arm_mean_f32.c 4.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_mean_f32.c
  4. * Description: Mean value of a floating-point vector
  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 groupStats
  31. */
  32. /**
  33. @defgroup mean Mean
  34. Calculates the mean of the input vector. Mean is defined as the average of the elements in the vector.
  35. The underlying algorithm is used:
  36. <pre>
  37. Result = (pSrc[0] + pSrc[1] + pSrc[2] + ... + pSrc[blockSize-1]) / blockSize;
  38. </pre>
  39. There are separate functions for floating-point, Q31, Q15, and Q7 data types.
  40. */
  41. /**
  42. @addtogroup mean
  43. @{
  44. */
  45. /**
  46. @brief Mean value of a floating-point vector.
  47. @param[in] pSrc points to the input vector.
  48. @param[in] blockSize number of samples in input vector.
  49. @param[out] pResult mean value returned here.
  50. @return none
  51. */
  52. #if defined(ARM_MATH_NEON_EXPERIMENTAL)
  53. void arm_mean_f32(
  54. const float32_t * pSrc,
  55. uint32_t blockSize,
  56. float32_t * pResult)
  57. {
  58. float32_t sum = 0.0f; /* Temporary result storage */
  59. float32x4_t sumV = vdupq_n_f32(0.0f); /* Temporary result storage */
  60. float32x2_t sumV2;
  61. uint32_t blkCnt; /* Loop counter */
  62. float32_t in1, in2, in3, in4;
  63. float32x4_t inV;
  64. blkCnt = blockSize >> 2U;
  65. /* Compute 4 outputs at a time.
  66. ** a second loop below computes the remaining 1 to 3 samples. */
  67. while (blkCnt > 0U)
  68. {
  69. /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
  70. inV = vld1q_f32(pSrc);
  71. sumV = vaddq_f32(sumV, inV);
  72. pSrc += 4;
  73. /* Decrement the loop counter */
  74. blkCnt--;
  75. }
  76. sumV2 = vpadd_f32(vget_low_f32(sumV),vget_high_f32(sumV));
  77. sum = sumV2[0] + sumV2[1];
  78. /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  79. ** No loop unrolling is used. */
  80. blkCnt = blockSize & 3;
  81. while (blkCnt > 0U)
  82. {
  83. /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
  84. sum += *pSrc++;
  85. /* Decrement the loop counter */
  86. blkCnt--;
  87. }
  88. /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
  89. /* Store the result to the destination */
  90. *pResult = sum / (float32_t) blockSize;
  91. }
  92. #else
  93. void arm_mean_f32(
  94. const float32_t * pSrc,
  95. uint32_t blockSize,
  96. float32_t * pResult)
  97. {
  98. uint32_t blkCnt; /* Loop counter */
  99. float32_t sum = 0.0f; /* Temporary result storage */
  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[0] + A[1] + A[2] + ... + A[blockSize-1]) */
  106. sum += *pSrc++;
  107. sum += *pSrc++;
  108. sum += *pSrc++;
  109. sum += *pSrc++;
  110. /* Decrement the loop counter */
  111. blkCnt--;
  112. }
  113. /* Loop unrolling: Compute remaining outputs */
  114. blkCnt = blockSize % 0x4U;
  115. #else
  116. /* Initialize blkCnt with number of samples */
  117. blkCnt = blockSize;
  118. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  119. while (blkCnt > 0U)
  120. {
  121. /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
  122. sum += *pSrc++;
  123. /* Decrement loop counter */
  124. blkCnt--;
  125. }
  126. /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
  127. /* Store result to destination */
  128. *pResult = (sum / blockSize);
  129. }
  130. #endif /* #if defined(ARM_MATH_NEON) */
  131. /**
  132. @} end of mean group
  133. */