arm_mult_q31.c 4.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_mult_q31.c
  4. * Description: Q31 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 Q31 vector multiplication.
  38. @param[in] pSrcA points to the first input vector.
  39. @param[in] pSrcB points to the second input vector.
  40. @param[out] pDst points to the 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 Q31 range[0x80000000 0x7FFFFFFF] are saturated.
  46. */
  47. #if defined(ARM_MATH_MVEI)
  48. #include "arm_helium_utils.h"
  49. void arm_mult_q31(
  50. const q31_t * pSrcA,
  51. const q31_t * pSrcB,
  52. q31_t * pDst,
  53. uint32_t blockSize)
  54. {
  55. uint32_t blkCnt; /* loop counters */
  56. q31x4_t vecA, vecB;
  57. /* Compute 4 outputs at a time */
  58. blkCnt = blockSize >> 2;
  59. while (blkCnt > 0U)
  60. {
  61. /*
  62. * C = A * B
  63. * Multiply the inputs and then store the results in the destination buffer.
  64. */
  65. vecA = vld1q(pSrcA);
  66. vecB = vld1q(pSrcB);
  67. vst1q(pDst, vqdmulhq(vecA, vecB));
  68. /*
  69. * Decrement the blockSize loop counter
  70. */
  71. blkCnt--;
  72. /*
  73. * advance vector source and destination pointers
  74. */
  75. pSrcA += 4;
  76. pSrcB += 4;
  77. pDst += 4;
  78. }
  79. /*
  80. * tail
  81. */
  82. blkCnt = blockSize & 3;
  83. if (blkCnt > 0U)
  84. {
  85. mve_pred16_t p0 = vctp32q(blkCnt);
  86. vecA = vld1q(pSrcA);
  87. vecB = vld1q(pSrcB);
  88. vstrwq_p(pDst, vqdmulhq(vecA, vecB), p0);
  89. }
  90. }
  91. #else
  92. void arm_mult_q31(
  93. const q31_t * pSrcA,
  94. const q31_t * pSrcB,
  95. q31_t * pDst,
  96. uint32_t blockSize)
  97. {
  98. uint32_t blkCnt; /* Loop counter */
  99. q31_t out; /* Temporary output variable */
  100. #if defined (ARM_MATH_LOOPUNROLL)
  101. /* Loop unrolling: Compute 4 outputs at a time */
  102. blkCnt = blockSize >> 2U;
  103. while (blkCnt > 0U)
  104. {
  105. /* C = A * B */
  106. /* Multiply inputs and store result in destination buffer. */
  107. out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
  108. out = __SSAT(out, 31);
  109. *pDst++ = out << 1U;
  110. out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
  111. out = __SSAT(out, 31);
  112. *pDst++ = out << 1U;
  113. out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
  114. out = __SSAT(out, 31);
  115. *pDst++ = out << 1U;
  116. out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
  117. out = __SSAT(out, 31);
  118. *pDst++ = out << 1U;
  119. /* Decrement loop counter */
  120. blkCnt--;
  121. }
  122. /* Loop unrolling: Compute remaining outputs */
  123. blkCnt = blockSize % 0x4U;
  124. #else
  125. /* Initialize blkCnt with number of samples */
  126. blkCnt = blockSize;
  127. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  128. while (blkCnt > 0U)
  129. {
  130. /* C = A * B */
  131. /* Multiply inputs and store result in destination buffer. */
  132. out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
  133. out = __SSAT(out, 31);
  134. *pDst++ = out << 1U;
  135. /* Decrement loop counter */
  136. blkCnt--;
  137. }
  138. }
  139. #endif /* defined(ARM_MATH_MVEI) */
  140. /**
  141. @} end of BasicMult group
  142. */