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_q31.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_q31.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Partial convolution of Q31 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 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 Q31 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.
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 * @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 62 *
emh203 0:3d9c67d97d6f 63 * See <code>arm_conv_partial_fast_q31()</code> for a faster but less precise implementation of this function for Cortex-M3 and Cortex-M4.
emh203 0:3d9c67d97d6f 64 */
emh203 0:3d9c67d97d6f 65
emh203 0:3d9c67d97d6f 66 arm_status arm_conv_partial_q31(
emh203 0:3d9c67d97d6f 67 q31_t * pSrcA,
emh203 0:3d9c67d97d6f 68 uint32_t srcALen,
emh203 0:3d9c67d97d6f 69 q31_t * pSrcB,
emh203 0:3d9c67d97d6f 70 uint32_t srcBLen,
emh203 0:3d9c67d97d6f 71 q31_t * pDst,
emh203 0:3d9c67d97d6f 72 uint32_t firstIndex,
emh203 0:3d9c67d97d6f 73 uint32_t numPoints)
emh203 0:3d9c67d97d6f 74 {
emh203 0:3d9c67d97d6f 75
emh203 0:3d9c67d97d6f 76
emh203 0:3d9c67d97d6f 77 #ifndef ARM_MATH_CM0_FAMILY
emh203 0:3d9c67d97d6f 78
emh203 0:3d9c67d97d6f 79 /* Run the below code for Cortex-M4 and Cortex-M3 */
emh203 0:3d9c67d97d6f 80
emh203 0:3d9c67d97d6f 81 q31_t *pIn1; /* inputA pointer */
emh203 0:3d9c67d97d6f 82 q31_t *pIn2; /* inputB pointer */
emh203 0:3d9c67d97d6f 83 q31_t *pOut = pDst; /* output pointer */
emh203 0:3d9c67d97d6f 84 q31_t *px; /* Intermediate inputA pointer */
emh203 0:3d9c67d97d6f 85 q31_t *py; /* Intermediate inputB pointer */
emh203 0:3d9c67d97d6f 86 q31_t *pSrc1, *pSrc2; /* Intermediate pointers */
emh203 0:3d9c67d97d6f 87 q63_t sum, acc0, acc1, acc2; /* Accumulator */
emh203 0:3d9c67d97d6f 88 q31_t x0, x1, x2, c0;
emh203 0:3d9c67d97d6f 89 uint32_t j, k, count, check, blkCnt;
emh203 0:3d9c67d97d6f 90 int32_t blockSize1, blockSize2, blockSize3; /* loop counter */
emh203 0:3d9c67d97d6f 91 arm_status status; /* status of Partial convolution */
emh203 0:3d9c67d97d6f 92
emh203 0:3d9c67d97d6f 93
emh203 0:3d9c67d97d6f 94 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 95 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 96 {
emh203 0:3d9c67d97d6f 97 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 98 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 99 }
emh203 0:3d9c67d97d6f 100 else
emh203 0:3d9c67d97d6f 101 {
emh203 0:3d9c67d97d6f 102
emh203 0:3d9c67d97d6f 103 /* The algorithm implementation is based on the lengths of the inputs. */
emh203 0:3d9c67d97d6f 104 /* srcB is always made to slide across srcA. */
emh203 0:3d9c67d97d6f 105 /* So srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 106 if(srcALen >= srcBLen)
emh203 0:3d9c67d97d6f 107 {
emh203 0:3d9c67d97d6f 108 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 109 pIn1 = pSrcA;
emh203 0:3d9c67d97d6f 110
emh203 0:3d9c67d97d6f 111 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 112 pIn2 = pSrcB;
emh203 0:3d9c67d97d6f 113 }
emh203 0:3d9c67d97d6f 114 else
emh203 0:3d9c67d97d6f 115 {
emh203 0:3d9c67d97d6f 116 /* Initialization of inputA pointer */
emh203 0:3d9c67d97d6f 117 pIn1 = pSrcB;
emh203 0:3d9c67d97d6f 118
emh203 0:3d9c67d97d6f 119 /* Initialization of inputB pointer */
emh203 0:3d9c67d97d6f 120 pIn2 = pSrcA;
emh203 0:3d9c67d97d6f 121
emh203 0:3d9c67d97d6f 122 /* srcBLen is always considered as shorter or equal to srcALen */
emh203 0:3d9c67d97d6f 123 j = srcBLen;
emh203 0:3d9c67d97d6f 124 srcBLen = srcALen;
emh203 0:3d9c67d97d6f 125 srcALen = j;
emh203 0:3d9c67d97d6f 126 }
emh203 0:3d9c67d97d6f 127
emh203 0:3d9c67d97d6f 128 /* Conditions to check which loopCounter holds
emh203 0:3d9c67d97d6f 129 * the first and last indices of the output samples to be calculated. */
emh203 0:3d9c67d97d6f 130 check = firstIndex + numPoints;
emh203 0:3d9c67d97d6f 131 blockSize3 = ((int32_t)check > (int32_t)srcALen) ? (int32_t)check - (int32_t)srcALen : 0;
emh203 0:3d9c67d97d6f 132 blockSize3 = ((int32_t)firstIndex > (int32_t)srcALen - 1) ? blockSize3 - (int32_t)firstIndex + (int32_t)srcALen : blockSize3;
emh203 0:3d9c67d97d6f 133 blockSize1 = (((int32_t) srcBLen - 1) - (int32_t) firstIndex);
emh203 0:3d9c67d97d6f 134 blockSize1 = (blockSize1 > 0) ? ((check > (srcBLen - 1u)) ? blockSize1 :
emh203 0:3d9c67d97d6f 135 (int32_t) numPoints) : 0;
emh203 0:3d9c67d97d6f 136 blockSize2 = (int32_t) check - ((blockSize3 + blockSize1) +
emh203 0:3d9c67d97d6f 137 (int32_t) firstIndex);
emh203 0:3d9c67d97d6f 138 blockSize2 = (blockSize2 > 0) ? blockSize2 : 0;
emh203 0:3d9c67d97d6f 139
emh203 0:3d9c67d97d6f 140 /* conv(x,y) at n = x[n] * y[0] + x[n-1] * y[1] + x[n-2] * y[2] + ...+ x[n-N+1] * y[N -1] */
emh203 0:3d9c67d97d6f 141 /* The function is internally
emh203 0:3d9c67d97d6f 142 * divided into three stages according to the number of multiplications that has to be
emh203 0:3d9c67d97d6f 143 * taken place between inputA samples and inputB samples. In the first stage of the
emh203 0:3d9c67d97d6f 144 * algorithm, the multiplications increase by one for every iteration.
emh203 0:3d9c67d97d6f 145 * In the second stage of the algorithm, srcBLen number of multiplications are done.
emh203 0:3d9c67d97d6f 146 * In the third stage of the algorithm, the multiplications decrease by one
emh203 0:3d9c67d97d6f 147 * for every iteration. */
emh203 0:3d9c67d97d6f 148
emh203 0:3d9c67d97d6f 149 /* Set the output pointer to point to the firstIndex
emh203 0:3d9c67d97d6f 150 * of the output sample to be calculated. */
emh203 0:3d9c67d97d6f 151 pOut = pDst + firstIndex;
emh203 0:3d9c67d97d6f 152
emh203 0:3d9c67d97d6f 153 /* --------------------------
emh203 0:3d9c67d97d6f 154 * Initializations of stage1
emh203 0:3d9c67d97d6f 155 * -------------------------*/
emh203 0:3d9c67d97d6f 156
emh203 0:3d9c67d97d6f 157 /* sum = x[0] * y[0]
emh203 0:3d9c67d97d6f 158 * sum = x[0] * y[1] + x[1] * y[0]
emh203 0:3d9c67d97d6f 159 * ....
emh203 0:3d9c67d97d6f 160 * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0]
emh203 0:3d9c67d97d6f 161 */
emh203 0:3d9c67d97d6f 162
emh203 0:3d9c67d97d6f 163 /* In this stage the MAC operations are increased by 1 for every iteration.
emh203 0:3d9c67d97d6f 164 The count variable holds the number of MAC operations performed.
emh203 0:3d9c67d97d6f 165 Since the partial convolution starts from firstIndex
emh203 0:3d9c67d97d6f 166 Number of Macs to be performed is firstIndex + 1 */
emh203 0:3d9c67d97d6f 167 count = 1u + firstIndex;
emh203 0:3d9c67d97d6f 168
emh203 0:3d9c67d97d6f 169 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 170 px = pIn1;
emh203 0:3d9c67d97d6f 171
emh203 0:3d9c67d97d6f 172 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 173 pSrc2 = pIn2 + firstIndex;
emh203 0:3d9c67d97d6f 174 py = pSrc2;
emh203 0:3d9c67d97d6f 175
emh203 0:3d9c67d97d6f 176 /* ------------------------
emh203 0:3d9c67d97d6f 177 * Stage1 process
emh203 0:3d9c67d97d6f 178 * ----------------------*/
emh203 0:3d9c67d97d6f 179
emh203 0:3d9c67d97d6f 180 /* The first loop starts here */
emh203 0:3d9c67d97d6f 181 while(blockSize1 > 0)
emh203 0:3d9c67d97d6f 182 {
emh203 0:3d9c67d97d6f 183 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 184 sum = 0;
emh203 0:3d9c67d97d6f 185
emh203 0:3d9c67d97d6f 186 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 187 k = count >> 2u;
emh203 0:3d9c67d97d6f 188
emh203 0:3d9c67d97d6f 189 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 190 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 191 while(k > 0u)
emh203 0:3d9c67d97d6f 192 {
emh203 0:3d9c67d97d6f 193 /* x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 194 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 195 /* x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 196 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 197 /* x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 198 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 199 /* x[3] * y[srcBLen - 4] */
emh203 0:3d9c67d97d6f 200 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 201
emh203 0:3d9c67d97d6f 202 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 203 k--;
emh203 0:3d9c67d97d6f 204 }
emh203 0:3d9c67d97d6f 205
emh203 0:3d9c67d97d6f 206 /* If the count is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 207 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 208 k = count % 0x4u;
emh203 0:3d9c67d97d6f 209
emh203 0:3d9c67d97d6f 210 while(k > 0u)
emh203 0:3d9c67d97d6f 211 {
emh203 0:3d9c67d97d6f 212 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 213 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 214
emh203 0:3d9c67d97d6f 215 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 216 k--;
emh203 0:3d9c67d97d6f 217 }
emh203 0:3d9c67d97d6f 218
emh203 0:3d9c67d97d6f 219 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 220 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 221
emh203 0:3d9c67d97d6f 222 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 223 py = ++pSrc2;
emh203 0:3d9c67d97d6f 224 px = pIn1;
emh203 0:3d9c67d97d6f 225
emh203 0:3d9c67d97d6f 226 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 227 count++;
emh203 0:3d9c67d97d6f 228
emh203 0:3d9c67d97d6f 229 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 230 blockSize1--;
emh203 0:3d9c67d97d6f 231 }
emh203 0:3d9c67d97d6f 232
emh203 0:3d9c67d97d6f 233 /* --------------------------
emh203 0:3d9c67d97d6f 234 * Initializations of stage2
emh203 0:3d9c67d97d6f 235 * ------------------------*/
emh203 0:3d9c67d97d6f 236
emh203 0:3d9c67d97d6f 237 /* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0]
emh203 0:3d9c67d97d6f 238 * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0]
emh203 0:3d9c67d97d6f 239 * ....
emh203 0:3d9c67d97d6f 240 * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0]
emh203 0:3d9c67d97d6f 241 */
emh203 0:3d9c67d97d6f 242
emh203 0:3d9c67d97d6f 243 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 244 if((int32_t)firstIndex - (int32_t)srcBLen + 1 > 0)
emh203 0:3d9c67d97d6f 245 {
emh203 0:3d9c67d97d6f 246 px = pIn1 + firstIndex - srcBLen + 1;
emh203 0:3d9c67d97d6f 247 }
emh203 0:3d9c67d97d6f 248 else
emh203 0:3d9c67d97d6f 249 {
emh203 0:3d9c67d97d6f 250 px = pIn1;
emh203 0:3d9c67d97d6f 251 }
emh203 0:3d9c67d97d6f 252
emh203 0:3d9c67d97d6f 253 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 254 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 255 py = pSrc2;
emh203 0:3d9c67d97d6f 256
emh203 0:3d9c67d97d6f 257 /* count is index by which the pointer pIn1 to be incremented */
emh203 0:3d9c67d97d6f 258 count = 0u;
emh203 0:3d9c67d97d6f 259
emh203 0:3d9c67d97d6f 260 /* -------------------
emh203 0:3d9c67d97d6f 261 * Stage2 process
emh203 0:3d9c67d97d6f 262 * ------------------*/
emh203 0:3d9c67d97d6f 263
emh203 0:3d9c67d97d6f 264 /* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed.
emh203 0:3d9c67d97d6f 265 * So, to loop unroll over blockSize2,
emh203 0:3d9c67d97d6f 266 * srcBLen should be greater than or equal to 4 */
emh203 0:3d9c67d97d6f 267 if(srcBLen >= 4u)
emh203 0:3d9c67d97d6f 268 {
emh203 0:3d9c67d97d6f 269 /* Loop unroll over blkCnt */
emh203 0:3d9c67d97d6f 270
emh203 0:3d9c67d97d6f 271 blkCnt = blockSize2 / 3;
emh203 0:3d9c67d97d6f 272 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 273 {
emh203 0:3d9c67d97d6f 274 /* Set all accumulators to zero */
emh203 0:3d9c67d97d6f 275 acc0 = 0;
emh203 0:3d9c67d97d6f 276 acc1 = 0;
emh203 0:3d9c67d97d6f 277 acc2 = 0;
emh203 0:3d9c67d97d6f 278
emh203 0:3d9c67d97d6f 279 /* read x[0], x[1] samples */
emh203 0:3d9c67d97d6f 280 x0 = *(px++);
emh203 0:3d9c67d97d6f 281 x1 = *(px++);
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283 /* Apply loop unrolling and compute 3 MACs simultaneously. */
emh203 0:3d9c67d97d6f 284 k = srcBLen / 3;
emh203 0:3d9c67d97d6f 285
emh203 0:3d9c67d97d6f 286 /* First part of the processing with loop unrolling. Compute 3 MACs at a time.
emh203 0:3d9c67d97d6f 287 ** a second loop below computes MACs for the remaining 1 to 2 samples. */
emh203 0:3d9c67d97d6f 288 do
emh203 0:3d9c67d97d6f 289 {
emh203 0:3d9c67d97d6f 290 /* Read y[srcBLen - 1] sample */
emh203 0:3d9c67d97d6f 291 c0 = *(py);
emh203 0:3d9c67d97d6f 292
emh203 0:3d9c67d97d6f 293 /* Read x[2] sample */
emh203 0:3d9c67d97d6f 294 x2 = *(px);
emh203 0:3d9c67d97d6f 295
emh203 0:3d9c67d97d6f 296 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 297 /* acc0 += x[0] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 298 acc0 += (q63_t) x0 *c0;
emh203 0:3d9c67d97d6f 299 /* acc1 += x[1] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 300 acc1 += (q63_t) x1 *c0;
emh203 0:3d9c67d97d6f 301 /* acc2 += x[2] * y[srcBLen - 1] */
emh203 0:3d9c67d97d6f 302 acc2 += (q63_t) x2 *c0;
emh203 0:3d9c67d97d6f 303
emh203 0:3d9c67d97d6f 304 /* Read y[srcBLen - 2] sample */
emh203 0:3d9c67d97d6f 305 c0 = *(py - 1u);
emh203 0:3d9c67d97d6f 306
emh203 0:3d9c67d97d6f 307 /* Read x[3] sample */
emh203 0:3d9c67d97d6f 308 x0 = *(px + 1u);
emh203 0:3d9c67d97d6f 309
emh203 0:3d9c67d97d6f 310 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 311 /* acc0 += x[1] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 312 acc0 += (q63_t) x1 *c0;
emh203 0:3d9c67d97d6f 313 /* acc1 += x[2] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 314 acc1 += (q63_t) x2 *c0;
emh203 0:3d9c67d97d6f 315 /* acc2 += x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 316 acc2 += (q63_t) x0 *c0;
emh203 0:3d9c67d97d6f 317
emh203 0:3d9c67d97d6f 318 /* Read y[srcBLen - 3] sample */
emh203 0:3d9c67d97d6f 319 c0 = *(py - 2u);
emh203 0:3d9c67d97d6f 320
emh203 0:3d9c67d97d6f 321 /* Read x[4] sample */
emh203 0:3d9c67d97d6f 322 x1 = *(px + 2u);
emh203 0:3d9c67d97d6f 323
emh203 0:3d9c67d97d6f 324 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 325 /* acc0 += x[2] * y[srcBLen - 3] */
emh203 0:3d9c67d97d6f 326 acc0 += (q63_t) x2 *c0;
emh203 0:3d9c67d97d6f 327 /* acc1 += x[3] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 328 acc1 += (q63_t) x0 *c0;
emh203 0:3d9c67d97d6f 329 /* acc2 += x[4] * y[srcBLen - 2] */
emh203 0:3d9c67d97d6f 330 acc2 += (q63_t) x1 *c0;
emh203 0:3d9c67d97d6f 331
emh203 0:3d9c67d97d6f 332
emh203 0:3d9c67d97d6f 333 px += 3u;
emh203 0:3d9c67d97d6f 334
emh203 0:3d9c67d97d6f 335 py -= 3u;
emh203 0:3d9c67d97d6f 336
emh203 0:3d9c67d97d6f 337 } while(--k);
emh203 0:3d9c67d97d6f 338
emh203 0:3d9c67d97d6f 339 /* If the srcBLen is not a multiple of 3, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 340 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 341 k = srcBLen - (3 * (srcBLen / 3));
emh203 0:3d9c67d97d6f 342
emh203 0:3d9c67d97d6f 343 while(k > 0u)
emh203 0:3d9c67d97d6f 344 {
emh203 0:3d9c67d97d6f 345 /* Read y[srcBLen - 5] sample */
emh203 0:3d9c67d97d6f 346 c0 = *(py--);
emh203 0:3d9c67d97d6f 347
emh203 0:3d9c67d97d6f 348 /* Read x[7] sample */
emh203 0:3d9c67d97d6f 349 x2 = *(px++);
emh203 0:3d9c67d97d6f 350
emh203 0:3d9c67d97d6f 351 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 352 /* acc0 += x[4] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 353 acc0 += (q63_t) x0 *c0;
emh203 0:3d9c67d97d6f 354 /* acc1 += x[5] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 355 acc1 += (q63_t) x1 *c0;
emh203 0:3d9c67d97d6f 356 /* acc2 += x[6] * y[srcBLen - 5] */
emh203 0:3d9c67d97d6f 357 acc2 += (q63_t) x2 *c0;
emh203 0:3d9c67d97d6f 358
emh203 0:3d9c67d97d6f 359 /* Reuse the present samples for the next MAC */
emh203 0:3d9c67d97d6f 360 x0 = x1;
emh203 0:3d9c67d97d6f 361 x1 = x2;
emh203 0:3d9c67d97d6f 362
emh203 0:3d9c67d97d6f 363 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 364 k--;
emh203 0:3d9c67d97d6f 365 }
emh203 0:3d9c67d97d6f 366
emh203 0:3d9c67d97d6f 367 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 368 *pOut++ = (q31_t) (acc0 >> 31);
emh203 0:3d9c67d97d6f 369 *pOut++ = (q31_t) (acc1 >> 31);
emh203 0:3d9c67d97d6f 370 *pOut++ = (q31_t) (acc2 >> 31);
emh203 0:3d9c67d97d6f 371
emh203 0:3d9c67d97d6f 372 /* Increment the pointer pIn1 index, count by 3 */
emh203 0:3d9c67d97d6f 373 count += 3u;
emh203 0:3d9c67d97d6f 374
emh203 0:3d9c67d97d6f 375 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 376 px = pIn1 + count;
emh203 0:3d9c67d97d6f 377 py = pSrc2;
emh203 0:3d9c67d97d6f 378
emh203 0:3d9c67d97d6f 379 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 380 blkCnt--;
emh203 0:3d9c67d97d6f 381 }
emh203 0:3d9c67d97d6f 382
emh203 0:3d9c67d97d6f 383 /* If the blockSize2 is not a multiple of 3, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 384 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 385 blkCnt = blockSize2 - 3 * (blockSize2 / 3);
emh203 0:3d9c67d97d6f 386
emh203 0:3d9c67d97d6f 387 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 388 {
emh203 0:3d9c67d97d6f 389 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 390 sum = 0;
emh203 0:3d9c67d97d6f 391
emh203 0:3d9c67d97d6f 392 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 393 k = srcBLen >> 2u;
emh203 0:3d9c67d97d6f 394
emh203 0:3d9c67d97d6f 395 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 396 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 397 while(k > 0u)
emh203 0:3d9c67d97d6f 398 {
emh203 0:3d9c67d97d6f 399 /* Perform the multiply-accumulates */
emh203 0:3d9c67d97d6f 400 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 401 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 402 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 403 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 404
emh203 0:3d9c67d97d6f 405 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 406 k--;
emh203 0:3d9c67d97d6f 407 }
emh203 0:3d9c67d97d6f 408
emh203 0:3d9c67d97d6f 409 /* If the srcBLen is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 410 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 411 k = srcBLen % 0x4u;
emh203 0:3d9c67d97d6f 412
emh203 0:3d9c67d97d6f 413 while(k > 0u)
emh203 0:3d9c67d97d6f 414 {
emh203 0:3d9c67d97d6f 415 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 416 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 417
emh203 0:3d9c67d97d6f 418 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 419 k--;
emh203 0:3d9c67d97d6f 420 }
emh203 0:3d9c67d97d6f 421
emh203 0:3d9c67d97d6f 422 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 423 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 424
emh203 0:3d9c67d97d6f 425 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 426 count++;
emh203 0:3d9c67d97d6f 427
emh203 0:3d9c67d97d6f 428 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 429 px = pIn1 + count;
emh203 0:3d9c67d97d6f 430 py = pSrc2;
emh203 0:3d9c67d97d6f 431
emh203 0:3d9c67d97d6f 432 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 433 blkCnt--;
emh203 0:3d9c67d97d6f 434 }
emh203 0:3d9c67d97d6f 435 }
emh203 0:3d9c67d97d6f 436 else
emh203 0:3d9c67d97d6f 437 {
emh203 0:3d9c67d97d6f 438 /* If the srcBLen is not a multiple of 4,
emh203 0:3d9c67d97d6f 439 * the blockSize2 loop cannot be unrolled by 4 */
emh203 0:3d9c67d97d6f 440 blkCnt = (uint32_t) blockSize2;
emh203 0:3d9c67d97d6f 441
emh203 0:3d9c67d97d6f 442 while(blkCnt > 0u)
emh203 0:3d9c67d97d6f 443 {
emh203 0:3d9c67d97d6f 444 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 445 sum = 0;
emh203 0:3d9c67d97d6f 446
emh203 0:3d9c67d97d6f 447 /* srcBLen number of MACS should be performed */
emh203 0:3d9c67d97d6f 448 k = srcBLen;
emh203 0:3d9c67d97d6f 449
emh203 0:3d9c67d97d6f 450 while(k > 0u)
emh203 0:3d9c67d97d6f 451 {
emh203 0:3d9c67d97d6f 452 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 453 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 454
emh203 0:3d9c67d97d6f 455 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 456 k--;
emh203 0:3d9c67d97d6f 457 }
emh203 0:3d9c67d97d6f 458
emh203 0:3d9c67d97d6f 459 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 460 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 461
emh203 0:3d9c67d97d6f 462 /* Increment the MAC count */
emh203 0:3d9c67d97d6f 463 count++;
emh203 0:3d9c67d97d6f 464
emh203 0:3d9c67d97d6f 465 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 466 px = pIn1 + count;
emh203 0:3d9c67d97d6f 467 py = pSrc2;
emh203 0:3d9c67d97d6f 468
emh203 0:3d9c67d97d6f 469 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 470 blkCnt--;
emh203 0:3d9c67d97d6f 471 }
emh203 0:3d9c67d97d6f 472 }
emh203 0:3d9c67d97d6f 473
emh203 0:3d9c67d97d6f 474
emh203 0:3d9c67d97d6f 475 /* --------------------------
emh203 0:3d9c67d97d6f 476 * Initializations of stage3
emh203 0:3d9c67d97d6f 477 * -------------------------*/
emh203 0:3d9c67d97d6f 478
emh203 0:3d9c67d97d6f 479 /* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1]
emh203 0:3d9c67d97d6f 480 * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2]
emh203 0:3d9c67d97d6f 481 * ....
emh203 0:3d9c67d97d6f 482 * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2]
emh203 0:3d9c67d97d6f 483 * sum += x[srcALen-1] * y[srcBLen-1]
emh203 0:3d9c67d97d6f 484 */
emh203 0:3d9c67d97d6f 485
emh203 0:3d9c67d97d6f 486 /* In this stage the MAC operations are decreased by 1 for every iteration.
emh203 0:3d9c67d97d6f 487 The blockSize3 variable holds the number of MAC operations performed */
emh203 0:3d9c67d97d6f 488 count = srcBLen - 1u;
emh203 0:3d9c67d97d6f 489
emh203 0:3d9c67d97d6f 490 /* Working pointer of inputA */
emh203 0:3d9c67d97d6f 491 pSrc1 = (pIn1 + srcALen) - (srcBLen - 1u);
emh203 0:3d9c67d97d6f 492 px = pSrc1;
emh203 0:3d9c67d97d6f 493
emh203 0:3d9c67d97d6f 494 /* Working pointer of inputB */
emh203 0:3d9c67d97d6f 495 pSrc2 = pIn2 + (srcBLen - 1u);
emh203 0:3d9c67d97d6f 496 py = pSrc2;
emh203 0:3d9c67d97d6f 497
emh203 0:3d9c67d97d6f 498 /* -------------------
emh203 0:3d9c67d97d6f 499 * Stage3 process
emh203 0:3d9c67d97d6f 500 * ------------------*/
emh203 0:3d9c67d97d6f 501
emh203 0:3d9c67d97d6f 502 while(blockSize3 > 0)
emh203 0:3d9c67d97d6f 503 {
emh203 0:3d9c67d97d6f 504 /* Accumulator is made zero for every iteration */
emh203 0:3d9c67d97d6f 505 sum = 0;
emh203 0:3d9c67d97d6f 506
emh203 0:3d9c67d97d6f 507 /* Apply loop unrolling and compute 4 MACs simultaneously. */
emh203 0:3d9c67d97d6f 508 k = count >> 2u;
emh203 0:3d9c67d97d6f 509
emh203 0:3d9c67d97d6f 510 /* First part of the processing with loop unrolling. Compute 4 MACs at a time.
emh203 0:3d9c67d97d6f 511 ** a second loop below computes MACs for the remaining 1 to 3 samples. */
emh203 0:3d9c67d97d6f 512 while(k > 0u)
emh203 0:3d9c67d97d6f 513 {
emh203 0:3d9c67d97d6f 514 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 515 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 516 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 517 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 518
emh203 0:3d9c67d97d6f 519 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 520 k--;
emh203 0:3d9c67d97d6f 521 }
emh203 0:3d9c67d97d6f 522
emh203 0:3d9c67d97d6f 523 /* If the blockSize3 is not a multiple of 4, compute any remaining MACs here.
emh203 0:3d9c67d97d6f 524 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 525 k = count % 0x4u;
emh203 0:3d9c67d97d6f 526
emh203 0:3d9c67d97d6f 527 while(k > 0u)
emh203 0:3d9c67d97d6f 528 {
emh203 0:3d9c67d97d6f 529 /* Perform the multiply-accumulate */
emh203 0:3d9c67d97d6f 530 sum += (q63_t) * px++ * (*py--);
emh203 0:3d9c67d97d6f 531
emh203 0:3d9c67d97d6f 532 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 533 k--;
emh203 0:3d9c67d97d6f 534 }
emh203 0:3d9c67d97d6f 535
emh203 0:3d9c67d97d6f 536 /* Store the result in the accumulator in the destination buffer. */
emh203 0:3d9c67d97d6f 537 *pOut++ = (q31_t) (sum >> 31);
emh203 0:3d9c67d97d6f 538
emh203 0:3d9c67d97d6f 539 /* Update the inputA and inputB pointers for next MAC calculation */
emh203 0:3d9c67d97d6f 540 px = ++pSrc1;
emh203 0:3d9c67d97d6f 541 py = pSrc2;
emh203 0:3d9c67d97d6f 542
emh203 0:3d9c67d97d6f 543 /* Decrement the MAC count */
emh203 0:3d9c67d97d6f 544 count--;
emh203 0:3d9c67d97d6f 545
emh203 0:3d9c67d97d6f 546 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 547 blockSize3--;
emh203 0:3d9c67d97d6f 548
emh203 0:3d9c67d97d6f 549 }
emh203 0:3d9c67d97d6f 550
emh203 0:3d9c67d97d6f 551 /* set status as ARM_MATH_SUCCESS */
emh203 0:3d9c67d97d6f 552 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 553 }
emh203 0:3d9c67d97d6f 554
emh203 0:3d9c67d97d6f 555 /* Return to application */
emh203 0:3d9c67d97d6f 556 return (status);
emh203 0:3d9c67d97d6f 557
emh203 0:3d9c67d97d6f 558 #else
emh203 0:3d9c67d97d6f 559
emh203 0:3d9c67d97d6f 560 /* Run the below code for Cortex-M0 */
emh203 0:3d9c67d97d6f 561
emh203 0:3d9c67d97d6f 562 q31_t *pIn1 = pSrcA; /* inputA pointer */
emh203 0:3d9c67d97d6f 563 q31_t *pIn2 = pSrcB; /* inputB pointer */
emh203 0:3d9c67d97d6f 564 q63_t sum; /* Accumulator */
emh203 0:3d9c67d97d6f 565 uint32_t i, j; /* loop counters */
emh203 0:3d9c67d97d6f 566 arm_status status; /* status of Partial convolution */
emh203 0:3d9c67d97d6f 567
emh203 0:3d9c67d97d6f 568 /* Check for range of output samples to be calculated */
emh203 0:3d9c67d97d6f 569 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emh203 0:3d9c67d97d6f 570 {
emh203 0:3d9c67d97d6f 571 /* Set status as ARM_ARGUMENT_ERROR */
emh203 0:3d9c67d97d6f 572 status = ARM_MATH_ARGUMENT_ERROR;
emh203 0:3d9c67d97d6f 573 }
emh203 0:3d9c67d97d6f 574 else
emh203 0:3d9c67d97d6f 575 {
emh203 0:3d9c67d97d6f 576 /* Loop to calculate convolution for output length number of values */
emh203 0:3d9c67d97d6f 577 for (i = firstIndex; i <= (firstIndex + numPoints - 1); i++)
emh203 0:3d9c67d97d6f 578 {
emh203 0:3d9c67d97d6f 579 /* Initialize sum with zero to carry on MAC operations */
emh203 0:3d9c67d97d6f 580 sum = 0;
emh203 0:3d9c67d97d6f 581
emh203 0:3d9c67d97d6f 582 /* Loop to perform MAC operations according to convolution equation */
emh203 0:3d9c67d97d6f 583 for (j = 0; j <= i; j++)
emh203 0:3d9c67d97d6f 584 {
emh203 0:3d9c67d97d6f 585 /* Check the array limitations */
emh203 0:3d9c67d97d6f 586 if(((i - j) < srcBLen) && (j < srcALen))
emh203 0:3d9c67d97d6f 587 {
emh203 0:3d9c67d97d6f 588 /* z[i] += x[i-j] * y[j] */
emh203 0:3d9c67d97d6f 589 sum += ((q63_t) pIn1[j] * (pIn2[i - j]));
emh203 0:3d9c67d97d6f 590 }
emh203 0:3d9c67d97d6f 591 }
emh203 0:3d9c67d97d6f 592
emh203 0:3d9c67d97d6f 593 /* Store the output in the destination buffer */
emh203 0:3d9c67d97d6f 594 pDst[i] = (q31_t) (sum >> 31u);
emh203 0:3d9c67d97d6f 595 }
emh203 0:3d9c67d97d6f 596 /* set status as ARM_SUCCESS as there are no argument errors */
emh203 0:3d9c67d97d6f 597 status = ARM_MATH_SUCCESS;
emh203 0:3d9c67d97d6f 598 }
emh203 0:3d9c67d97d6f 599 return (status);
emh203 0:3d9c67d97d6f 600
emh203 0:3d9c67d97d6f 601 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
emh203 0:3d9c67d97d6f 602
emh203 0:3d9c67d97d6f 603 }
emh203 0:3d9c67d97d6f 604
emh203 0:3d9c67d97d6f 605 /**
emh203 0:3d9c67d97d6f 606 * @} end of PartialConv group
emh203 0:3d9c67d97d6f 607 */