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_opt_q7.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_opt_q7.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Convolution of Q7 sequences.
emh203 0:3d9c67d97d6f 11 *
emh203 0:3d9c67d97d6f 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
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 Conv
emh203 0:3d9c67d97d6f 49 * @{
emh203 0:3d9c67d97d6f 50 */
emh203 0:3d9c67d97d6f 51
emh203 0:3d9c67d97d6f 52 /**
emh203 0:3d9c67d97d6f 53 * @brief Convolution of Q7 sequences.
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. Length srcALen+srcBLen-1.
emh203 0:3d9c67d97d6f 59 * @param[in] *pScratch1 points to scratch buffer(of type q15_t) of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emh203 0:3d9c67d97d6f 60 * @param[in] *pScratch2 points to scratch buffer (of type q15_t) of size min(srcALen, srcBLen).
emh203 0:3d9c67d97d6f 61 * @return none.
emh203 0:3d9c67d97d6f 62 *
emh203 0:3d9c67d97d6f 63 * \par Restrictions
emh203 0:3d9c67d97d6f 64 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emh203 0:3d9c67d97d6f 65 * In this case input, output, scratch1 and scratch2 buffers should be aligned by 32-bit
emh203 0:3d9c67d97d6f 66 *
emh203 0:3d9c67d97d6f 67 * @details
emh203 0:3d9c67d97d6f 68 * <b>Scaling and Overflow Behavior:</b>
emh203 0:3d9c67d97d6f 69 *
emh203 0:3d9c67d97d6f 70 * \par
emh203 0:3d9c67d97d6f 71 * The function is implemented using a 32-bit internal accumulator.
emh203 0:3d9c67d97d6f 72 * Both the inputs are represented in 1.7 format and multiplications yield a 2.14 result.
emh203 0:3d9c67d97d6f 73 * The 2.14 intermediate results are accumulated in a 32-bit accumulator in 18.14 format.
emh203 0:3d9c67d97d6f 74 * This approach provides 17 guard bits and there is no risk of overflow as long as <code>max(srcALen, srcBLen)<131072</code>.
emh203 0:3d9c67d97d6f 75 * The 18.14 result is then truncated to 18.7 format by discarding the low 7 bits and then saturated to 1.7 format.
emh203 0:3d9c67d97d6f 76 *
emh203 0:3d9c67d97d6f 77 */
emh203 0:3d9c67d97d6f 78
emh203 0:3d9c67d97d6f 79 void arm_conv_opt_q7(
emh203 0:3d9c67d97d6f 80 q7_t * pSrcA,
emh203 0:3d9c67d97d6f 81 uint32_t srcALen,
emh203 0:3d9c67d97d6f 82 q7_t * pSrcB,
emh203 0:3d9c67d97d6f 83 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 84 q7_t * pDst,
emh203 0:3d9c67d97d6f 85 q15_t * pScratch1,
emh203 0:3d9c67d97d6f 86 q15_t * pScratch2)
emh203 0:3d9c67d97d6f 87 {
emh203 0:3d9c67d97d6f 88
emh203 0:3d9c67d97d6f 89 q15_t *pScr2, *pScr1; /* Intermediate pointers for scratch pointers */
emh203 0:3d9c67d97d6f 90 q15_t x4; /* Temporary input variable */
emh203 0:3d9c67d97d6f 91 q7_t *pIn1, *pIn2; /* inputA and inputB pointer */
emh203 0:3d9c67d97d6f 92 uint32_t j, k, blkCnt, tapCnt; /* loop counter */
emh203 0:3d9c67d97d6f 93 q7_t *px; /* Temporary input1 pointer */
emh203 0:3d9c67d97d6f 94 q15_t *py; /* Temporary input2 pointer */
emh203 0:3d9c67d97d6f 95 q31_t acc0, acc1, acc2, acc3; /* Accumulator */
emh203 0:3d9c67d97d6f 96 q31_t x1, x2, x3, y1; /* Temporary input variables */
emh203 0:3d9c67d97d6f 97 q7_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 98 q7_t out0, out1, out2, out3; /* temporary variables */
emh203 0:3d9c67d97d6f 99
emh203 0:3d9c67d97d6f 100 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 101 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 102 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 103 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 104 {
emh203 0:3d9c67d97d6f 105 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 106 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 107
emh203 0:3d9c67d97d6f 108 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 109 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 110 }
emh203 0:3d9c67d97d6f 111 else
emh203 0:3d9c67d97d6f 112 {
emh203 0:3d9c67d97d6f 113 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 114 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 115
emh203 0:3d9c67d97d6f 116 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 117 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 118
emh203 0:3d9c67d97d6f 119 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 120 j = srcBLen;
emh203 0:3d9c67d97d6f 121 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 122 srcALen = j;
emh203 0:3d9c67d97d6f 123 }
emh203 0:3d9c67d97d6f 124
emh203 0:3d9c67d97d6f 125 /* pointer to take end of scratch2 buffer */
emh203 0:3d9c67d97d6f 126 pScr2 = pScratch2;
emh203 0:3d9c67d97d6f 127
emh203 0:3d9c67d97d6f 128 /* points to smaller length sequence */
emh203 0:3d9c67d97d6f 129 px = pIn2 + srcBLen - 1;
emh203 0:3d9c67d97d6f 130
emh203 0:3d9c67d97d6f 131 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 132 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 133
emh203 0:3d9c67d97d6f 134 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 135 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 136 while(k > 0u)
emh203 0:3d9c67d97d6f 137 {
emh203 0:3d9c67d97d6f 138 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 139 x4 = (q15_t) * px--;
emh203 0:3d9c67d97d6f 140 *pScr2++ = x4;
emh203 0:3d9c67d97d6f 141 x4 = (q15_t) * px--;
emh203 0:3d9c67d97d6f 142 *pScr2++ = x4;
emh203 0:3d9c67d97d6f 143 x4 = (q15_t) * px--;
emh203 0:3d9c67d97d6f 144 *pScr2++ = x4;
emh203 0:3d9c67d97d6f 145 x4 = (q15_t) * px--;
emh203 0:3d9c67d97d6f 146 *pScr2++ = x4;
emh203 0:3d9c67d97d6f 147
emh203 0:3d9c67d97d6f 148 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 149 k--;
emh203 0:3d9c67d97d6f 150 }
emh203 0:3d9c67d97d6f 151
emh203 0:3d9c67d97d6f 152 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 153 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 154 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 155
emh203 0:3d9c67d97d6f 156 while(k > 0u)
emh203 0:3d9c67d97d6f 157 {
emh203 0:3d9c67d97d6f 158 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 159 x4 = (q15_t) * px--;
emh203 0:3d9c67d97d6f 160 *pScr2++ = x4;
emh203 0:3d9c67d97d6f 161
emh203 0:3d9c67d97d6f 162 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 163 k--;
emh203 0:3d9c67d97d6f 164 }
emh203 0:3d9c67d97d6f 165
emh203 0:3d9c67d97d6f 166 /* Initialze temporary scratch pointer */
emh203 0:3d9c67d97d6f 167 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 168
emh203 0:3d9c67d97d6f 169 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emh203 0:3d9c67d97d6f 170 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emh203 0:3d9c67d97d6f 171
emh203 0:3d9c67d97d6f 172 /* Update temporary scratch pointer */
emh203 0:3d9c67d97d6f 173 pScr1 += (srcBLen - 1u);
emh203 0:3d9c67d97d6f 174
emh203 0:3d9c67d97d6f 175 /* Copy (srcALen) samples in scratch buffer */
emh203 0:3d9c67d97d6f 176 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 177 k = srcALen >> 2u;
emh203 0:3d9c67d97d6f 178
emh203 0:3d9c67d97d6f 179 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 180 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 181 while(k > 0u)
emh203 0:3d9c67d97d6f 182 {
emh203 0:3d9c67d97d6f 183 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 184 x4 = (q15_t) * pIn1++;
emh203 0:3d9c67d97d6f 185 *pScr1++ = x4;
emh203 0:3d9c67d97d6f 186 x4 = (q15_t) * pIn1++;
emh203 0:3d9c67d97d6f 187 *pScr1++ = x4;
emh203 0:3d9c67d97d6f 188 x4 = (q15_t) * pIn1++;
emh203 0:3d9c67d97d6f 189 *pScr1++ = x4;
emh203 0:3d9c67d97d6f 190 x4 = (q15_t) * pIn1++;
emh203 0:3d9c67d97d6f 191 *pScr1++ = x4;
emh203 0:3d9c67d97d6f 192
emh203 0:3d9c67d97d6f 193 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 194 k--;
emh203 0:3d9c67d97d6f 195 }
emh203 0:3d9c67d97d6f 196
emh203 0:3d9c67d97d6f 197 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 198 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 199 k = srcALen % 0x4u;
emh203 0:3d9c67d97d6f 200
emh203 0:3d9c67d97d6f 201 while(k > 0u)
emh203 0:3d9c67d97d6f 202 {
emh203 0:3d9c67d97d6f 203 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 204 x4 = (q15_t) * pIn1++;
emh203 0:3d9c67d97d6f 205 *pScr1++ = x4;
emh203 0:3d9c67d97d6f 206
emh203 0:3d9c67d97d6f 207 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 208 k--;
emh203 0:3d9c67d97d6f 209 }
emh203 0:3d9c67d97d6f 210
emh203 0:3d9c67d97d6f 211 #ifndef UNALIGNED_SUPPORT_DISABLE
emh203 0:3d9c67d97d6f 212
emh203 0:3d9c67d97d6f 213 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emh203 0:3d9c67d97d6f 214 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emh203 0:3d9c67d97d6f 215
emh203 0:3d9c67d97d6f 216 /* Update pointer */
emh203 0:3d9c67d97d6f 217 pScr1 += (srcBLen - 1u);
emh203 0:3d9c67d97d6f 218
emh203 0:3d9c67d97d6f 219 #else
emh203 0:3d9c67d97d6f 220
emh203 0:3d9c67d97d6f 221 /* Apply loop unrolling and do 4 Copies simultaneously. */
emh203 0:3d9c67d97d6f 222 k = (srcBLen - 1u) >> 2u;
emh203 0:3d9c67d97d6f 223
emh203 0:3d9c67d97d6f 224 /* First part of the processing with loop unrolling copies 4 data points at a time.
emh203 0:3d9c67d97d6f 225 ** a second loop below copies for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 226 while(k > 0u)
emh203 0:3d9c67d97d6f 227 {
emh203 0:3d9c67d97d6f 228 /* copy second buffer in reversal manner */
emh203 0:3d9c67d97d6f 229 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 230 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 231 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 232 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 233
emh203 0:3d9c67d97d6f 234 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 235 k--;
emh203 0:3d9c67d97d6f 236 }
emh203 0:3d9c67d97d6f 237
emh203 0:3d9c67d97d6f 238 /* If the count is not a multiple of 4, copy remaining samples here.
emh203 0:3d9c67d97d6f 239 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 240 k = (srcBLen - 1u) % 0x4u;
emh203 0:3d9c67d97d6f 241
emh203 0:3d9c67d97d6f 242 while(k > 0u)
emh203 0:3d9c67d97d6f 243 {
emh203 0:3d9c67d97d6f 244 /* copy second buffer in reversal manner for remaining samples */
emh203 0:3d9c67d97d6f 245 *pScr1++ = 0;
emh203 0:3d9c67d97d6f 246
emh203 0:3d9c67d97d6f 247 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 248 k--;
emh203 0:3d9c67d97d6f 249 }
emh203 0:3d9c67d97d6f 250
emh203 0:3d9c67d97d6f 251 #endif
emh203 0:3d9c67d97d6f 252
emh203 0:3d9c67d97d6f 253 /* Temporary pointer for scratch2 */
emh203 0:3d9c67d97d6f 254 py = pScratch2;
emh203 0:3d9c67d97d6f 255
emh203 0:3d9c67d97d6f 256 /* Initialization of pIn2 pointer */
emh203 0:3d9c67d97d6f 257 pIn2 = (q7_t *) py;
emh203 0:3d9c67d97d6f 258
emh203 0:3d9c67d97d6f 259 pScr2 = py;
emh203 0:3d9c67d97d6f 260
emh203 0:3d9c67d97d6f 261 /* Actual convolution process starts here */
emh203 0:3d9c67d97d6f 262 blkCnt = (srcALen + srcBLen - 1u) >> 2;
emh203 0:3d9c67d97d6f 263
emh203 0:3d9c67d97d6f 264 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 265 {
emh203 0:3d9c67d97d6f 266 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 267 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 268
emh203 0:3d9c67d97d6f 269 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 270 acc0 = 0;
emh203 0:3d9c67d97d6f 271 acc1 = 0;
emh203 0:3d9c67d97d6f 272 acc2 = 0;
emh203 0:3d9c67d97d6f 273 acc3 = 0;
emh203 0:3d9c67d97d6f 274
emh203 0:3d9c67d97d6f 275 /* Read two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 276 x1 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 277
emh203 0:3d9c67d97d6f 278 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 279 x2 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 280
emh203 0:3d9c67d97d6f 281 tapCnt = (srcBLen) >> 2u;
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 284 {
emh203 0:3d9c67d97d6f 285
emh203 0:3d9c67d97d6f 286 /* Read four samples from smaller buffer */
emh203 0:3d9c67d97d6f 287 y1 = _SIMD32_OFFSET(pScr2);
emh203 0:3d9c67d97d6f 288
emh203 0:3d9c67d97d6f 289 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 290 acc0 = __SMLAD(x1, y1, acc0);
emh203 0:3d9c67d97d6f 291 acc2 = __SMLAD(x2, y1, acc2);
emh203 0:3d9c67d97d6f 292
emh203 0:3d9c67d97d6f 293 /* pack input data */
emh203 0:3d9c67d97d6f 294 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 295 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 296 #else
emh203 0:3d9c67d97d6f 297 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 298 #endif
emh203 0:3d9c67d97d6f 299
emh203 0:3d9c67d97d6f 300 /* multiply and accumlate */
emh203 0:3d9c67d97d6f 301 acc1 = __SMLADX(x3, y1, acc1);
emh203 0:3d9c67d97d6f 302
emh203 0:3d9c67d97d6f 303 /* Read next two samples from scratch1 buffer */
emh203 0:3d9c67d97d6f 304 x1 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 305
emh203 0:3d9c67d97d6f 306 /* pack input data */
emh203 0:3d9c67d97d6f 307 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 308 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 309 #else
emh203 0:3d9c67d97d6f 310 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 311 #endif
emh203 0:3d9c67d97d6f 312
emh203 0:3d9c67d97d6f 313 acc3 = __SMLADX(x3, y1, acc3);
emh203 0:3d9c67d97d6f 314
emh203 0:3d9c67d97d6f 315 /* Read four samples from smaller buffer */
emh203 0:3d9c67d97d6f 316 y1 = _SIMD32_OFFSET(pScr2 + 2u);
emh203 0:3d9c67d97d6f 317
emh203 0:3d9c67d97d6f 318 acc0 = __SMLAD(x2, y1, acc0);
emh203 0:3d9c67d97d6f 319
emh203 0:3d9c67d97d6f 320 acc2 = __SMLAD(x1, y1, acc2);
emh203 0:3d9c67d97d6f 321
emh203 0:3d9c67d97d6f 322 acc1 = __SMLADX(x3, y1, acc1);
emh203 0:3d9c67d97d6f 323
emh203 0:3d9c67d97d6f 324 x2 = *__SIMD32(pScr1)++;
emh203 0:3d9c67d97d6f 325
emh203 0:3d9c67d97d6f 326 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 327 x3 = __PKHBT(x2, x1, 0);
emh203 0:3d9c67d97d6f 328 #else
emh203 0:3d9c67d97d6f 329 x3 = __PKHBT(x1, x2, 0);
emh203 0:3d9c67d97d6f 330 #endif
emh203 0:3d9c67d97d6f 331
emh203 0:3d9c67d97d6f 332 acc3 = __SMLADX(x3, y1, acc3);
emh203 0:3d9c67d97d6f 333
emh203 0:3d9c67d97d6f 334 pScr2 += 4u;
emh203 0:3d9c67d97d6f 335
emh203 0:3d9c67d97d6f 336
emh203 0:3d9c67d97d6f 337 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 338 tapCnt--;
emh203 0:3d9c67d97d6f 339 }
emh203 0:3d9c67d97d6f 340
emh203 0:3d9c67d97d6f 341
emh203 0:3d9c67d97d6f 342
emh203 0:3d9c67d97d6f 343 /* Update scratch pointer for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 344 pScr1 -= 4u;
emh203 0:3d9c67d97d6f 345
emh203 0:3d9c67d97d6f 346
emh203 0:3d9c67d97d6f 347 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 348 tapCnt = (srcBLen) & 3u;
emh203 0:3d9c67d97d6f 349
emh203 0:3d9c67d97d6f 350 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 351 {
emh203 0:3d9c67d97d6f 352
emh203 0:3d9c67d97d6f 353 /* accumlate the results */
emh203 0:3d9c67d97d6f 354 acc0 += (*pScr1++ * *pScr2);
emh203 0:3d9c67d97d6f 355 acc1 += (*pScr1++ * *pScr2);
emh203 0:3d9c67d97d6f 356 acc2 += (*pScr1++ * *pScr2);
emh203 0:3d9c67d97d6f 357 acc3 += (*pScr1++ * *pScr2++);
emh203 0:3d9c67d97d6f 358
emh203 0:3d9c67d97d6f 359 pScr1 -= 3u;
emh203 0:3d9c67d97d6f 360
emh203 0:3d9c67d97d6f 361 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 362 tapCnt--;
emh203 0:3d9c67d97d6f 363 }
emh203 0:3d9c67d97d6f 364
emh203 0:3d9c67d97d6f 365 blkCnt--;
emh203 0:3d9c67d97d6f 366
emh203 0:3d9c67d97d6f 367 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 368 out0 = (q7_t) (__SSAT(acc0 >> 7u, 8));
emh203 0:3d9c67d97d6f 369 out1 = (q7_t) (__SSAT(acc1 >> 7u, 8));
emh203 0:3d9c67d97d6f 370 out2 = (q7_t) (__SSAT(acc2 >> 7u, 8));
emh203 0:3d9c67d97d6f 371 out3 = (q7_t) (__SSAT(acc3 >> 7u, 8));
emh203 0:3d9c67d97d6f 372
emh203 0:3d9c67d97d6f 373 *__SIMD32(pOut)++ = __PACKq7(out0, out1, out2, out3);
emh203 0:3d9c67d97d6f 374
emh203 0:3d9c67d97d6f 375 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 376 pScr2 = py;
emh203 0:3d9c67d97d6f 377
emh203 0:3d9c67d97d6f 378 pScratch1 += 4u;
emh203 0:3d9c67d97d6f 379
emh203 0:3d9c67d97d6f 380 }
emh203 0:3d9c67d97d6f 381
emh203 0:3d9c67d97d6f 382
emh203 0:3d9c67d97d6f 383 blkCnt = (srcALen + srcBLen - 1u) & 0x3;
emh203 0:3d9c67d97d6f 384
emh203 0:3d9c67d97d6f 385 /* Calculate convolution for remaining samples of Bigger length sequence */
emh203 0:3d9c67d97d6f 386 while(blkCnt > 0)
emh203 0:3d9c67d97d6f 387 {
emh203 0:3d9c67d97d6f 388 /* Initialze temporary scratch pointer as scratch1 */
emh203 0:3d9c67d97d6f 389 pScr1 = pScratch1;
emh203 0:3d9c67d97d6f 390
emh203 0:3d9c67d97d6f 391 /* Clear Accumlators */
emh203 0:3d9c67d97d6f 392 acc0 = 0;
emh203 0:3d9c67d97d6f 393
emh203 0:3d9c67d97d6f 394 tapCnt = (srcBLen) >> 1u;
emh203 0:3d9c67d97d6f 395
emh203 0:3d9c67d97d6f 396 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 397 {
emh203 0:3d9c67d97d6f 398 acc0 += (*pScr1++ * *pScr2++);
emh203 0:3d9c67d97d6f 399 acc0 += (*pScr1++ * *pScr2++);
emh203 0:3d9c67d97d6f 400
emh203 0:3d9c67d97d6f 401 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 402 tapCnt--;
emh203 0:3d9c67d97d6f 403 }
emh203 0:3d9c67d97d6f 404
emh203 0:3d9c67d97d6f 405 tapCnt = (srcBLen) & 1u;
emh203 0:3d9c67d97d6f 406
emh203 0:3d9c67d97d6f 407 /* apply same above for remaining samples of smaller length sequence */
emh203 0:3d9c67d97d6f 408 while(tapCnt > 0u)
emh203 0:3d9c67d97d6f 409 {
emh203 0:3d9c67d97d6f 410
emh203 0:3d9c67d97d6f 411 /* accumlate the results */
emh203 0:3d9c67d97d6f 412 acc0 += (*pScr1++ * *pScr2++);
emh203 0:3d9c67d97d6f 413
emh203 0:3d9c67d97d6f 414 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 415 tapCnt--;
emh203 0:3d9c67d97d6f 416 }
emh203 0:3d9c67d97d6f 417
emh203 0:3d9c67d97d6f 418 blkCnt--;
emh203 0:3d9c67d97d6f 419
emh203 0:3d9c67d97d6f 420 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 421 *pOut++ = (q7_t) (__SSAT(acc0 >> 7u, 8));
emh203 0:3d9c67d97d6f 422
emh203 0:3d9c67d97d6f 423 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 424 pScr2 = py;
emh203 0:3d9c67d97d6f 425
emh203 0:3d9c67d97d6f 426 pScratch1 += 1u;
emh203 0:3d9c67d97d6f 427
emh203 0:3d9c67d97d6f 428 }
emh203 0:3d9c67d97d6f 429
emh203 0:3d9c67d97d6f 430 }
emh203 0:3d9c67d97d6f 431
emh203 0:3d9c67d97d6f 432
emh203 0:3d9c67d97d6f 433 /**
emh203 0:3d9c67d97d6f 434 * @} end of Conv group
emh203 0:3d9c67d97d6f 435 */