arm_abs_q7.c 4.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_abs_q7.c
  4. * Description: Q7 vector absolute value
  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 BasicAbs
  34. @{
  35. */
  36. /**
  37. @brief Q7 vector absolute value.
  38. @param[in] pSrc points to the input vector
  39. @param[out] pDst points to the output vector
  40. @param[in] blockSize number of samples in each vector
  41. @return none
  42. @par Conditions for optimum performance
  43. Input and output buffers should be aligned by 32-bit
  44. @par Scaling and Overflow Behavior
  45. The function uses saturating arithmetic.
  46. The Q7 value -1 (0x80) will be saturated to the maximum allowable positive value 0x7F.
  47. */
  48. #if defined(ARM_MATH_MVEI)
  49. #include "arm_helium_utils.h"
  50. void arm_abs_q7(
  51. const q7_t * pSrc,
  52. q7_t * pDst,
  53. uint32_t blockSize)
  54. {
  55. uint32_t blkCnt; /* loop counters */
  56. q7x16_t vecSrc;
  57. /* Compute 16 outputs at a time */
  58. blkCnt = blockSize >> 4;
  59. while (blkCnt > 0U)
  60. {
  61. /*
  62. * C = |A|
  63. * Calculate absolute and then store the results in the destination buffer.
  64. */
  65. vecSrc = vld1q(pSrc);
  66. vst1q(pDst, vqabsq(vecSrc));
  67. /*
  68. * Decrement the blockSize loop counter
  69. */
  70. blkCnt--;
  71. /*
  72. * advance vector source and destination pointers
  73. */
  74. pSrc += 16;
  75. pDst += 16;
  76. }
  77. /*
  78. * tail
  79. */
  80. blkCnt = blockSize & 0xF;
  81. if (blkCnt > 0U)
  82. {
  83. mve_pred16_t p0 = vctp8q(blkCnt);
  84. vecSrc = vld1q(pSrc);
  85. vstrbq_p(pDst, vqabsq(vecSrc), p0);
  86. }
  87. }
  88. #else
  89. void arm_abs_q7(
  90. const q7_t * pSrc,
  91. q7_t * pDst,
  92. uint32_t blockSize)
  93. {
  94. uint32_t blkCnt; /* Loop counter */
  95. q7_t in; /* Temporary input variable */
  96. #if defined (ARM_MATH_LOOPUNROLL)
  97. /* Loop unrolling: Compute 4 outputs at a time */
  98. blkCnt = blockSize >> 2U;
  99. while (blkCnt > 0U)
  100. {
  101. /* C = |A| */
  102. /* Calculate absolute of input (if -1 then saturated to 0x7f) and store result in destination buffer. */
  103. in = *pSrc++;
  104. #if defined (ARM_MATH_DSP)
  105. *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
  106. #else
  107. *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
  108. #endif
  109. in = *pSrc++;
  110. #if defined (ARM_MATH_DSP)
  111. *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
  112. #else
  113. *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
  114. #endif
  115. in = *pSrc++;
  116. #if defined (ARM_MATH_DSP)
  117. *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
  118. #else
  119. *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
  120. #endif
  121. in = *pSrc++;
  122. #if defined (ARM_MATH_DSP)
  123. *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
  124. #else
  125. *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
  126. #endif
  127. /* Decrement loop counter */
  128. blkCnt--;
  129. }
  130. /* Loop unrolling: Compute remaining outputs */
  131. blkCnt = blockSize % 0x4U;
  132. #else
  133. /* Initialize blkCnt with number of samples */
  134. blkCnt = blockSize;
  135. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  136. while (blkCnt > 0U)
  137. {
  138. /* C = |A| */
  139. /* Calculate absolute of input (if -1 then saturated to 0x7f) and store result in destination buffer. */
  140. in = *pSrc++;
  141. #if defined (ARM_MATH_DSP)
  142. *pDst++ = (in > 0) ? in : (q7_t) __QSUB8(0, in);
  143. #else
  144. *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
  145. #endif
  146. /* Decrement loop counter */
  147. blkCnt--;
  148. }
  149. }
  150. #endif /* defined(ARM_MATH_MVEI) */
  151. /**
  152. @} end of BasicAbs group
  153. */