You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

139 lines
5.0KB

  1. /*
  2. * MMX optimized FLAC DSP utils
  3. * Copyright (c) 2007 Loren Merritt
  4. *
  5. * This file is part of FFmpeg.
  6. *
  7. * FFmpeg is free software; you can redistribute it and/or
  8. * modify it under the terms of the GNU Lesser General Public
  9. * License as published by the Free Software Foundation; either
  10. * version 2.1 of the License, or (at your option) any later version.
  11. *
  12. * FFmpeg is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  15. * Lesser General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public
  18. * License along with FFmpeg; if not, write to the Free Software
  19. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20. */
  21. #include "dsputil_mmx.h"
  22. static void apply_welch_window_sse2(const int32_t *data, int len, double *w_data)
  23. {
  24. double c = 2.0 / (len-1.0);
  25. int n2 = len>>1;
  26. long i = -n2*sizeof(int32_t);
  27. long j = n2*sizeof(int32_t);
  28. asm volatile(
  29. "movsd %0, %%xmm7 \n\t"
  30. "movapd %1, %%xmm6 \n\t"
  31. "movapd %2, %%xmm5 \n\t"
  32. "movlhps %%xmm7, %%xmm7 \n\t"
  33. "subpd %%xmm5, %%xmm7 \n\t"
  34. "addsd %%xmm6, %%xmm7 \n\t"
  35. ::"m"(c), "m"(*ff_pd_1), "m"(*ff_pd_2)
  36. );
  37. #define WELCH(MOVPD, offset)\
  38. asm volatile(\
  39. "1: \n\t"\
  40. "movapd %%xmm7, %%xmm1 \n\t"\
  41. "mulpd %%xmm1, %%xmm1 \n\t"\
  42. "movapd %%xmm6, %%xmm0 \n\t"\
  43. "subpd %%xmm1, %%xmm0 \n\t"\
  44. "pshufd $0x4e, %%xmm0, %%xmm1 \n\t"\
  45. "cvtpi2pd (%3,%0), %%xmm2 \n\t"\
  46. "cvtpi2pd "#offset"*4(%3,%1), %%xmm3 \n\t"\
  47. "mulpd %%xmm0, %%xmm2 \n\t"\
  48. "mulpd %%xmm1, %%xmm3 \n\t"\
  49. "movapd %%xmm2, (%2,%0,2) \n\t"\
  50. MOVPD" %%xmm3, "#offset"*8(%2,%1,2) \n\t"\
  51. "subpd %%xmm5, %%xmm7 \n\t"\
  52. "sub $8, %1 \n\t"\
  53. "add $8, %0 \n\t"\
  54. "jl 1b \n\t"\
  55. :"+&r"(i), "+&r"(j)\
  56. :"r"(w_data+n2), "r"(data+n2)\
  57. );
  58. if(len&1)
  59. WELCH("movupd", -1)
  60. else
  61. WELCH("movapd", -2)
  62. #undef WELCH
  63. }
  64. void ff_flac_compute_autocorr_sse2(const int32_t *data, int len, int lag,
  65. double *autoc)
  66. {
  67. double tmp[len + lag + 2];
  68. double *data1 = tmp + lag;
  69. int j;
  70. if((long)data1 & 15)
  71. data1++;
  72. apply_welch_window_sse2(data, len, data1);
  73. for(j=0; j<lag; j++)
  74. data1[j-lag]= 0.0;
  75. data1[len] = 0.0;
  76. for(j=0; j<lag; j+=2){
  77. long i = -len*sizeof(double);
  78. if(j == lag-2) {
  79. asm volatile(
  80. "movsd %6, %%xmm0 \n\t"
  81. "movsd %6, %%xmm1 \n\t"
  82. "movsd %6, %%xmm2 \n\t"
  83. "1: \n\t"
  84. "movapd (%4,%0), %%xmm3 \n\t"
  85. "movupd -8(%5,%0), %%xmm4 \n\t"
  86. "movapd (%5,%0), %%xmm5 \n\t"
  87. "mulpd %%xmm3, %%xmm4 \n\t"
  88. "mulpd %%xmm3, %%xmm5 \n\t"
  89. "mulpd -16(%5,%0), %%xmm3 \n\t"
  90. "addpd %%xmm4, %%xmm1 \n\t"
  91. "addpd %%xmm5, %%xmm0 \n\t"
  92. "addpd %%xmm3, %%xmm2 \n\t"
  93. "add $16, %0 \n\t"
  94. "jl 1b \n\t"
  95. "movhlps %%xmm0, %%xmm3 \n\t"
  96. "movhlps %%xmm1, %%xmm4 \n\t"
  97. "movhlps %%xmm2, %%xmm5 \n\t"
  98. "addsd %%xmm3, %%xmm0 \n\t"
  99. "addsd %%xmm4, %%xmm1 \n\t"
  100. "addsd %%xmm5, %%xmm2 \n\t"
  101. "movsd %%xmm0, %1 \n\t"
  102. "movsd %%xmm1, %2 \n\t"
  103. "movsd %%xmm2, %3 \n\t"
  104. :"+&r"(i), "=m"(autoc[j]), "=m"(autoc[j+1]), "=m"(autoc[j+2])
  105. :"r"(data1+len), "r"(data1+len-j), "m"(*ff_pd_1)
  106. );
  107. } else {
  108. asm volatile(
  109. "movsd %5, %%xmm0 \n\t"
  110. "movsd %5, %%xmm1 \n\t"
  111. "1: \n\t"
  112. "movapd (%3,%0), %%xmm3 \n\t"
  113. "movupd -8(%4,%0), %%xmm4 \n\t"
  114. "mulpd %%xmm3, %%xmm4 \n\t"
  115. "mulpd (%4,%0), %%xmm3 \n\t"
  116. "addpd %%xmm4, %%xmm1 \n\t"
  117. "addpd %%xmm3, %%xmm0 \n\t"
  118. "add $16, %0 \n\t"
  119. "jl 1b \n\t"
  120. "movhlps %%xmm0, %%xmm3 \n\t"
  121. "movhlps %%xmm1, %%xmm4 \n\t"
  122. "addsd %%xmm3, %%xmm0 \n\t"
  123. "addsd %%xmm4, %%xmm1 \n\t"
  124. "movsd %%xmm0, %1 \n\t"
  125. "movsd %%xmm1, %2 \n\t"
  126. :"+&r"(i), "=m"(autoc[j]), "=m"(autoc[j+1])
  127. :"r"(data1+len), "r"(data1+len-j), "m"(*ff_pd_1)
  128. );
  129. }
  130. }
  131. }