arm_abs_q31.c 5.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_abs_q31.c
  4. * Description: Q31 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 Q31 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 Scaling and Overflow Behavior
  43. The function uses saturating arithmetic.
  44. The Q31 value -1 (0x80000000) will be saturated to the maximum allowable positive value 0x7FFFFFFF.
  45. */
  46. #if defined(ARM_MATH_MVEI)
  47. #include "arm_helium_utils.h"
  48. void arm_abs_q31(
  49. const q31_t * pSrc,
  50. q31_t * pDst,
  51. uint32_t blockSize)
  52. {
  53. uint32_t blkCnt; /* Loop counters */
  54. q31x4_t vecSrc;
  55. /* Compute 4 outputs at a time */
  56. blkCnt = blockSize >> 2;
  57. while (blkCnt > 0U)
  58. {
  59. /*
  60. * C = |A|
  61. * Calculate absolute and then store the results in the destination buffer.
  62. */
  63. vecSrc = vld1q(pSrc);
  64. vst1q(pDst, vqabsq(vecSrc));
  65. /*
  66. * Decrement the blockSize loop counter
  67. */
  68. blkCnt--;
  69. /*
  70. * Advance vector source and destination pointers
  71. */
  72. pSrc += 4;
  73. pDst += 4;
  74. }
  75. /*
  76. * Tail
  77. */
  78. blkCnt = blockSize & 3;
  79. if (blkCnt > 0U)
  80. {
  81. mve_pred16_t p0 = vctp32q(blkCnt);
  82. vecSrc = vld1q(pSrc);
  83. vstrwq_p(pDst, vqabsq(vecSrc), p0);
  84. }
  85. }
  86. #else
  87. void arm_abs_q31(
  88. const q31_t * pSrc,
  89. q31_t * pDst,
  90. uint32_t blockSize)
  91. {
  92. uint32_t blkCnt; /* Loop counter */
  93. q31_t in; /* Temporary variable */
  94. #if defined(ARM_MATH_NEON)
  95. int32x4_t vec1;
  96. int32x4_t res;
  97. /* Compute 4 outputs at a time */
  98. blkCnt = blockSize >> 2U;
  99. while (blkCnt > 0U)
  100. {
  101. /* C = |A| */
  102. /* Calculate absolute and then store the results in the destination buffer. */
  103. vec1 = vld1q_s32(pSrc);
  104. res = vqabsq_s32(vec1);
  105. vst1q_s32(pDst, res);
  106. /* Increment pointers */
  107. pSrc += 4;
  108. pDst += 4;
  109. /* Decrement the blockSize loop counter */
  110. blkCnt--;
  111. }
  112. /* Tail */
  113. blkCnt = blockSize & 0x3;
  114. #else
  115. #if defined (ARM_MATH_LOOPUNROLL)
  116. /* Loop unrolling: Compute 4 outputs at a time */
  117. blkCnt = blockSize >> 2U;
  118. while (blkCnt > 0U)
  119. {
  120. /* C = |A| */
  121. /* Calculate absolute of input (if -1 then saturated to 0x7fffffff) and store result in destination buffer. */
  122. in = *pSrc++;
  123. #if defined (ARM_MATH_DSP)
  124. *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
  125. #else
  126. *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
  127. #endif
  128. in = *pSrc++;
  129. #if defined (ARM_MATH_DSP)
  130. *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
  131. #else
  132. *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
  133. #endif
  134. in = *pSrc++;
  135. #if defined (ARM_MATH_DSP)
  136. *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
  137. #else
  138. *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
  139. #endif
  140. in = *pSrc++;
  141. #if defined (ARM_MATH_DSP)
  142. *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
  143. #else
  144. *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
  145. #endif
  146. /* Decrement loop counter */
  147. blkCnt--;
  148. }
  149. /* Loop unrolling: Compute remaining outputs */
  150. blkCnt = blockSize % 0x4U;
  151. #else
  152. /* Initialize blkCnt with number of samples */
  153. blkCnt = blockSize;
  154. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  155. #endif /* #if defined (ARM_MATH_NEON) */
  156. while (blkCnt > 0U)
  157. {
  158. /* C = |A| */
  159. /* Calculate absolute of input (if -1 then saturated to 0x7fffffff) and store result in destination buffer. */
  160. in = *pSrc++;
  161. #if defined (ARM_MATH_DSP)
  162. *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
  163. #else
  164. *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
  165. #endif
  166. /* Decrement loop counter */
  167. blkCnt--;
  168. }
  169. }
  170. #endif /* #if defined (ARM_MATH_MVEI) */
  171. /**
  172. @} end of BasicAbs group
  173. */