arm_scale_q31.c 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244
  1. /* ----------------------------------------------------------------------
  2. * Project: CMSIS DSP Library
  3. * Title: arm_scale_q31.c
  4. * Description: Multiplies a Q31 vector by a scalar
  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 BasicScale
  34. @{
  35. */
  36. /**
  37. @brief Multiplies a Q31 vector by a scalar.
  38. @param[in] pSrc points to the input vector
  39. @param[in] scaleFract fractional portion of the scale value
  40. @param[in] shift number of bits to shift the result by
  41. @param[out] pDst points to the output vector
  42. @param[in] blockSize number of samples in each vector
  43. @return none
  44. @par Scaling and Overflow Behavior
  45. The input data <code>*pSrc</code> and <code>scaleFract</code> are in 1.31 format.
  46. These are multiplied to yield a 2.62 intermediate result and this is shifted with saturation to 1.31 format.
  47. */
  48. #if defined(ARM_MATH_MVEI)
  49. #include "arm_helium_utils.h"
  50. void arm_scale_q31(
  51. const q31_t * pSrc,
  52. q31_t scaleFract,
  53. int8_t shift,
  54. q31_t * pDst,
  55. uint32_t blockSize)
  56. {
  57. uint32_t blkCnt; /* loop counters */
  58. q31x4_t vecSrc;
  59. q31x4_t vecDst;
  60. /* Compute 4 outputs at a time */
  61. blkCnt = blockSize >> 2;
  62. while (blkCnt > 0U)
  63. {
  64. /*
  65. * C = A * scale
  66. * Scale the input and then store the result in the destination buffer.
  67. */
  68. vecSrc = vld1q(pSrc);
  69. vecDst = vmulhq(vecSrc, vdupq_n_s32(scaleFract));
  70. vecDst = vqshlq_r(vecDst, shift + 1);
  71. vst1q(pDst, vecDst);
  72. /*
  73. * Decrement the blockSize loop counter
  74. */
  75. blkCnt--;
  76. /*
  77. * advance vector source and destination pointers
  78. */
  79. pSrc += 4;
  80. pDst += 4;
  81. }
  82. /*
  83. * tail
  84. */
  85. blkCnt = blockSize & 3;
  86. if (blkCnt > 0U)
  87. {
  88. mve_pred16_t p0 = vctp32q(blkCnt);
  89. vecSrc = vld1q(pSrc);
  90. vecDst = vmulhq(vecSrc, vdupq_n_s32(scaleFract));
  91. vecDst = vqshlq_r(vecDst, shift + 1);
  92. vstrwq_p(pDst, vecDst, p0);
  93. }
  94. }
  95. #else
  96. void arm_scale_q31(
  97. const q31_t *pSrc,
  98. q31_t scaleFract,
  99. int8_t shift,
  100. q31_t *pDst,
  101. uint32_t blockSize)
  102. {
  103. uint32_t blkCnt; /* Loop counter */
  104. q31_t in, out; /* Temporary variables */
  105. int8_t kShift = shift + 1; /* Shift to apply after scaling */
  106. int8_t sign = (kShift & 0x80);
  107. #if defined (ARM_MATH_LOOPUNROLL)
  108. /* Loop unrolling: Compute 4 outputs at a time */
  109. blkCnt = blockSize >> 2U;
  110. if (sign == 0U)
  111. {
  112. while (blkCnt > 0U)
  113. {
  114. /* C = A * scale */
  115. /* Scale input and store result in destination buffer. */
  116. in = *pSrc++; /* read input from source */
  117. in = ((q63_t) in * scaleFract) >> 32; /* multiply input with scaler value */
  118. out = in << kShift; /* apply shifting */
  119. if (in != (out >> kShift)) /* saturate the result */
  120. out = 0x7FFFFFFF ^ (in >> 31);
  121. *pDst++ = out; /* Store result destination */
  122. in = *pSrc++;
  123. in = ((q63_t) in * scaleFract) >> 32;
  124. out = in << kShift;
  125. if (in != (out >> kShift))
  126. out = 0x7FFFFFFF ^ (in >> 31);
  127. *pDst++ = out;
  128. in = *pSrc++;
  129. in = ((q63_t) in * scaleFract) >> 32;
  130. out = in << kShift;
  131. if (in != (out >> kShift))
  132. out = 0x7FFFFFFF ^ (in >> 31);
  133. *pDst++ = out;
  134. in = *pSrc++;
  135. in = ((q63_t) in * scaleFract) >> 32;
  136. out = in << kShift;
  137. if (in != (out >> kShift))
  138. out = 0x7FFFFFFF ^ (in >> 31);
  139. *pDst++ = out;
  140. /* Decrement loop counter */
  141. blkCnt--;
  142. }
  143. }
  144. else
  145. {
  146. while (blkCnt > 0U)
  147. {
  148. /* C = A * scale */
  149. /* Scale input and store result in destination buffer. */
  150. in = *pSrc++; /* read four inputs from source */
  151. in = ((q63_t) in * scaleFract) >> 32; /* multiply input with scaler value */
  152. out = in >> -kShift; /* apply shifting */
  153. *pDst++ = out; /* Store result destination */
  154. in = *pSrc++;
  155. in = ((q63_t) in * scaleFract) >> 32;
  156. out = in >> -kShift;
  157. *pDst++ = out;
  158. in = *pSrc++;
  159. in = ((q63_t) in * scaleFract) >> 32;
  160. out = in >> -kShift;
  161. *pDst++ = out;
  162. in = *pSrc++;
  163. in = ((q63_t) in * scaleFract) >> 32;
  164. out = in >> -kShift;
  165. *pDst++ = out;
  166. /* Decrement loop counter */
  167. blkCnt--;
  168. }
  169. }
  170. /* Loop unrolling: Compute remaining outputs */
  171. blkCnt = blockSize % 0x4U;
  172. #else
  173. /* Initialize blkCnt with number of samples */
  174. blkCnt = blockSize;
  175. #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
  176. if (sign == 0U)
  177. {
  178. while (blkCnt > 0U)
  179. {
  180. /* C = A * scale */
  181. /* Scale input and store result in destination buffer. */
  182. in = *pSrc++;
  183. in = ((q63_t) in * scaleFract) >> 32;
  184. out = in << kShift;
  185. if (in != (out >> kShift))
  186. out = 0x7FFFFFFF ^ (in >> 31);
  187. *pDst++ = out;
  188. /* Decrement loop counter */
  189. blkCnt--;
  190. }
  191. }
  192. else
  193. {
  194. while (blkCnt > 0U)
  195. {
  196. /* C = A * scale */
  197. /* Scale input and store result in destination buffer. */
  198. in = *pSrc++;
  199. in = ((q63_t) in * scaleFract) >> 32;
  200. out = in >> -kShift;
  201. *pDst++ = out;
  202. /* Decrement loop counter */
  203. blkCnt--;
  204. }
  205. }
  206. }
  207. #endif /* defined(ARM_MATH_MVEI) */
  208. /**
  209. @} end of BasicScale group
  210. */