arm_mult_q15.c 4.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_mult_q15.c
  4. * Description: Q15 vector multiplication
  5. *
  6. * $Date: 18. March 2019
  7. * $Revision: V1.6.0
  8. *
  9. * Target Processor: Cortex-M cores
  10. * -------------------------------------------------------------------- */
  11. /*
  12. * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
  13. *
  14. * SPDX-License-Identifier: Apache-2.0
  15. *
  16. * Licensed under the Apache License, Version 2.0 (the License); you may
  17. * not use this file except in compliance with the License.
  18. * You may obtain a copy of the License at
  19. *
  20. * www.apache.org/licenses/LICENSE-2.0
  21. *
  22. * Unless required by applicable law or agreed to in writing, software
  23. * distributed under the License is distributed on an AS IS BASIS, WITHOUT
  24. * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  25. * See the License for the specific language governing permissions and
  26. * limitations under the License.
  27. */
  28. #include "arm_math.h"
  29. /**
  30. @ingroup groupMath
  31. */
  32. /**
  33. @addtogroup BasicMult
  34. @{
  35. */
  36. /**
  37. @brief Q15 vector multiplication
  38. @param[in] pSrcA points to first input vector
  39. @param[in] pSrcB points to second input vector
  40. @param[out] pDst points to output vector
  41. @param[in] blockSize number of samples in each vector
  42. @return none
  43. @par Scaling and Overflow Behavior
  44. The function uses saturating arithmetic.
  45. Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
  46. */
  47. void arm_mult_q15(
  48. const q15_t * pSrcA,
  49. const q15_t * pSrcB,
  50. q15_t * pDst,
  51. uint32_t blockSize)
  52. {
  53. uint32_t blkCnt; /* Loop counter */
  54. #if defined (ARM_MATH_LOOPUNROLL)
  55. #if defined (ARM_MATH_DSP)
  56. q31_t inA1, inA2, inB1, inB2; /* Temporary input variables */
  57. q15_t out1, out2, out3, out4; /* Temporary output variables */
  58. q31_t mul1, mul2, mul3, mul4; /* Temporary variables */
  59. #endif
  60. /* Loop unrolling: Compute 4 outputs at a time */
  61. blkCnt = blockSize >> 2U;
  62. while (blkCnt > 0U)
  63. {
  64. /* C = A * B */
  65. #if defined (ARM_MATH_DSP)
  66. /* read 2 samples at a time from sourceA */
  67. inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
  68. /* read 2 samples at a time from sourceB */
  69. inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
  70. /* read 2 samples at a time from sourceA */
  71. inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
  72. /* read 2 samples at a time from sourceB */
  73. inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
  74. /* multiply mul = sourceA * sourceB */
  75. mul1 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
  76. mul2 = (q31_t) ((q15_t) (inA1 ) * (q15_t) (inB1 ));
  77. mul3 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB2 >> 16));
  78. mul4 = (q31_t) ((q15_t) (inA2 ) * (q15_t) (inB2 ));
  79. /* saturate result to 16 bit */
  80. out1 = (q15_t) __SSAT(mul1 >> 15, 16);
  81. out2 = (q15_t) __SSAT(mul2 >> 15, 16);
  82. out3 = (q15_t) __SSAT(mul3 >> 15, 16);
  83. out4 = (q15_t) __SSAT(mul4 >> 15, 16);
  84. /* store result to destination */
  85. #ifndef ARM_MATH_BIG_ENDIAN
  86. write_q15x2_ia (&pDst, __PKHBT(out2, out1, 16));
  87. write_q15x2_ia (&pDst, __PKHBT(out4, out3, 16));
  88. #else
  89. write_q15x2_ia (&pDst, __PKHBT(out1, out2, 16));
  90. write_q15x2_ia (&pDst, __PKHBT(out3, out4, 16));
  91. #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
  92. #else
  93. *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
  94. *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
  95. *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
  96. *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
  97. #endif
  98. /* Decrement loop counter */
  99. blkCnt--;
  100. }
  101. /* Loop unrolling: Compute remaining outputs */
  102. blkCnt = blockSize % 0x4U;
  103. #else
  104. /* Initialize blkCnt with number of samples */
  105. blkCnt = blockSize;
  106. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  107. while (blkCnt > 0U)
  108. {
  109. /* C = A * B */
  110. /* Multiply inputs and store result in destination buffer. */
  111. *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
  112. /* Decrement loop counter */
  113. blkCnt--;
  114. }
  115. }
  116. /**
  117. @} end of BasicMult group
  118. */