tmk_core/tool/mbed/mbed-sdk/libraries/dsp/cmsis_dsp/ComplexMathFunctions/arm_cmplx_mult_cmplx_q31.c

   1 /* ----------------------------------------------------------------------
   2 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.
   3 *
   4 * $Date:        17. January 2013
   5 * $Revision:    V1.4.1
   6 *
   7 * Project:          CMSIS DSP Library
   8 * Title:            arm_cmplx_mult_cmplx_q31.c
   9 *
  10 * Description:  Q31 complex-by-complex multiplication
  11 *
  12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
  13 *
  14 * Redistribution and use in source and binary forms, with or without
  15 * modification, are permitted provided that the following conditions
  16 * are met:
  17 *   - Redistributions of source code must retain the above copyright
  18 *     notice, this list of conditions and the following disclaimer.
  19 *   - Redistributions in binary form must reproduce the above copyright
  20 *     notice, this list of conditions and the following disclaimer in
  21 *     the documentation and/or other materials provided with the
  22 *     distribution.
  23 *   - Neither the name of ARM LIMITED nor the names of its contributors
  24 *     may be used to endorse or promote products derived from this
  25 *     software without specific prior written permission.
  26 *
  27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  30 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
  31 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  32 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
  33 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  34 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  35 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
  37 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  38 * POSSIBILITY OF SUCH DAMAGE.
  39 * -------------------------------------------------------------------- */
  40
  41 #include "arm_math.h"
  42
  43 /**
  44  * @ingroup groupCmplxMath
  45  */
  46
  47 /**
  48  * @addtogroup CmplxByCmplxMult
  49  * @{
  50  */
  51
  52
  53 /**
  54  * @brief  Q31 complex-by-complex multiplication
  55  * @param[in]  *pSrcA points to the first input vector
  56  * @param[in]  *pSrcB points to the second input vector
  57  * @param[out]  *pDst  points to the output vector
  58  * @param[in]  numSamples number of complex samples in each vector
  59  * @return none.
  60  *
  61  * <b>Scaling and Overflow Behavior:</b>
  62  * \par
  63  * The function implements 1.31 by 1.31 multiplications and finally output is converted into 3.29 format.
  64  * Input down scaling is not required.
  65  */
  66
  67 void arm_cmplx_mult_cmplx_q31(
  68   q31_t * pSrcA,
  69   q31_t * pSrcB,
  70   q31_t * pDst,
  71   uint32_t numSamples)
  72 {
  73   q31_t a, b, c, d;                              /* Temporary variables to store real and imaginary values */
  74   uint32_t blkCnt;                               /* loop counters */
  75   q31_t mul1, mul2, mul3, mul4;
  76   q31_t out1, out2;
  77
  78 #ifndef ARM_MATH_CM0_FAMILY
  79
  80   /* Run the below code for Cortex-M4 and Cortex-M3 */
  81
  82   /* loop Unrolling */
  83   blkCnt = numSamples >> 2u;
  84
  85   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  86    ** a second loop below computes the remaining 1 to 3 samples. */
  87   while(blkCnt > 0u)
  88   {
  89     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
  90     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
  91     a = *pSrcA++;
  92     b = *pSrcA++;
  93     c = *pSrcB++;
  94     d = *pSrcB++;
  95
  96     mul1 = (q31_t) (((q63_t) a * c) >> 32);
  97     mul2 = (q31_t) (((q63_t) b * d) >> 32);
  98     mul3 = (q31_t) (((q63_t) a * d) >> 32);
  99     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 100
 101     mul1 = (mul1 >> 1);
 102     mul2 = (mul2 >> 1);
 103     mul3 = (mul3 >> 1);
 104     mul4 = (mul4 >> 1);
 105
 106     out1 = mul1 - mul2;
 107     out2 = mul3 + mul4;
 108
 109     /* store the real result in 3.29 format in the destination buffer. */
 110     *pDst++ = out1;
 111     /* store the imag result in 3.29 format in the destination buffer. */
 112     *pDst++ = out2;
 113
 114     a = *pSrcA++;
 115     b = *pSrcA++;
 116     c = *pSrcB++;
 117     d = *pSrcB++;
 118
 119     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 120     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 121     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 122     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 123
 124     mul1 = (mul1 >> 1);
 125     mul2 = (mul2 >> 1);
 126     mul3 = (mul3 >> 1);
 127     mul4 = (mul4 >> 1);
 128
 129     out1 = mul1 - mul2;
 130     out2 = mul3 + mul4;
 131
 132     /* store the real result in 3.29 format in the destination buffer. */
 133     *pDst++ = out1;
 134     /* store the imag result in 3.29 format in the destination buffer. */
 135     *pDst++ = out2;
 136
 137     a = *pSrcA++;
 138     b = *pSrcA++;
 139     c = *pSrcB++;
 140     d = *pSrcB++;
 141
 142     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 143     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 144     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 145     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 146
 147     mul1 = (mul1 >> 1);
 148     mul2 = (mul2 >> 1);
 149     mul3 = (mul3 >> 1);
 150     mul4 = (mul4 >> 1);
 151
 152     out1 = mul1 - mul2;
 153     out2 = mul3 + mul4;
 154
 155     /* store the real result in 3.29 format in the destination buffer. */
 156     *pDst++ = out1;
 157     /* store the imag result in 3.29 format in the destination buffer. */
 158     *pDst++ = out2;
 159
 160     a = *pSrcA++;
 161     b = *pSrcA++;
 162     c = *pSrcB++;
 163     d = *pSrcB++;
 164
 165     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 166     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 167     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 168     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 169
 170     mul1 = (mul1 >> 1);
 171     mul2 = (mul2 >> 1);
 172     mul3 = (mul3 >> 1);
 173     mul4 = (mul4 >> 1);
 174
 175     out1 = mul1 - mul2;
 176     out2 = mul3 + mul4;
 177
 178     /* store the real result in 3.29 format in the destination buffer. */
 179     *pDst++ = out1;
 180     /* store the imag result in 3.29 format in the destination buffer. */
 181     *pDst++ = out2;
 182
 183     /* Decrement the blockSize loop counter */
 184     blkCnt--;
 185   }
 186
 187   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
 188    ** No loop unrolling is used. */
 189   blkCnt = numSamples % 0x4u;
 190
 191   while(blkCnt > 0u)
 192   {
 193     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
 194     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
 195     a = *pSrcA++;
 196     b = *pSrcA++;
 197     c = *pSrcB++;
 198     d = *pSrcB++;
 199
 200     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 201     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 202     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 203     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 204
 205     mul1 = (mul1 >> 1);
 206     mul2 = (mul2 >> 1);
 207     mul3 = (mul3 >> 1);
 208     mul4 = (mul4 >> 1);
 209
 210     out1 = mul1 - mul2;
 211     out2 = mul3 + mul4;
 212
 213     /* store the real result in 3.29 format in the destination buffer. */
 214     *pDst++ = out1;
 215     /* store the imag result in 3.29 format in the destination buffer. */
 216     *pDst++ = out2;
 217
 218     /* Decrement the blockSize loop counter */
 219     blkCnt--;
 220   }
 221
 222 #else
 223
 224   /* Run the below code for Cortex-M0 */
 225
 226   /* loop Unrolling */
 227   blkCnt = numSamples >> 1u;
 228
 229   /* First part of the processing with loop unrolling.  Compute 2 outputs at a time.
 230    ** a second loop below computes the remaining 1 sample. */
 231   while(blkCnt > 0u)
 232   {
 233     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
 234     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
 235     a = *pSrcA++;
 236     b = *pSrcA++;
 237     c = *pSrcB++;
 238     d = *pSrcB++;
 239
 240     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 241     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 242     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 243     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 244
 245     mul1 = (mul1 >> 1);
 246     mul2 = (mul2 >> 1);
 247     mul3 = (mul3 >> 1);
 248     mul4 = (mul4 >> 1);
 249
 250     out1 = mul1 - mul2;
 251     out2 = mul3 + mul4;
 252
 253     /* store the real result in 3.29 format in the destination buffer. */
 254     *pDst++ = out1;
 255     /* store the imag result in 3.29 format in the destination buffer. */
 256     *pDst++ = out2;
 257
 258     a = *pSrcA++;
 259     b = *pSrcA++;
 260     c = *pSrcB++;
 261     d = *pSrcB++;
 262
 263     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 264     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 265     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 266     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 267
 268     mul1 = (mul1 >> 1);
 269     mul2 = (mul2 >> 1);
 270     mul3 = (mul3 >> 1);
 271     mul4 = (mul4 >> 1);
 272
 273     out1 = mul1 - mul2;
 274     out2 = mul3 + mul4;
 275
 276     /* store the real result in 3.29 format in the destination buffer. */
 277     *pDst++ = out1;
 278     /* store the imag result in 3.29 format in the destination buffer. */
 279     *pDst++ = out2;
 280
 281     /* Decrement the blockSize loop counter */
 282     blkCnt--;
 283   }
 284
 285   /* If the blockSize is not a multiple of 2, compute any remaining output samples here.
 286    ** No loop unrolling is used. */
 287   blkCnt = numSamples % 0x2u;
 288
 289   while(blkCnt > 0u)
 290   {
 291     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
 292     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
 293     a = *pSrcA++;
 294     b = *pSrcA++;
 295     c = *pSrcB++;
 296     d = *pSrcB++;
 297
 298     mul1 = (q31_t) (((q63_t) a * c) >> 32);
 299     mul2 = (q31_t) (((q63_t) b * d) >> 32);
 300     mul3 = (q31_t) (((q63_t) a * d) >> 32);
 301     mul4 = (q31_t) (((q63_t) b * c) >> 32);
 302
 303     mul1 = (mul1 >> 1);
 304     mul2 = (mul2 >> 1);
 305     mul3 = (mul3 >> 1);
 306     mul4 = (mul4 >> 1);
 307
 308     out1 = mul1 - mul2;
 309     out2 = mul3 + mul4;
 310
 311     /* store the real result in 3.29 format in the destination buffer. */
 312     *pDst++ = out1;
 313     /* store the imag result in 3.29 format in the destination buffer. */
 314     *pDst++ = out2;
 315
 316     /* Decrement the blockSize loop counter */
 317     blkCnt--;
 318   }
 319
 320 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
 321
 322 }
 323
 324 /**
 325  * @} end of CmplxByCmplxMult group
 326  */