Eli Hughes / CMSIS_DSP_401

V4.0.1 of the ARM CMSIS DSP libraries. Note that arm_bitreversal2.s, arm_cfft_f32.c and arm_rfft_fast_f32.c had to be removed. arm_bitreversal2.s will not assemble with the online tools. So, the fast f32 FFT functions are not yet available. All the other FFT functions are available.

Dependents:   MPU9150_Example fir_f32 fir_f32 MPU9150_nucleo_noni2cdev ... more

FilteringFunctions/arm_conv_partial_fast_opt_q15.c@0:3d9c67d97d6f, 2014-07-28 (annotated)

Committer:
emh203
Date:
Mon Jul 28 15:03:15 2014 +0000
Revision:
0:3d9c67d97d6f
1st working commit.   Had to remove arm_bitreversal2.s     arm_cfft_f32.c and arm_rfft_fast_f32.c.    The .s will not assemble.      For now I removed these functions so we could at least have a library for the other functions.

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emh203 0:3d9c67d97d6f 1 /* ----------------------------------------------------------------------
emh203 0:3d9c67d97d6f 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emh203 0:3d9c67d97d6f 3 *
emh203 0:3d9c67d97d6f 4 * $Date: 12. March 2014
emh203 0:3d9c67d97d6f 5 * $Revision: V1.4.3
emh203 0:3d9c67d97d6f 6 *
emh203 0:3d9c67d97d6f 7 * Project: CMSIS DSP Library
emh203 0:3d9c67d97d6f 8 * Title: arm_conv_partial_fast_opt_q15.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Fast Q15 Partial convolution.
emh203 0:3d9c67d97d6f 11 *
emh203 0:3d9c67d97d6f 12 * Target Processor: Cortex-M4/Cortex-M3
emh203 0:3d9c67d97d6f 13 *
emh203 0:3d9c67d97d6f 14 * Redistribution and use in source and binary forms, with or without
emh203 0:3d9c67d97d6f 15 * modification, are permitted provided that the following conditions
emh203 0:3d9c67d97d6f 16 * are met:
emh203 0:3d9c67d97d6f 17 * - Redistributions of source code must retain the above copyright
emh203 0:3d9c67d97d6f 18 * notice, this list of conditions and the following disclaimer.
emh203 0:3d9c67d97d6f 19 * - Redistributions in binary form must reproduce the above copyright
emh203 0:3d9c67d97d6f 20 * notice, this list of conditions and the following disclaimer in
emh203 0:3d9c67d97d6f 21 * the documentation and/or other materials provided with the
emh203 0:3d9c67d97d6f 22 * distribution.
emh203 0:3d9c67d97d6f 23 * - Neither the name of ARM LIMITED nor the names of its contributors
emh203 0:3d9c67d97d6f 24 * may be used to endorse or promote products derived from this
emh203 0:3d9c67d97d6f 25 * software without specific prior written permission.
emh203 0:3d9c67d97d6f 26 *
emh203 0:3d9c67d97d6f 27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
emh203 0:3d9c67d97d6f 28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
emh203 0:3d9c67d97d6f 29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
emh203 0:3d9c67d97d6f 30 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
emh203 0:3d9c67d97d6f 31 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
emh203 0:3d9c67d97d6f 32 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
emh203 0:3d9c67d97d6f 33 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
emh203 0:3d9c67d97d6f 34 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
emh203 0:3d9c67d97d6f 35 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
emh203 0:3d9c67d97d6f 36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
emh203 0:3d9c67d97d6f 37 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
emh203 0:3d9c67d97d6f 38 * POSSIBILITY OF SUCH DAMAGE.
emh203 0:3d9c67d97d6f 39 * -------------------------------------------------------------------- */
emh203 0:3d9c67d97d6f 40
emh203 0:3d9c67d97d6f 41 #include "arm_math.h"
emh203 0:3d9c67d97d6f 42
emh203 0:3d9c67d97d6f 43 /**
emh203 0:3d9c67d97d6f 44 * @ingroup groupFilters
emh203 0:3d9c67d97d6f 45 */
emh203 0:3d9c67d97d6f 46
emh203 0:3d9c67d97d6f 47 /**
emh203 0:3d9c67d97d6f 48 * @addtogroup PartialConv
emh203 0:3d9c67d97d6f 49 * @{
emh203 0:3d9c67d97d6f 50 */
emh203 0:3d9c67d97d6f 51
emh203 0:3d9c67d97d6f 52 /**
emh203 0:3d9c67d97d6f 53 * @brief Partial convolution of Q15 sequences (fast version) for Cortex-M3 and Cortex-M4.
emh203 0:3d9c67d97d6f 54 * @param[in] *pSrcA points to the first input sequence.
emh203 0:3d9c67d97d6f 55 * @param[in] srcALen length of the first input sequence.
emh203 0:3d9c67d97d6f 56 * @param[in] *pSrcB points to the second input sequence.
emh203 0:3d9c67d97d6f 57 * @param[in] srcBLen length of the second input sequence.
emh203 0:3d9c67d97d6f 58 * @param[out] *pDst points to the location where the output result is written.
emh203 0:3d9c67d97d6f 59 * @param[in] firstIndex is the first output sample to start with.
emh203 0:3d9c67d97d6f 60 * @param[in] numPoints is the number of output points to be computed.
emh203 0:3d9c67d97d6f 61 * @param[in] *pScratch1 points to scratch buffer of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emh203 0:3d9c67d97d6f 62 * @param[in] *pScratch2 points to scratch buffer of size min(srcALen, srcBLen).
emh203 0:3d9c67d97d6f 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].
emh203 0:3d9c67d97d6f 64 *
emh203 0:3d9c67d97d6f 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.
emh203 0:3d9c67d97d6f 66 *
emh203 0:3d9c67d97d6f 67 * \par Restrictions
emh203 0:3d9c67d97d6f 68 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emh203 0:3d9c67d97d6f 69 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emh203 0:3d9c67d97d6f 70 *
emh203 0:3d9c67d97d6f 71 */
emh203 0:3d9c67d97d6f 72
emh203 0:3d9c67d97d6f 73 #ifndef UNALIGNED_SUPPORT_DISABLE
emh203 0:3d9c67d97d6f 74
emh203 0:3d9c67d97d6f 75 arm_status arm_conv_partial_fast_opt_q15(
emh203 0:3d9c67d97d6f 76 q15_t * pSrcA,
emh203 0:3d9c67d97d6f 77 uint32_t srcALen,
emh203 0:3d9c67d97d6f 78 q15_t * pSrcB,
emh203 0:3d9c67d97d6f 79 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 80 q15_t * pDst,
emh203 0:3d9c67d97d6f 81 uint32_t firstIndex,
emh203 0:3d9c67d97d6f 82 uint32_t numPoints,
emh203 0:3d9c67d97d6f 83 q15_t * pScratch1,
emh203 0:3d9c67d97d6f 84 q15_t * pScratch2)
emh203 0:3d9c67d97d6f 85 {
emh203 0:3d9c67d97d6f 86
emh203 0:3d9c67d97d6f 87 q15_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 88 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emh203 0:3d9c67d97d6f 89 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emh203 0:3d9c67d97d6f 90 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emh203 0:3d9c67d97d6f 91 q31_t x1, x2, x3; /* Temporary variables to hold state and coefficient values */
emh203 0:3d9c67d97d6f 92 q31_t y1, y2; /* State variables */
emh203 0:3d9c67d97d6f 93 q15_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 94 q15_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 95 q15_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 96 q15_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 97 uint32_t j, k, blkCnt; /* loop counter */
emh203 0:3d9c67d97d6f 98 arm_status status;
emh203 0:3d9c67d97d6f 99
emh203 0:3d9c67d97d6f 100 uint32_t tapCnt; /* loop count */
emh203 0:3d9c67d97d6f 101
emh203 0:3d9c67d97d6f 102 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 103 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 104 {
emh203 0:3d9c67d97d6f 105 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 106 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 107 }
emh203 0:3d9c67d97d6f 108 else
emh203 0:3d9c67d97d6f 109 {
emh203 0:3d9c67d97d6f 110
emh203 0:3d9c67d97d6f 111 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 112 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 113 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 114 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 115 {
emh203 0:3d9c67d97d6f 116 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 117 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 118
emh203 0:3d9c67d97d6f 119 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 120 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 121 }
emh203 0:3d9c67d97d6f 122 else
emh203 0:3d9c67d97d6f 123 {
emh203 0:3d9c67d97d6f 124 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 125 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 126
emh203 0:3d9c67d97d6f 127 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 128 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 129
emh203 0:3d9c67d97d6f 130 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 131 j = srcBLen;
emh203 0:3d9c67d97d6f 132 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 133 srcALen = j;
emh203 0:3d9c67d97d6f 134 }
emh203 0:3d9c67d97d6f 135
emh203 0:3d9c67d97d6f 136 /* Temporary pointer for scratch2 */
emh203 0:3d9c67d97d6f 137 py = pScratch2;
emh203 0:3d9c67d97d6f 138
emh203 0:3d9c67d97d6f 139 /* pointer to take end of scratch2 buffer */
emh203 0:3d9c67d97d6f 140 pScr2 = pScratch2 + srcBLen - 1;
emh203 0:3d9c67d97d6f 141
emh203 0:3d9c67d97d6f 142 /* points to smaller length sequence */
emh203 0:3d9c67d97d6f 143 px = pIn2;
emh203 0:3d9c67d97d6f 144
emh203 0:3d9c67d97d6f 145 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 146 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 147
emh203 0:3d9c67d97d6f 148 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 149 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 150
emh203 0:3d9c67d97d6f 151 /* Copy smaller length input sequence in reverse order into second scratch buffer */
emh203 0:3d9c67d97d6f 152 while(k > 0u)
emh203 0:3d9c67d97d6f 153 {
emh203 0:3d9c67d97d6f 154 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 155 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 156 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 157 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 158 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 159
emh203 0:3d9c67d97d6f 160 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 161 k--;
emh203 0:3d9c67d97d6f 162 }
emh203 0:3d9c67d97d6f 163
emh203 0:3d9c67d97d6f 164 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 165 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 166 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 167
emh203 0:3d9c67d97d6f 168 while(k > 0u)
emh203 0:3d9c67d97d6f 169 {
emh203 0:3d9c67d97d6f 170 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 171 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 172
emh203 0:3d9c67d97d6f 173 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 174 k--;
emh203 0:3d9c67d97d6f 175 }
emh203 0:3d9c67d97d6f 176
emh203 0:3d9c67d97d6f 177 /* Initialze temporary scratch pointer */
emh203 0:3d9c67d97d6f 178 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 179
emh203 0:3d9c67d97d6f 180 /* Assuming scratch1 buffer is aligned by 32-bit */
emh203 0:3d9c67d97d6f 181 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emh203 0:3d9c67d97d6f 182 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emh203 0:3d9c67d97d6f 183
emh203 0:3d9c67d97d6f 184 /* Update temporary scratch pointer */
emh203 0:3d9c67d97d6f 185 pScr1 += (srcBLen - 1u);
emh203 0:3d9c67d97d6f 186
emh203 0:3d9c67d97d6f 187 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emh203 0:3d9c67d97d6f 188
emh203 0:3d9c67d97d6f 189 /* Copy (srcALen) samples in scratch buffer */
emh203 0:3d9c67d97d6f 190 arm_copy_q15(pIn1, pScr1, srcALen);
emh203 0:3d9c67d97d6f 191
emh203 0:3d9c67d97d6f 192 /* Update pointers */
emh203 0:3d9c67d97d6f 193 pScr1 += srcALen;
emh203 0:3d9c67d97d6f 194
emh203 0:3d9c67d97d6f 195 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emh203 0:3d9c67d97d6f 196 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emh203 0:3d9c67d97d6f 197
emh203 0:3d9c67d97d6f 198 /* Update pointer */
emh203 0:3d9c67d97d6f 199 pScr1 += (srcBLen - 1u);
emh203 0:3d9c67d97d6f 200
emh203 0:3d9c67d97d6f 201 /* Initialization of pIn2 pointer */
emh203 0:3d9c67d97d6f 202 pIn2 = py;
emh203 0:3d9c67d97d6f 203
emh203 0:3d9c67d97d6f 204 pScratch1 += firstIndex;
emh203 0:3d9c67d97d6f 205
emh203 0:3d9c67d97d6f 206 pOut = pDst + firstIndex;
emh203 0:3d9c67d97d6f 207
emh203 0:3d9c67d97d6f 208 /* First part of the processing with loop unrolling process 4 data points at a time.
emh203 0:3d9c67d97d6f 209 ** a second loop below process for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 210
emh203 0:3d9c67d97d6f 211 /* Actual convolution process starts here */
emh203 0:3d9c67d97d6f 212 blkCnt = (numPoints) >> 2;
emh203 0:3d9c67d97d6f 213
emh203 0:3d9c67d97d6f 214 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 215 {
emh203 0:3d9c67d97d6f 216 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 217 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 218
emh203 0:3d9c67d97d6f 219 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 220 acc0 = 0;
emh203 0:3d9c67d97d6f 221 acc1 = 0;
emh203 0:3d9c67d97d6f 222 acc2 = 0;
emh203 0:3d9c67d97d6f 223 acc3 = 0;
emh203 0:3d9c67d97d6f 224
emh203 0:3d9c67d97d6f 225 /* Read two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 226 x1 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 227
emh203 0:3d9c67d97d6f 228 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 229 x2 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 230
emh203 0:3d9c67d97d6f 231 tapCnt = (srcBLen) >> 2u;
emh203 0:3d9c67d97d6f 232
emh203 0:3d9c67d97d6f 233 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 234 {
emh203 0:3d9c67d97d6f 235
emh203 0:3d9c67d97d6f 236 /* Read four samples from smaller buffer */
emh203 0:3d9c67d97d6f 237 y1 = _SIMD32_OFFSET(pIn2);
emh203 0:3d9c67d97d6f 238 y2 = _SIMD32_OFFSET(pIn2 + 2u);
emh203 0:3d9c67d97d6f 239
emh203 0:3d9c67d97d6f 240 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 241 acc0 = __SMLAD(x1, y1, acc0);
emh203 0:3d9c67d97d6f 242 acc2 = __SMLAD(x2, y1, acc2);
emh203 0:3d9c67d97d6f 243
emh203 0:3d9c67d97d6f 244 /* pack input data */
emh203 0:3d9c67d97d6f 245 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 246 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 247 #else
emh203 0:3d9c67d97d6f 248 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 249 #endif
emh203 0:3d9c67d97d6f 250
emh203 0:3d9c67d97d6f 251 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 252 acc1 = __SMLADX(x3, y1, acc1);
emh203 0:3d9c67d97d6f 253
emh203 0:3d9c67d97d6f 254 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 255 x1 = _SIMD32_OFFSET(pScr1);
emh203 0:3d9c67d97d6f 256
emh203 0:3d9c67d97d6f 257 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 258 acc0 = __SMLAD(x2, y2, acc0);
emh203 0:3d9c67d97d6f 259
emh203 0:3d9c67d97d6f 260 acc2 = __SMLAD(x1, y2, acc2);
emh203 0:3d9c67d97d6f 261
emh203 0:3d9c67d97d6f 262 /* pack input data */
emh203 0:3d9c67d97d6f 263 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 264 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 265 #else
emh203 0:3d9c67d97d6f 266 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 267 #endif
emh203 0:3d9c67d97d6f 268
emh203 0:3d9c67d97d6f 269 acc3 = __SMLADX(x3, y1, acc3);
emh203 0:3d9c67d97d6f 270 acc1 = __SMLADX(x3, y2, acc1);
emh203 0:3d9c67d97d6f 271
emh203 0:3d9c67d97d6f 272 x2 = _SIMD32_OFFSET(pScr1 + 2u);
emh203 0:3d9c67d97d6f 273
emh203 0:3d9c67d97d6f 274 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 275 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 276 #else
emh203 0:3d9c67d97d6f 277 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 278 #endif
emh203 0:3d9c67d97d6f 279
emh203 0:3d9c67d97d6f 280 acc3 = __SMLADX(x3, y2, acc3);
emh203 0:3d9c67d97d6f 281
emh203 0:3d9c67d97d6f 282 /* update scratch pointers */
emh203 0:3d9c67d97d6f 283 pIn2 += 4u;
emh203 0:3d9c67d97d6f 284 pScr1 += 4u;
emh203 0:3d9c67d97d6f 285
emh203 0:3d9c67d97d6f 286
emh203 0:3d9c67d97d6f 287 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 288 tapCnt--;
emh203 0:3d9c67d97d6f 289 }
emh203 0:3d9c67d97d6f 290
emh203 0:3d9c67d97d6f 291 /* Update scratch pointer for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 292 pScr1 -= 4u;
emh203 0:3d9c67d97d6f 293
emh203 0:3d9c67d97d6f 294 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 295 tapCnt = (srcBLen) & 3u;
emh203 0:3d9c67d97d6f 296
emh203 0:3d9c67d97d6f 297 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 298 {
emh203 0:3d9c67d97d6f 299
emh203 0:3d9c67d97d6f 300 /* accumlate the results */
emh203 0:3d9c67d97d6f 301 acc0 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 302 acc1 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 303 acc2 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 304 acc3 += (*pScr1++ * *pIn2++);
emh203 0:3d9c67d97d6f 305
emh203 0:3d9c67d97d6f 306 pScr1 -= 3u;
emh203 0:3d9c67d97d6f 307
emh203 0:3d9c67d97d6f 308 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 309 tapCnt--;
emh203 0:3d9c67d97d6f 310 }
emh203 0:3d9c67d97d6f 311
emh203 0:3d9c67d97d6f 312 blkCnt--;
emh203 0:3d9c67d97d6f 313
emh203 0:3d9c67d97d6f 314
emh203 0:3d9c67d97d6f 315 /* Store the results in the accumulators in the destination buffer. */
emh203 0:3d9c67d97d6f 316
emh203 0:3d9c67d97d6f 317 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 318
emh203 0:3d9c67d97d6f 319 *__SIMD32(pOut)++ =
emh203 0:3d9c67d97d6f 320 __PKHBT(__SSAT((acc0 >> 15), 16), __SSAT((acc1 >> 15), 16), 16);
emh203 0:3d9c67d97d6f 321 *__SIMD32(pOut)++ =
emh203 0:3d9c67d97d6f 322 __PKHBT(__SSAT((acc2 >> 15), 16), __SSAT((acc3 >> 15), 16), 16);
emh203 0:3d9c67d97d6f 323
emh203 0:3d9c67d97d6f 324 #else
emh203 0:3d9c67d97d6f 325
emh203 0:3d9c67d97d6f 326 *__SIMD32(pOut)++ =
emh203 0:3d9c67d97d6f 327 __PKHBT(__SSAT((acc1 >> 15), 16), __SSAT((acc0 >> 15), 16), 16);
emh203 0:3d9c67d97d6f 328 *__SIMD32(pOut)++ =
emh203 0:3d9c67d97d6f 329 __PKHBT(__SSAT((acc3 >> 15), 16), __SSAT((acc2 >> 15), 16), 16);
emh203 0:3d9c67d97d6f 330
emh203 0:3d9c67d97d6f 331 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 332
emh203 0:3d9c67d97d6f 333 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 334 pIn2 = py;
emh203 0:3d9c67d97d6f 335
emh203 0:3d9c67d97d6f 336 pScratch1 += 4u;
emh203 0:3d9c67d97d6f 337
emh203 0:3d9c67d97d6f 338 }
emh203 0:3d9c67d97d6f 339
emh203 0:3d9c67d97d6f 340
emh203 0:3d9c67d97d6f 341 blkCnt = numPoints & 0x3;
emh203 0:3d9c67d97d6f 342
emh203 0:3d9c67d97d6f 343 /* Calculate convolution for remaining samples of Bigger length sequence */
emh203 0:3d9c67d97d6f 344 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 345 {
emh203 0:3d9c67d97d6f 346 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 347 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 348
emh203 0:3d9c67d97d6f 349 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 350 acc0 = 0;
emh203 0:3d9c67d97d6f 351
emh203 0:3d9c67d97d6f 352 tapCnt = (srcBLen) >> 1u;
emh203 0:3d9c67d97d6f 353
emh203 0:3d9c67d97d6f 354 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 355 {
emh203 0:3d9c67d97d6f 356
emh203 0:3d9c67d97d6f 357 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 358 x1 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 359
emh203 0:3d9c67d97d6f 360 /* Read two samples from smaller buffer */
emh203 0:3d9c67d97d6f 361 y1 = *__SIMD32(pIn2)++;
emh203 0:3d9c67d97d6f 362
emh203 0:3d9c67d97d6f 363 acc0 = __SMLAD(x1, y1, acc0);
emh203 0:3d9c67d97d6f 364
emh203 0:3d9c67d97d6f 365 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 366 tapCnt--;
emh203 0:3d9c67d97d6f 367 }
emh203 0:3d9c67d97d6f 368
emh203 0:3d9c67d97d6f 369 tapCnt = (srcBLen) & 1u;
emh203 0:3d9c67d97d6f 370
emh203 0:3d9c67d97d6f 371 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 372 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 373 {
emh203 0:3d9c67d97d6f 374
emh203 0:3d9c67d97d6f 375 /* accumlate the results */
emh203 0:3d9c67d97d6f 376 acc0 += (*pScr1++ * *pIn2++);
emh203 0:3d9c67d97d6f 377
emh203 0:3d9c67d97d6f 378 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 379 tapCnt--;
emh203 0:3d9c67d97d6f 380 }
emh203 0:3d9c67d97d6f 381
emh203 0:3d9c67d97d6f 382 blkCnt--;
emh203 0:3d9c67d97d6f 383
emh203 0:3d9c67d97d6f 384 /* The result is in 2.30 format. Convert to 1.15 with saturation.
emh203 0:3d9c67d97d6f 385 ** Then store the output in the destination buffer. */
emh203 0:3d9c67d97d6f 386 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emh203 0:3d9c67d97d6f 387
emh203 0:3d9c67d97d6f 388 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 389 pIn2 = py;
emh203 0:3d9c67d97d6f 390
emh203 0:3d9c67d97d6f 391 pScratch1 += 1u;
emh203 0:3d9c67d97d6f 392
emh203 0:3d9c67d97d6f 393 }
emh203 0:3d9c67d97d6f 394 /* set status as ARM_MATH_SUCCESS */
emh203 0:3d9c67d97d6f 395 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 396 }
emh203 0:3d9c67d97d6f 397 /* Return to application */
emh203 0:3d9c67d97d6f 398 return (status);
emh203 0:3d9c67d97d6f 399 }
emh203 0:3d9c67d97d6f 400
emh203 0:3d9c67d97d6f 401 #else
emh203 0:3d9c67d97d6f 402
emh203 0:3d9c67d97d6f 403 arm_status arm_conv_partial_fast_opt_q15(
emh203 0:3d9c67d97d6f 404 q15_t * pSrcA,
emh203 0:3d9c67d97d6f 405 uint32_t srcALen,
emh203 0:3d9c67d97d6f 406 q15_t * pSrcB,
emh203 0:3d9c67d97d6f 407 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 408 q15_t * pDst,
emh203 0:3d9c67d97d6f 409 uint32_t firstIndex,
emh203 0:3d9c67d97d6f 410 uint32_t numPoints,
emh203 0:3d9c67d97d6f 411 q15_t * pScratch1,
emh203 0:3d9c67d97d6f 412 q15_t * pScratch2)
emh203 0:3d9c67d97d6f 413 {
emh203 0:3d9c67d97d6f 414
emh203 0:3d9c67d97d6f 415 q15_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 416 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emh203 0:3d9c67d97d6f 417 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emh203 0:3d9c67d97d6f 418 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emh203 0:3d9c67d97d6f 419 q15_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 420 q15_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 421 q15_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 422 q15_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 423 uint32_t j, k, blkCnt; /* loop counter */
emh203 0:3d9c67d97d6f 424 arm_status status; /* Status variable */
emh203 0:3d9c67d97d6f 425 uint32_t tapCnt; /* loop count */
emh203 0:3d9c67d97d6f 426 q15_t x10, x11, x20, x21; /* Temporary variables to hold srcA buffer */
emh203 0:3d9c67d97d6f 427 q15_t y10, y11; /* Temporary variables to hold srcB buffer */
emh203 0:3d9c67d97d6f 428
emh203 0:3d9c67d97d6f 429
emh203 0:3d9c67d97d6f 430 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 431 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 432 {
emh203 0:3d9c67d97d6f 433 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 434 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 435 }
emh203 0:3d9c67d97d6f 436 else
emh203 0:3d9c67d97d6f 437 {
emh203 0:3d9c67d97d6f 438
emh203 0:3d9c67d97d6f 439 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 440 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 441 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 442 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 443 {
emh203 0:3d9c67d97d6f 444 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 445 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 446
emh203 0:3d9c67d97d6f 447 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 448 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 449 }
emh203 0:3d9c67d97d6f 450 else
emh203 0:3d9c67d97d6f 451 {
emh203 0:3d9c67d97d6f 452 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 453 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 454
emh203 0:3d9c67d97d6f 455 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 456 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 457
emh203 0:3d9c67d97d6f 458 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 459 j = srcBLen;
emh203 0:3d9c67d97d6f 460 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 461 srcALen = j;
emh203 0:3d9c67d97d6f 462 }
emh203 0:3d9c67d97d6f 463
emh203 0:3d9c67d97d6f 464 /* Temporary pointer for scratch2 */
emh203 0:3d9c67d97d6f 465 py = pScratch2;
emh203 0:3d9c67d97d6f 466
emh203 0:3d9c67d97d6f 467 /* pointer to take end of scratch2 buffer */
emh203 0:3d9c67d97d6f 468 pScr2 = pScratch2 + srcBLen - 1;
emh203 0:3d9c67d97d6f 469
emh203 0:3d9c67d97d6f 470 /* points to smaller length sequence */
emh203 0:3d9c67d97d6f 471 px = pIn2;
emh203 0:3d9c67d97d6f 472
emh203 0:3d9c67d97d6f 473 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 474 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 475
emh203 0:3d9c67d97d6f 476 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 477 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 478 while(k > 0u)
emh203 0:3d9c67d97d6f 479 {
emh203 0:3d9c67d97d6f 480 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 481 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 482 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 483 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 484 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 485
emh203 0:3d9c67d97d6f 486 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 487 k--;
emh203 0:3d9c67d97d6f 488 }
emh203 0:3d9c67d97d6f 489
emh203 0:3d9c67d97d6f 490 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 491 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 492 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 493
emh203 0:3d9c67d97d6f 494 while(k > 0u)
emh203 0:3d9c67d97d6f 495 {
emh203 0:3d9c67d97d6f 496 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 497 *pScr2-- = *px++;
emh203 0:3d9c67d97d6f 498
emh203 0:3d9c67d97d6f 499 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 500 k--;
emh203 0:3d9c67d97d6f 501 }
emh203 0:3d9c67d97d6f 502
emh203 0:3d9c67d97d6f 503 /* Initialze temporary scratch pointer */
emh203 0:3d9c67d97d6f 504 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 505
emh203 0:3d9c67d97d6f 506 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emh203 0:3d9c67d97d6f 507 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emh203 0:3d9c67d97d6f 508
emh203 0:3d9c67d97d6f 509 /* Update temporary scratch pointer */
emh203 0:3d9c67d97d6f 510 pScr1 += (srcBLen - 1u);
emh203 0:3d9c67d97d6f 511
emh203 0:3d9c67d97d6f 512 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emh203 0:3d9c67d97d6f 513
emh203 0:3d9c67d97d6f 514
emh203 0:3d9c67d97d6f 515 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 516 k = srcALen >> 2u;
emh203 0:3d9c67d97d6f 517
emh203 0:3d9c67d97d6f 518 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 519 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 520 while(k > 0u)
emh203 0:3d9c67d97d6f 521 {
emh203 0:3d9c67d97d6f 522 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 523 *pScr1++ = *pIn1++;
emh203 0:3d9c67d97d6f 524 *pScr1++ = *pIn1++;
emh203 0:3d9c67d97d6f 525 *pScr1++ = *pIn1++;
emh203 0:3d9c67d97d6f 526 *pScr1++ = *pIn1++;
emh203 0:3d9c67d97d6f 527
emh203 0:3d9c67d97d6f 528 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 529 k--;
emh203 0:3d9c67d97d6f 530 }
emh203 0:3d9c67d97d6f 531
emh203 0:3d9c67d97d6f 532 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 533 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 534 k = srcALen % 0x4u;
emh203 0:3d9c67d97d6f 535
emh203 0:3d9c67d97d6f 536 while(k > 0u)
emh203 0:3d9c67d97d6f 537 {
emh203 0:3d9c67d97d6f 538 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 539 *pScr1++ = *pIn1++;
emh203 0:3d9c67d97d6f 540
emh203 0:3d9c67d97d6f 541 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 542 k--;
emh203 0:3d9c67d97d6f 543 }
emh203 0:3d9c67d97d6f 544
emh203 0:3d9c67d97d6f 545
emh203 0:3d9c67d97d6f 546 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 547 k = (srcBLen - 1u) >> 2u;
emh203 0:3d9c67d97d6f 548
emh203 0:3d9c67d97d6f 549 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 550 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 551 while(k > 0u)
emh203 0:3d9c67d97d6f 552 {
emh203 0:3d9c67d97d6f 553 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 554 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 555 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 556 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 557 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 558
emh203 0:3d9c67d97d6f 559 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 560 k--;
emh203 0:3d9c67d97d6f 561 }
emh203 0:3d9c67d97d6f 562
emh203 0:3d9c67d97d6f 563 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 564 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 565 k = (srcBLen - 1u) % 0x4u;
emh203 0:3d9c67d97d6f 566
emh203 0:3d9c67d97d6f 567 while(k > 0u)
emh203 0:3d9c67d97d6f 568 {
emh203 0:3d9c67d97d6f 569 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 570 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 571
emh203 0:3d9c67d97d6f 572 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 573 k--;
emh203 0:3d9c67d97d6f 574 }
emh203 0:3d9c67d97d6f 575
emh203 0:3d9c67d97d6f 576
emh203 0:3d9c67d97d6f 577 /* Initialization of pIn2 pointer */
emh203 0:3d9c67d97d6f 578 pIn2 = py;
emh203 0:3d9c67d97d6f 579
emh203 0:3d9c67d97d6f 580 pScratch1 += firstIndex;
emh203 0:3d9c67d97d6f 581
emh203 0:3d9c67d97d6f 582 pOut = pDst + firstIndex;
emh203 0:3d9c67d97d6f 583
emh203 0:3d9c67d97d6f 584 /* Actual convolution process starts here */
emh203 0:3d9c67d97d6f 585 blkCnt = (numPoints) >> 2;
emh203 0:3d9c67d97d6f 586
emh203 0:3d9c67d97d6f 587 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 588 {
emh203 0:3d9c67d97d6f 589 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 590 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 591
emh203 0:3d9c67d97d6f 592 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 593 acc0 = 0;
emh203 0:3d9c67d97d6f 594 acc1 = 0;
emh203 0:3d9c67d97d6f 595 acc2 = 0;
emh203 0:3d9c67d97d6f 596 acc3 = 0;
emh203 0:3d9c67d97d6f 597
emh203 0:3d9c67d97d6f 598 /* Read two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 599 x10 = *pScr1++;
emh203 0:3d9c67d97d6f 600 x11 = *pScr1++;
emh203 0:3d9c67d97d6f 601
emh203 0:3d9c67d97d6f 602 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 603 x20 = *pScr1++;
emh203 0:3d9c67d97d6f 604 x21 = *pScr1++;
emh203 0:3d9c67d97d6f 605
emh203 0:3d9c67d97d6f 606 tapCnt = (srcBLen) >> 2u;
emh203 0:3d9c67d97d6f 607
emh203 0:3d9c67d97d6f 608 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 609 {
emh203 0:3d9c67d97d6f 610
emh203 0:3d9c67d97d6f 611 /* Read two samples from smaller buffer */
emh203 0:3d9c67d97d6f 612 y10 = *pIn2;
emh203 0:3d9c67d97d6f 613 y11 = *(pIn2 + 1u);
emh203 0:3d9c67d97d6f 614
emh203 0:3d9c67d97d6f 615 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 616 acc0 += (q31_t) x10 *y10;
emh203 0:3d9c67d97d6f 617 acc0 += (q31_t) x11 *y11;
emh203 0:3d9c67d97d6f 618 acc2 += (q31_t) x20 *y10;
emh203 0:3d9c67d97d6f 619 acc2 += (q31_t) x21 *y11;
emh203 0:3d9c67d97d6f 620
emh203 0:3d9c67d97d6f 621 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 622 acc1 += (q31_t) x11 *y10;
emh203 0:3d9c67d97d6f 623 acc1 += (q31_t) x20 *y11;
emh203 0:3d9c67d97d6f 624
emh203 0:3d9c67d97d6f 625 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 626 x10 = *pScr1;
emh203 0:3d9c67d97d6f 627 x11 = *(pScr1 + 1u);
emh203 0:3d9c67d97d6f 628
emh203 0:3d9c67d97d6f 629 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 630 acc3 += (q31_t) x21 *y10;
emh203 0:3d9c67d97d6f 631 acc3 += (q31_t) x10 *y11;
emh203 0:3d9c67d97d6f 632
emh203 0:3d9c67d97d6f 633 /* Read next two samples from scratch2 buffer */
emh203 0:3d9c67d97d6f 634 y10 = *(pIn2 + 2u);
emh203 0:3d9c67d97d6f 635 y11 = *(pIn2 + 3u);
emh203 0:3d9c67d97d6f 636
emh203 0:3d9c67d97d6f 637 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 638 acc0 += (q31_t) x20 *y10;
emh203 0:3d9c67d97d6f 639 acc0 += (q31_t) x21 *y11;
emh203 0:3d9c67d97d6f 640 acc2 += (q31_t) x10 *y10;
emh203 0:3d9c67d97d6f 641 acc2 += (q31_t) x11 *y11;
emh203 0:3d9c67d97d6f 642 acc1 += (q31_t) x21 *y10;
emh203 0:3d9c67d97d6f 643 acc1 += (q31_t) x10 *y11;
emh203 0:3d9c67d97d6f 644
emh203 0:3d9c67d97d6f 645 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 646 x20 = *(pScr1 + 2);
emh203 0:3d9c67d97d6f 647 x21 = *(pScr1 + 3);
emh203 0:3d9c67d97d6f 648
emh203 0:3d9c67d97d6f 649 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 650 acc3 += (q31_t) x11 *y10;
emh203 0:3d9c67d97d6f 651 acc3 += (q31_t) x20 *y11;
emh203 0:3d9c67d97d6f 652
emh203 0:3d9c67d97d6f 653 /* update scratch pointers */
emh203 0:3d9c67d97d6f 654 pIn2 += 4u;
emh203 0:3d9c67d97d6f 655 pScr1 += 4u;
emh203 0:3d9c67d97d6f 656
emh203 0:3d9c67d97d6f 657 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 658 tapCnt--;
emh203 0:3d9c67d97d6f 659 }
emh203 0:3d9c67d97d6f 660
emh203 0:3d9c67d97d6f 661 /* Update scratch pointer for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 662 pScr1 -= 4u;
emh203 0:3d9c67d97d6f 663
emh203 0:3d9c67d97d6f 664 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 665 tapCnt = (srcBLen) & 3u;
emh203 0:3d9c67d97d6f 666
emh203 0:3d9c67d97d6f 667 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 668 {
emh203 0:3d9c67d97d6f 669 /* accumlate the results */
emh203 0:3d9c67d97d6f 670 acc0 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 671 acc1 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 672 acc2 += (*pScr1++ * *pIn2);
emh203 0:3d9c67d97d6f 673 acc3 += (*pScr1++ * *pIn2++);
emh203 0:3d9c67d97d6f 674
emh203 0:3d9c67d97d6f 675 pScr1 -= 3u;
emh203 0:3d9c67d97d6f 676
emh203 0:3d9c67d97d6f 677 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 678 tapCnt--;
emh203 0:3d9c67d97d6f 679 }
emh203 0:3d9c67d97d6f 680
emh203 0:3d9c67d97d6f 681 blkCnt--;
emh203 0:3d9c67d97d6f 682
emh203 0:3d9c67d97d6f 683
emh203 0:3d9c67d97d6f 684 /* Store the results in the accumulators in the destination buffer. */
emh203 0:3d9c67d97d6f 685 *pOut++ = __SSAT((acc0 >> 15), 16);
emh203 0:3d9c67d97d6f 686 *pOut++ = __SSAT((acc1 >> 15), 16);
emh203 0:3d9c67d97d6f 687 *pOut++ = __SSAT((acc2 >> 15), 16);
emh203 0:3d9c67d97d6f 688 *pOut++ = __SSAT((acc3 >> 15), 16);
emh203 0:3d9c67d97d6f 689
emh203 0:3d9c67d97d6f 690 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 691 pIn2 = py;
emh203 0:3d9c67d97d6f 692
emh203 0:3d9c67d97d6f 693 pScratch1 += 4u;
emh203 0:3d9c67d97d6f 694
emh203 0:3d9c67d97d6f 695 }
emh203 0:3d9c67d97d6f 696
emh203 0:3d9c67d97d6f 697
emh203 0:3d9c67d97d6f 698 blkCnt = numPoints & 0x3;
emh203 0:3d9c67d97d6f 699
emh203 0:3d9c67d97d6f 700 /* Calculate convolution for remaining samples of Bigger length sequence */
emh203 0:3d9c67d97d6f 701 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 702 {
emh203 0:3d9c67d97d6f 703 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 704 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 705
emh203 0:3d9c67d97d6f 706 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 707 acc0 = 0;
emh203 0:3d9c67d97d6f 708
emh203 0:3d9c67d97d6f 709 tapCnt = (srcBLen) >> 1u;
emh203 0:3d9c67d97d6f 710
emh203 0:3d9c67d97d6f 711 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 712 {
emh203 0:3d9c67d97d6f 713
emh203 0:3d9c67d97d6f 714 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 715 x10 = *pScr1++;
emh203 0:3d9c67d97d6f 716 x11 = *pScr1++;
emh203 0:3d9c67d97d6f 717
emh203 0:3d9c67d97d6f 718 /* Read two samples from smaller buffer */
emh203 0:3d9c67d97d6f 719 y10 = *pIn2++;
emh203 0:3d9c67d97d6f 720 y11 = *pIn2++;
emh203 0:3d9c67d97d6f 721
emh203 0:3d9c67d97d6f 722 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 723 acc0 += (q31_t) x10 *y10;
emh203 0:3d9c67d97d6f 724 acc0 += (q31_t) x11 *y11;
emh203 0:3d9c67d97d6f 725
emh203 0:3d9c67d97d6f 726 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 727 tapCnt--;
emh203 0:3d9c67d97d6f 728 }
emh203 0:3d9c67d97d6f 729
emh203 0:3d9c67d97d6f 730 tapCnt = (srcBLen) & 1u;
emh203 0:3d9c67d97d6f 731
emh203 0:3d9c67d97d6f 732 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 733 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 734 {
emh203 0:3d9c67d97d6f 735
emh203 0:3d9c67d97d6f 736 /* accumlate the results */
emh203 0:3d9c67d97d6f 737 acc0 += (*pScr1++ * *pIn2++);
emh203 0:3d9c67d97d6f 738
emh203 0:3d9c67d97d6f 739 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 740 tapCnt--;
emh203 0:3d9c67d97d6f 741 }
emh203 0:3d9c67d97d6f 742
emh203 0:3d9c67d97d6f 743 blkCnt--;
emh203 0:3d9c67d97d6f 744
emh203 0:3d9c67d97d6f 745 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 746 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emh203 0:3d9c67d97d6f 747
emh203 0:3d9c67d97d6f 748 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 749 pIn2 = py;
emh203 0:3d9c67d97d6f 750
emh203 0:3d9c67d97d6f 751 pScratch1 += 1u;
emh203 0:3d9c67d97d6f 752
emh203 0:3d9c67d97d6f 753 }
emh203 0:3d9c67d97d6f 754
emh203 0:3d9c67d97d6f 755 /* set status as ARM_MATH_SUCCESS */
emh203 0:3d9c67d97d6f 756 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 757
emh203 0:3d9c67d97d6f 758 }
emh203 0:3d9c67d97d6f 759
emh203 0:3d9c67d97d6f 760 /* Return to application */
emh203 0:3d9c67d97d6f 761 return (status);
emh203 0:3d9c67d97d6f 762 }
emh203 0:3d9c67d97d6f 763
emh203 0:3d9c67d97d6f 764 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emh203 0:3d9c67d97d6f 765
emh203 0:3d9c67d97d6f 766 /**
emh203 0:3d9c67d97d6f 767 * @} end of PartialConv group
emh203 0:3d9c67d97d6f 768 */