ac3dsp_mmx.c 4.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293
  1. /*
  2. * x86-optimized AC-3 DSP utils
  3. * Copyright (c) 2011 Justin Ruggles
  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 "libavutil/x86_cpu.h"
  22. #include "dsputil_mmx.h"
  23. #include "libavcodec/ac3dsp.h"
  24. extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs);
  25. extern void ff_ac3_exponent_min_mmxext(uint8_t *exp, int num_reuse_blocks, int nb_coefs);
  26. extern void ff_ac3_exponent_min_sse2 (uint8_t *exp, int num_reuse_blocks, int nb_coefs);
  27. extern int ff_ac3_max_msb_abs_int16_mmx (const int16_t *src, int len);
  28. extern int ff_ac3_max_msb_abs_int16_mmx2 (const int16_t *src, int len);
  29. extern int ff_ac3_max_msb_abs_int16_sse2 (const int16_t *src, int len);
  30. extern int ff_ac3_max_msb_abs_int16_ssse3(const int16_t *src, int len);
  31. extern void ff_ac3_lshift_int16_mmx (int16_t *src, unsigned int len, unsigned int shift);
  32. extern void ff_ac3_lshift_int16_sse2(int16_t *src, unsigned int len, unsigned int shift);
  33. extern void ff_ac3_rshift_int32_mmx (int32_t *src, unsigned int len, unsigned int shift);
  34. extern void ff_ac3_rshift_int32_sse2(int32_t *src, unsigned int len, unsigned int shift);
  35. extern void ff_float_to_fixed24_3dnow(int32_t *dst, const float *src, unsigned int len);
  36. extern void ff_float_to_fixed24_sse (int32_t *dst, const float *src, unsigned int len);
  37. extern void ff_float_to_fixed24_sse2 (int32_t *dst, const float *src, unsigned int len);
  38. extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]);
  39. extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_coefs);
  40. extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs);
  41. extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs);
  42. av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
  43. {
  44. #if HAVE_YASM
  45. int mm_flags = av_get_cpu_flags();
  46. if (mm_flags & AV_CPU_FLAG_MMX) {
  47. c->ac3_exponent_min = ff_ac3_exponent_min_mmx;
  48. c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx;
  49. c->ac3_lshift_int16 = ff_ac3_lshift_int16_mmx;
  50. c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx;
  51. }
  52. if (mm_flags & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) {
  53. c->extract_exponents = ff_ac3_extract_exponents_3dnow;
  54. if (!bit_exact) {
  55. c->float_to_fixed24 = ff_float_to_fixed24_3dnow;
  56. }
  57. }
  58. if (mm_flags & AV_CPU_FLAG_MMXEXT && HAVE_MMXEXT) {
  59. c->ac3_exponent_min = ff_ac3_exponent_min_mmxext;
  60. c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx2;
  61. }
  62. if (mm_flags & AV_CPU_FLAG_SSE && HAVE_SSE) {
  63. c->float_to_fixed24 = ff_float_to_fixed24_sse;
  64. }
  65. if (mm_flags & AV_CPU_FLAG_SSE2 && HAVE_SSE) {
  66. c->ac3_exponent_min = ff_ac3_exponent_min_sse2;
  67. c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2;
  68. c->float_to_fixed24 = ff_float_to_fixed24_sse2;
  69. c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2;
  70. c->extract_exponents = ff_ac3_extract_exponents_sse2;
  71. if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
  72. c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2;
  73. c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2;
  74. }
  75. }
  76. if (mm_flags & AV_CPU_FLAG_SSSE3 && HAVE_SSSE3) {
  77. c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3;
  78. if (!(mm_flags & AV_CPU_FLAG_ATOM)) {
  79. c->extract_exponents = ff_ac3_extract_exponents_ssse3;
  80. }
  81. }
  82. #endif
  83. }