilmont 1:fdd22bb7aa52 585 }
emilmont 1:fdd22bb7aa52 586
emilmont 1:fdd22bb7aa52 587 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 588 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 589 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 590
emilmont 1:fdd22bb7aa52 591 while(k > 0u)
emilmont 1:fdd22bb7aa52 592 {
emilmont 1:fdd22bb7aa52 593 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 594 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 595
emilmont 1:fdd22bb7aa52 596 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 597 k--;
emilmont 1:fdd22bb7aa52 598 }
emilmont 1:fdd22bb7aa52 599
emilmont 1:fdd22bb7aa52 600
emilmont 1:fdd22bb7aa52 601 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 602 py = pScratch2;
emilmont 1:fdd22bb7aa52 603
emilmont 1:fdd22bb7aa52 604 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 605 pIn2 = (q7_t *) py;
emilmont 1:fdd22bb7aa52 606
emilmont 1:fdd22bb7aa52 607 pScr2 = py;
emilmont 1:fdd22bb7aa52 608
emilmont 1:fdd22bb7aa52 609 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 610
emilmont 1:fdd22bb7aa52 611 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 612
emilmont 1:fdd22bb7aa52 613 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 614 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 615
emilmont 1:fdd22bb7aa52 616
emilmont 1:fdd22bb7aa52 617 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 618 {
emilmont 1:fdd22bb7aa52 619 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 620 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 621
emilmont 1:fdd22bb7aa52 622 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 623 acc0 = 0;
emilmont 1:fdd22bb7aa52 624 acc1 = 0;
emilmont 1:fdd22bb7aa52 625 acc2 = 0;
emilmont 1:fdd22bb7aa52 626 acc3 = 0;
emilmont 1:fdd22bb7aa52 627
emilmont 1:fdd22bb7aa52 628 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 629 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 630 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 631
emilmont 1:fdd22bb7aa52 632 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 633 x20 = *pScr1++;
emilmont 1:fdd22bb7aa52 634 x21 = *pScr1++;
emilmont 1:fdd22bb7aa52 635
emilmont 1:fdd22bb7aa52 636 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 637
emilmont 1:fdd22bb7aa52 638 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 639 {
emilmont 1:fdd22bb7aa52 640
emilmont 1:fdd22bb7aa52 641 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 642 y10 = *pScr2;
emilmont 1:fdd22bb7aa52 643 y11 = *(pScr2 + 1u);
emilmont 1:fdd22bb7aa52 644
emilmont 1:fdd22bb7aa52 645 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 646 acc0 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 647 acc0 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 648 acc2 += (q31_t) x20 *y10;
emilmont 1:fdd22bb7aa52 649 acc2 += (q31_t) x21 *y11;
emilmont 1:fdd22bb7aa52 650
emilmont 1:fdd22bb7aa52 651
emilmont 1:fdd22bb7aa52 652 acc1 += (q31_t) x11 *y10;
emilmont 1:fdd22bb7aa52 653 acc1 += (q31_t) x20 *y11;
emilmont 1:fdd22bb7aa52 654
emilmont 1:fdd22bb7aa52 655 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 656 x10 = *pScr1;
emilmont 1:fdd22bb7aa52 657 x11 = *(pScr1 + 1u);
emilmont 1:fdd22bb7aa52 658
emilmont 1:fdd22bb7aa52 659 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 660 acc3 += (q31_t) x21 *y10;
emilmont 1:fdd22bb7aa52 661 acc3 += (q31_t) x10 *y11;
emilmont 1:fdd22bb7aa52 662
emilmont 1:fdd22bb7aa52 663 /* Read next two samples from scratch2 buffer */
emilmont 1:fdd22bb7aa52 664 y10 = *(pScr2 + 2u);
emilmont 1:fdd22bb7aa52 665 y11 = *(pScr2 + 3u);
emilmont 1:fdd22bb7aa52 666
emilmont 1:fdd22bb7aa52 667 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 668 acc0 += (q31_t) x20 *y10;
emilmont 1:fdd22bb7aa52 669 acc0 += (q31_t) x21 *y11;
emilmont 1:fdd22bb7aa52 670 acc2 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 671 acc2 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 672 acc1 += (q31_t) x21 *y10;
emilmont 1:fdd22bb7aa52 673 acc1 += (q31_t) x10 *y11;
emilmont 1:fdd22bb7aa52 674
emilmont 1:fdd22bb7aa52 675 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 676 x20 = *(pScr1 + 2);
emilmont 1:fdd22bb7aa52 677 x21 = *(pScr1 + 3);
emilmont 1:fdd22bb7aa52 678
emilmont 1:fdd22bb7aa52 679 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 680 acc3 += (q31_t) x11 *y10;
emilmont 1:fdd22bb7aa52 681 acc3 += (q31_t) x20 *y11;
emilmont 1:fdd22bb7aa52 682
emilmont 1:fdd22bb7aa52 683 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 684
emilmont 1:fdd22bb7aa52 685 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 686 pScr2 += 4u;
emilmont 1:fdd22bb7aa52 687
emilmont 1:fdd22bb7aa52 688 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 689 tapCnt--;
emilmont 1:fdd22bb7aa52 690 }
emilmont 1:fdd22bb7aa52 691
emilmont 1:fdd22bb7aa52 692
emilmont 1:fdd22bb7aa52 693
emilmont 1:fdd22bb7aa52 694 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 695 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 696
emilmont 1:fdd22bb7aa52 697
emilmont 1:fdd22bb7aa52 698 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 699 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 700
emilmont 1:fdd22bb7aa52 701 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 702 {
emilmont 1:fdd22bb7aa52 703
emilmont 1:fdd22bb7aa52 704 /* accumlate the results */
emilmont 1:fdd22bb7aa52 705 acc0 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 706 acc1 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 707 acc2 += (*pScr1++ * *pScr2);
emilmont 1:fdd22bb7aa52 708 acc3 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 709
emilmont 1:fdd22bb7aa52 710 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 711
emilmont 1:fdd22bb7aa52 712 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 713 tapCnt--;
emilmont 1:fdd22bb7aa52 714 }
emilmont 1:fdd22bb7aa52 715
emilmont 1:fdd22bb7aa52 716 blkCnt--;
emilmont 1:fdd22bb7aa52 717
emilmont 1:fdd22bb7aa52 718 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 719 out0 = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 720 out1 = (q7_t) (__SSAT(acc1 >> 7u, 8));
emilmont 1:fdd22bb7aa52 721 out2 = (q7_t) (__SSAT(acc2 >> 7u, 8));
emilmont 1:fdd22bb7aa52 722 out3 = (q7_t) (__SSAT(acc3 >> 7u, 8));
emilmont 1:fdd22bb7aa52 723
emilmont 1:fdd22bb7aa52 724
emilmont 1:fdd22bb7aa52 725 *__SIMD32(pOut)++ = __PACKq7(out0, out1, out2, out3);
emilmont 1:fdd22bb7aa52 726
emilmont 1:fdd22bb7aa52 727 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 728 pScr2 = py;
emilmont 1:fdd22bb7aa52 729
emilmont 1:fdd22bb7aa52 730 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 731
emilmont 1:fdd22bb7aa52 732 }
emilmont 1:fdd22bb7aa52 733
emilmont 1:fdd22bb7aa52 734 blkCnt = (numPoints) & 0x3;
emilmont 1:fdd22bb7aa52 735
emilmont 1:fdd22bb7aa52 736 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 737 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 738 {
emilmont 1:fdd22bb7aa52 739 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 740 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 741
emilmont 1:fdd22bb7aa52 742 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 743 acc0 = 0;
emilmont 1:fdd22bb7aa52 744
emilmont 1:fdd22bb7aa52 745 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 746
emilmont 1:fdd22bb7aa52 747 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 748 {
emilmont 1:fdd22bb7aa52 749
emilmont 1:fdd22bb7aa52 750 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 751 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 752 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 753
emilmont 1:fdd22bb7aa52 754 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 755 y10 = *pScr2++;
emilmont 1:fdd22bb7aa52 756 y11 = *pScr2++;
emilmont 1:fdd22bb7aa52 757
emilmont 1:fdd22bb7aa52 758 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 759 acc0 += (q31_t) x10 *y10;
emilmont 1:fdd22bb7aa52 760 acc0 += (q31_t) x11 *y11;
emilmont 1:fdd22bb7aa52 761
emilmont 1:fdd22bb7aa52 762 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 763 tapCnt--;
emilmont 1:fdd22bb7aa52 764 }
emilmont 1:fdd22bb7aa52 765
emilmont 1:fdd22bb7aa52 766 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 767
emilmont 1:fdd22bb7aa52 768 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 769 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 770 {
emilmont 1:fdd22bb7aa52 771
emilmont 1:fdd22bb7aa52 772 /* accumlate the results */
emilmont 1:fdd22bb7aa52 773 acc0 += (*pScr1++ * *pScr2++);
emilmont 1:fdd22bb7aa52 774
emilmont 1:fdd22bb7aa52 775 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 776 tapCnt--;
emilmont 1:fdd22bb7aa52 777 }
emilmont 1:fdd22bb7aa52 778
emilmont 1:fdd22bb7aa52 779 blkCnt--;
emilmont 1:fdd22bb7aa52 780
emilmont 1:fdd22bb7aa52 781 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 782 *pOut++ = (q7_t) (__SSAT(acc0 >> 7u, 8));
emilmont 1:fdd22bb7aa52 783
emilmont 1:fdd22bb7aa52 784 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 785 pScr2 = py;
emilmont 1:fdd22bb7aa52 786
emilmont 1:fdd22bb7aa52 787 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 788
emilmont 1:fdd22bb7aa52 789 }
emilmont 1:fdd22bb7aa52 790
emilmont 1:fdd22bb7aa52 791 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 792 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 793
emilmont 1:fdd22bb7aa52 794 }
emilmont 1:fdd22bb7aa52 795
emilmont 1:fdd22bb7aa52 796 return (status);
emilmont 1:fdd22bb7aa52 797
emilmont 1:fdd22bb7aa52 798 }
emilmont 1:fdd22bb7aa52 799
emilmont 2:da51fb522205 800 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 801
emilmont 1:fdd22bb7aa52 802
emilmont 1:fdd22bb7aa52 803
emilmont 1:fdd22bb7aa52 804 /**
emilmont 1:fdd22bb7aa52 805 * @} end of PartialConv group
emilmont 1:fdd22bb7aa52 806 */