arm_add_q15.c 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_add_q15.c
  4. * Description: Q15 vector addition
  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 BasicAdd
  34. @{
  35. */
  36. /**
  37. @brief Q15 vector addition.
  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 Q15 range [0x8000 0x7FFF] are saturated.
  46. */
  47. #if defined(ARM_MATH_MVEI)
  48. #include "arm_helium_utils.h"
  49. void arm_add_q15(
  50. const q15_t * pSrcA,
  51. const q15_t * pSrcB,
  52. q15_t * pDst,
  53. uint32_t blockSize)
  54. {
  55. uint32_t blkCnt; /* loop counters */
  56. q15x8_t vecA;
  57. q15x8_t vecB;
  58. /* Compute 8 outputs at a time */
  59. blkCnt = blockSize >> 3;
  60. while (blkCnt > 0U)
  61. {
  62. /*
  63. * C = A + B
  64. * Add and then store the results in the destination buffer.
  65. */
  66. vecA = vld1q(pSrcA);
  67. vecB = vld1q(pSrcB);
  68. vst1q(pDst, vqaddq(vecA, vecB));
  69. /*
  70. * Decrement the blockSize loop counter
  71. */
  72. blkCnt--;
  73. /*
  74. * advance vector source and destination pointers
  75. */
  76. pSrcA += 8;
  77. pSrcB += 8;
  78. pDst += 8;
  79. }
  80. /*
  81. * tail
  82. */
  83. blkCnt = blockSize & 7;
  84. if (blkCnt > 0U)
  85. {
  86. mve_pred16_t p0 = vctp16q(blkCnt);
  87. vecA = vld1q(pSrcA);
  88. vecB = vld1q(pSrcB);
  89. vstrhq_p(pDst, vqaddq(vecA, vecB), p0);
  90. }
  91. }
  92. #else
  93. void arm_add_q15(
  94. const q15_t * pSrcA,
  95. const q15_t * pSrcB,
  96. q15_t * pDst,
  97. uint32_t blockSize)
  98. {
  99. uint32_t blkCnt; /* Loop counter */
  100. #if defined (ARM_MATH_LOOPUNROLL)
  101. #if defined (ARM_MATH_DSP)
  102. q31_t inA1, inA2;
  103. q31_t inB1, inB2;
  104. #endif
  105. /* Loop unrolling: Compute 4 outputs at a time */
  106. blkCnt = blockSize >> 2U;
  107. while (blkCnt > 0U)
  108. {
  109. /* C = A + B */
  110. #if defined (ARM_MATH_DSP)
  111. /* read 2 times 2 samples at a time from sourceA */
  112. inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
  113. inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
  114. /* read 2 times 2 samples at a time from sourceB */
  115. inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
  116. inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
  117. /* Add and store 2 times 2 samples at a time */
  118. write_q15x2_ia (&pDst, __QADD16(inA1, inB1));
  119. write_q15x2_ia (&pDst, __QADD16(inA2, inB2));
  120. #else
  121. *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
  122. *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
  123. *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
  124. *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
  125. #endif
  126. /* Decrement loop counter */
  127. blkCnt--;
  128. }
  129. /* Loop unrolling: Compute remaining outputs */
  130. blkCnt = blockSize % 0x4U;
  131. #else
  132. /* Initialize blkCnt with number of samples */
  133. blkCnt = blockSize;
  134. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  135. while (blkCnt > 0U)
  136. {
  137. /* C = A + B */
  138. /* Add and store result in destination buffer. */
  139. #if defined (ARM_MATH_DSP)
  140. *pDst++ = (q15_t) __QADD16(*pSrcA++, *pSrcB++);
  141. #else
  142. *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
  143. #endif
  144. /* Decrement loop counter */
  145. blkCnt--;
  146. }
  147. }
  148. #endif /* defined(ARM_MATH_MVEI) */
  149. /**
  150. @} end of BasicAdd group
  151. */