History

Merge remote-tracking branch 'qatar/master'

* qatar/master: os_support: Choose between direct.h and io.h using a configure check os_support: Include io.h instead of direct.h on mingw32ce x86: ac3dsp: Only refer to the ac3_downmix_sse symbol if it has been declared swscale: Remove two bogus asserts ac3: move ac3_downmix() from dsputil to ac3dsp lavr/audio_mix_matrix: acknowledge the existence of LFE2. mlp_parser: avoid mapping multiple disctinct TrueHD channels to the same Libav channel. lavu/audioconvert: add a second low frequency channel. Conflicts: doc/APIchanges libavcodec/ac3dsp.c libavcodec/ac3dsp.h libavcodec/mlp_parser.c libavutil/audioconvert.c libavutil/version.h

Michael Niedermayer committed 14 years ago

Commit e3e09f22 parents c942e8b ca6b54417 files · +171 −158Browse files

Unified · Split

17 files
  1. configure +2 −0
  2. doc/APIchanges +3 −0
  3. libavcodec/ac3dec.c +6 −34
  4. libavcodec/ac3dec.h +0 −3
  5. libavcodec/ac3dsp.c +26 −0
  6. libavcodec/ac3dsp.h +3 −0
  7. libavcodec/dsputil.c +0 −4
  8. libavcodec/dsputil.h +0 −1
  9. libavcodec/mlp_parser.c +2 −2
  10. libavcodec/x86/ac3dsp_init.c +121 −0
  11. libavcodec/x86/dsputil_mmx.c +0 −111
  12. libavformat/os_support.h +4 −0
  13. libavresample/audio_mix_matrix.c +1 −0
  14. libavutil/audioconvert.c +1 −0
  15. libavutil/audioconvert.h +1 −0
  16. libavutil/version.h +1 −1
  17. libswscale/swscale.c +0 −2
configure+2−0
@@ -1276,6 +1276,7 @@ HAVE_LIST=" · 1275 unmodified lines
1276 dev_ic_bt8xx_h1276 dev_ic_bt8xx_h
1277 dev_video_bktr_ioctl_bt848_h1277 dev_video_bktr_ioctl_bt848_h
1278 dev_video_meteor_ioctl_meteor_h1278 dev_video_meteor_ioctl_meteor_h
1279 direct_h
1279 dlfcn_h1280 dlfcn_h
1280 dlopen1281 dlopen
1281 dos_paths1282 dos_paths
@@ -3504,6 +3505,7 @@ check_func_headers windows.h Sleep · 2222 unmodified lines
3504check_func_headers windows.h VirtualAlloc3505check_func_headers windows.h VirtualAlloc
3505check_func_headers glob.h glob3506check_func_headers glob.h glob
3506 3507
3508check_header direct.h
3507check_header dlfcn.h3509check_header dlfcn.h
3508check_header dxva.h3510check_header dxva.h
3509check_header dxva2api.h -D_WIN32_WINNT=0x06003511check_header dxva2api.h -D_WIN32_WINNT=0x0600
doc/APIchanges+3−0
@@ -83,6 +83,9 @@ API changes, most recent first: · 82 unmodified lines
832012-03-26 - a67d9cf - lavfi 2.66.100832012-03-26 - a67d9cf - lavfi 2.66.100
84 Add avfilter_fill_frame_from_{audio_,}buffer_ref() functions.84 Add avfilter_fill_frame_from_{audio_,}buffer_ref() functions.
85 85
862012-09-12 - xxxxxxx - lavu 51.41.0 - audioconvert.h
87 Added AV_CH_LOW_FREQUENCY_2 channel mask value.
88
862012-09-04 - xxxxxxx - lavu 51.40.0 - opt.h892012-09-04 - xxxxxxx - lavu 51.40.0 - opt.h
87 Reordered the fields in default_val in AVOption, changed which90 Reordered the fields in default_val in AVOption, changed which
88 default_val field is used for which AVOptionType.91 default_val field is used for which AVOptionType.
libavcodec/ac3dec.c+6−34
@@ -620,34 +620,6 @@ static inline void do_imdct(AC3DecodeContext *s, int channels) · 619 unmodified lines
620 }620 }
621}621}
622 622
623/**
624 * Downmix the output to mono or stereo.
625 */
626void ff_ac3_downmix_c(float (*samples)[256], float (*matrix)[2],
627 int out_ch, int in_ch, int len)
628{
629 int i, j;
630 float v0, v1;
631 if (out_ch == 2) {
632 for (i = 0; i < len; i++) {
633 v0 = v1 = 0.0f;
634 for (j = 0; j < in_ch; j++) {
635 v0 += samples[j][i] * matrix[j][0];
636 v1 += samples[j][i] * matrix[j][1];
637 }
638 samples[0][i] = v0;
639 samples[1][i] = v1;
640 }
641 } else if (out_ch == 1) {
642 for (i = 0; i < len; i++) {
643 v0 = 0.0f;
644 for (j = 0; j < in_ch; j++)
645 v0 += samples[j][i] * matrix[j][0];
646 samples[0][i] = v0;
647 }
648 }
649}
650
651/**623/**
652 * Upmix delay samples from stereo to original channel layout.624 * Upmix delay samples from stereo to original channel layout.
653 */625 */
@@ -1266,19 +1238,19 @@ static int decode_audio_block(AC3DecodeContext *s, int blk) · 612 unmodified lines
1266 do_imdct(s, s->channels);1238 do_imdct(s, s->channels);
1267 1239
1268 if (downmix_output) {1240 if (downmix_output) {
1269 s->dsp.ac3_downmix(s->output, s->downmix_coeffs,1241 s->ac3dsp.downmix(s->output, s->downmix_coeffs,
1270 s->out_channels, s->fbw_channels, 256);1242 s->out_channels, s->fbw_channels, 256);
1271 }1243 }
1272 } else {1244 } else {
1273 if (downmix_output) {1245 if (downmix_output) {
1274 s->dsp.ac3_downmix(s->transform_coeffs + 1, s->downmix_coeffs,1246 s->ac3dsp.downmix(s->transform_coeffs + 1, s->downmix_coeffs,
1275 s->out_channels, s->fbw_channels, 256);1247 s->out_channels, s->fbw_channels, 256);
1276 }1248 }
1277 1249
1278 if (downmix_output && !s->downmixed) {1250 if (downmix_output && !s->downmixed) {
1279 s->downmixed = 1;1251 s->downmixed = 1;
1280 s->dsp.ac3_downmix(s->delay, s->downmix_coeffs, s->out_channels,1252 s->ac3dsp.downmix(s->delay, s->downmix_coeffs, s->out_channels,
1281 s->fbw_channels, 128);1253 s->fbw_channels, 128);
1282 }1254 }
1283 1255
1284 do_imdct(s, s->out_channels);1256 do_imdct(s, s->out_channels);
libavcodec/ac3dec.h+0−3
@@ -227,9 +227,6 @@ int ff_eac3_parse_header(AC3DecodeContext *s); · 226 unmodified lines
227 */227 */
228void ff_eac3_decode_transform_coeffs_aht_ch(AC3DecodeContext *s, int ch);228void ff_eac3_decode_transform_coeffs_aht_ch(AC3DecodeContext *s, int ch);
229 229
230void ff_ac3_downmix_c(float (*samples)[256], float (*matrix)[2],
231 int out_ch, int in_ch, int len);
232
233/**230/**
234 * Apply spectral extension to each channel by copying lower frequency231 * Apply spectral extension to each channel by copying lower frequency
235 * coefficients to higher frequency bins and applying side information to232 * coefficients to higher frequency bins and applying side information to
libavcodec/ac3dsp.c+26−0
@@ -214,6 +214,31 @@ static void ac3_sum_square_butterfly_float_c(float sum[4], · 213 unmodified lines
214 }214 }
215}215}
216 216
217static void ac3_downmix_c(float (*samples)[256], float (*matrix)[2],
218 int out_ch, int in_ch, int len)
219{
220 int i, j;
221 float v0, v1;
222 if (out_ch == 2) {
223 for (i = 0; i < len; i++) {
224 v0 = v1 = 0.0f;
225 for (j = 0; j < in_ch; j++) {
226 v0 += samples[j][i] * matrix[j][0];
227 v1 += samples[j][i] * matrix[j][1];
228 }
229 samples[0][i] = v0;
230 samples[1][i] = v1;
231 }
232 } else if (out_ch == 1) {
233 for (i = 0; i < len; i++) {
234 v0 = 0.0f;
235 for (j = 0; j < in_ch; j++)
236 v0 += samples[j][i] * matrix[j][0];
237 samples[0][i] = v0;
238 }
239 }
240}
241
217av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact)242av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact)
218{243{
219 c->ac3_exponent_min = ac3_exponent_min_c;244 c->ac3_exponent_min = ac3_exponent_min_c;
@@ -227,6 +252,7 @@ av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact) · 7 unmodified lines
227 c->extract_exponents = ac3_extract_exponents_c;252 c->extract_exponents = ac3_extract_exponents_c;
228 c->sum_square_butterfly_int32 = ac3_sum_square_butterfly_int32_c;253 c->sum_square_butterfly_int32 = ac3_sum_square_butterfly_int32_c;
229 c->sum_square_butterfly_float = ac3_sum_square_butterfly_float_c;254 c->sum_square_butterfly_float = ac3_sum_square_butterfly_float_c;
255 c->downmix = ac3_downmix_c;
230 256
231 if (ARCH_ARM)257 if (ARCH_ARM)
232 ff_ac3dsp_init_arm(c, bit_exact);258 ff_ac3dsp_init_arm(c, bit_exact);
libavcodec/ac3dsp.h+3−0
@@ -131,6 +131,9 @@ typedef struct AC3DSPContext { · 130 unmodified lines
131 131
132 void (*sum_square_butterfly_float)(float sum[4], const float *coef0,132 void (*sum_square_butterfly_float)(float sum[4], const float *coef0,
133 const float *coef1, int len);133 const float *coef1, int len);
134
135 void (*downmix)(float (*samples)[256], float (*matrix)[2], int out_ch,
136 int in_ch, int len);
134} AC3DSPContext;137} AC3DSPContext;
135 138
136void ff_ac3dsp_init (AC3DSPContext *c, int bit_exact);139void ff_ac3dsp_init (AC3DSPContext *c, int bit_exact);
libavcodec/dsputil.c+0−4
@@ -36,7 +36,6 @@ · 35 unmodified lines
36#include "mathops.h"36#include "mathops.h"
37#include "mpegvideo.h"37#include "mpegvideo.h"
38#include "config.h"38#include "config.h"
39#include "ac3dec.h"
40#include "vorbis.h"39#include "vorbis.h"
41#include "diracdsp.h"40#include "diracdsp.h"
42 41
@@ -3032,9 +3031,6 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx) · 2989 unmodified lines
3032 3031
3033#if CONFIG_VORBIS_DECODER3032#if CONFIG_VORBIS_DECODER
3034 c->vorbis_inverse_coupling = ff_vorbis_inverse_coupling;3033 c->vorbis_inverse_coupling = ff_vorbis_inverse_coupling;
3035#endif
3036#if CONFIG_AC3_DECODER
3037 c->ac3_downmix = ff_ac3_downmix_c;
3038#endif3034#endif
3039 c->vector_fmul_reverse = vector_fmul_reverse_c;3035 c->vector_fmul_reverse = vector_fmul_reverse_c;
3040 c->vector_fmul_add = vector_fmul_add_c;3036 c->vector_fmul_add = vector_fmul_add_c;
libavcodec/dsputil.h+0−1
@@ -384,7 +384,6 @@ typedef struct DSPContext { · 383 unmodified lines
384 384
385 /* assume len is a multiple of 4, and arrays are 16-byte aligned */385 /* assume len is a multiple of 4, and arrays are 16-byte aligned */
386 void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);386 void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
387 void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
388 /* assume len is a multiple of 16, and arrays are 32-byte aligned */387 /* assume len is a multiple of 16, and arrays are 32-byte aligned */
389 void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);388 void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
390 /* assume len is a multiple of 8, and src arrays are 16-byte aligned */389 /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
libavcodec/mlp_parser.c+2−2
@@ -83,10 +83,10 @@ static const uint64_t thd_layout[13] = { · 82 unmodified lines
83 AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT, // LRrs83 AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT, // LRrs
84 AV_CH_BACK_CENTER, // Cs84 AV_CH_BACK_CENTER, // Cs
85 AV_CH_TOP_CENTER, // Ts85 AV_CH_TOP_CENTER, // Ts
86 AV_CH_SIDE_LEFT|AV_CH_SIDE_RIGHT, // LRsd - TODO: Surround Direct86 AV_CH_SURROUND_DIRECT_LEFT|AV_CH_SURROUND_DIRECT_RIGHT, // LRsd
87 AV_CH_WIDE_LEFT|AV_CH_WIDE_RIGHT, // LRw87 AV_CH_WIDE_LEFT|AV_CH_WIDE_RIGHT, // LRw
88 AV_CH_TOP_FRONT_CENTER, // Cvh88 AV_CH_TOP_FRONT_CENTER, // Cvh
89 AV_CH_LOW_FREQUENCY // LFE289 AV_CH_LOW_FREQUENCY_2, // LFE2
90};90};
91 91
92static int mlp_samplerate(int in)92static int mlp_samplerate(int in)
libavcodec/x86/ac3dsp_init.c+121−0
@@ -19,9 +19,11 @@ · 18 unmodified lines
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */20 */
21 21
22#include "libavutil/mem.h"
22#include "libavutil/x86/asm.h"23#include "libavutil/x86/asm.h"
23#include "libavutil/x86/cpu.h"24#include "libavutil/x86/cpu.h"
24#include "dsputil_mmx.h"25#include "dsputil_mmx.h"
26#include "libavcodec/ac3.h"
25#include "libavcodec/ac3dsp.h"27#include "libavcodec/ac3dsp.h"
26 28
27extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs);29extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs);
@@ -49,6 +51,119 @@ extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_c · 21 unmodified lines
49extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs);51extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs);
50extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs);52extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs);
51 53
54#if HAVE_SSE_INLINE
55
56#define IF1(x) x
57#define IF0(x)
58
59#define MIX5(mono, stereo) \
60 __asm__ volatile ( \
61 "movss 0(%2), %%xmm5 \n" \
62 "movss 8(%2), %%xmm6 \n" \
63 "movss 24(%2), %%xmm7 \n" \
64 "shufps $0, %%xmm5, %%xmm5 \n" \
65 "shufps $0, %%xmm6, %%xmm6 \n" \
66 "shufps $0, %%xmm7, %%xmm7 \n" \
67 "1: \n" \
68 "movaps (%0, %1), %%xmm0 \n" \
69 "movaps 0x400(%0, %1), %%xmm1 \n" \
70 "movaps 0x800(%0, %1), %%xmm2 \n" \
71 "movaps 0xc00(%0, %1), %%xmm3 \n" \
72 "movaps 0x1000(%0, %1), %%xmm4 \n" \
73 "mulps %%xmm5, %%xmm0 \n" \
74 "mulps %%xmm6, %%xmm1 \n" \
75 "mulps %%xmm5, %%xmm2 \n" \
76 "mulps %%xmm7, %%xmm3 \n" \
77 "mulps %%xmm7, %%xmm4 \n" \
78 stereo("addps %%xmm1, %%xmm0 \n") \
79 "addps %%xmm1, %%xmm2 \n" \
80 "addps %%xmm3, %%xmm0 \n" \
81 "addps %%xmm4, %%xmm2 \n" \
82 mono("addps %%xmm2, %%xmm0 \n") \
83 "movaps %%xmm0, (%0, %1) \n" \
84 stereo("movaps %%xmm2, 0x400(%0, %1) \n") \
85 "add $16, %0 \n" \
86 "jl 1b \n" \
87 : "+&r"(i) \
88 : "r"(samples[0] + len), "r"(matrix) \
89 : XMM_CLOBBERS("%xmm0", "%xmm1", "%xmm2", "%xmm3", \
90 "%xmm4", "%xmm5", "%xmm6", "%xmm7",) \
91 "memory" \
92 );
93
94#define MIX_MISC(stereo) \
95 __asm__ volatile ( \
96 "1: \n" \
97 "movaps (%3, %0), %%xmm0 \n" \
98 stereo("movaps %%xmm0, %%xmm1 \n") \
99 "mulps %%xmm4, %%xmm0 \n" \
100 stereo("mulps %%xmm5, %%xmm1 \n") \
101 "lea 1024(%3, %0), %1 \n" \
102 "mov %5, %2 \n" \
103 "2: \n" \
104 "movaps (%1), %%xmm2 \n" \
105 stereo("movaps %%xmm2, %%xmm3 \n") \
106 "mulps (%4, %2), %%xmm2 \n" \
107 stereo("mulps 16(%4, %2), %%xmm3 \n") \
108 "addps %%xmm2, %%xmm0 \n" \
109 stereo("addps %%xmm3, %%xmm1 \n") \
110 "add $1024, %1 \n" \
111 "add $32, %2 \n" \
112 "jl 2b \n" \
113 "movaps %%xmm0, (%3, %0) \n" \
114 stereo("movaps %%xmm1, 1024(%3, %0) \n") \
115 "add $16, %0 \n" \
116 "jl 1b \n" \
117 : "+&r"(i), "=&r"(j), "=&r"(k) \
118 : "r"(samples[0] + len), "r"(matrix_simd + in_ch), \
119 "g"((intptr_t) - 32 * (in_ch - 1)) \
120 : "memory" \
121 );
122
123static void ac3_downmix_sse(float (*samples)[256], float (*matrix)[2],
124 int out_ch, int in_ch, int len)
125{
126 int (*matrix_cmp)[2] = (int(*)[2])matrix;
127 intptr_t i, j, k;
128
129 i = -len * sizeof(float);
130 if (in_ch == 5 && out_ch == 2 &&
131 !(matrix_cmp[0][1] | matrix_cmp[2][0] |
132 matrix_cmp[3][1] | matrix_cmp[4][0] |
133 (matrix_cmp[1][0] ^ matrix_cmp[1][1]) |
134 (matrix_cmp[0][0] ^ matrix_cmp[2][1]))) {
135 MIX5(IF0, IF1);
136 } else if (in_ch == 5 && out_ch == 1 &&
137 matrix_cmp[0][0] == matrix_cmp[2][0] &&
138 matrix_cmp[3][0] == matrix_cmp[4][0]) {
139 MIX5(IF1, IF0);
140 } else {
141 DECLARE_ALIGNED(16, float, matrix_simd)[AC3_MAX_CHANNELS][2][4];
142 j = 2 * in_ch * sizeof(float);
143 __asm__ volatile (
144 "1: \n"
145 "sub $8, %0 \n"
146 "movss (%2, %0), %%xmm4 \n"
147 "movss 4(%2, %0), %%xmm5 \n"
148 "shufps $0, %%xmm4, %%xmm4 \n"
149 "shufps $0, %%xmm5, %%xmm5 \n"
150 "movaps %%xmm4, (%1, %0, 4) \n"
151 "movaps %%xmm5, 16(%1, %0, 4) \n"
152 "jg 1b \n"
153 : "+&r"(j)
154 : "r"(matrix_simd), "r"(matrix)
155 : "memory"
156 );
157 if (out_ch == 2) {
158 MIX_MISC(IF1);
159 } else {
160 MIX_MISC(IF0);
161 }
162 }
163}
164
165#endif /* HAVE_SSE_INLINE */
166
52av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)167av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
53{168{
54 int mm_flags = av_get_cpu_flags();169 int mm_flags = av_get_cpu_flags();
@@ -89,4 +204,10 @@ av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact) · 34 unmodified lines
89 c->extract_exponents = ff_ac3_extract_exponents_ssse3;204 c->extract_exponents = ff_ac3_extract_exponents_ssse3;
90 }205 }
91 }206 }
207
208#if HAVE_SSE_INLINE
209 if (INLINE_SSE(mm_flags)) {
210 c->downmix = ac3_downmix_sse;
211 }
212#endif
92}213}
libavcodec/x86/dsputil_mmx.c+0−111
@@ -28,7 +28,6 @@ · 27 unmodified lines
28#include "libavcodec/h264dsp.h"28#include "libavcodec/h264dsp.h"
29#include "libavcodec/mpegvideo.h"29#include "libavcodec/mpegvideo.h"
30#include "libavcodec/simple_idct.h"30#include "libavcodec/simple_idct.h"
31#include "libavcodec/ac3dec.h"
32#include "dsputil_mmx.h"31#include "dsputil_mmx.h"
33#include "idct_xvid.h"32#include "idct_xvid.h"
34#include "diracdsp_mmx.h"33#include "diracdsp_mmx.h"
@@ -2375,115 +2374,6 @@ static void vorbis_inverse_coupling_sse(float *mag, float *ang, int blocksize) · 2340 unmodified lines
2375 }2374 }
2376}2375}
2377 2376
2378#define IF1(x) x
2379#define IF0(x)
2380
2381#define MIX5(mono, stereo) \
2382 __asm__ volatile ( \
2383 "movss 0(%2), %%xmm5 \n" \
2384 "movss 8(%2), %%xmm6 \n" \
2385 "movss 24(%2), %%xmm7 \n" \
2386 "shufps $0, %%xmm5, %%xmm5 \n" \
2387 "shufps $0, %%xmm6, %%xmm6 \n" \
2388 "shufps $0, %%xmm7, %%xmm7 \n" \
2389 "1: \n" \
2390 "movaps (%0, %1), %%xmm0 \n" \
2391 "movaps 0x400(%0, %1), %%xmm1 \n" \
2392 "movaps 0x800(%0, %1), %%xmm2 \n" \
2393 "movaps 0xc00(%0, %1), %%xmm3 \n" \
2394 "movaps 0x1000(%0, %1), %%xmm4 \n" \
2395 "mulps %%xmm5, %%xmm0 \n" \
2396 "mulps %%xmm6, %%xmm1 \n" \
2397 "mulps %%xmm5, %%xmm2 \n" \
2398 "mulps %%xmm7, %%xmm3 \n" \
2399 "mulps %%xmm7, %%xmm4 \n" \
2400 stereo("addps %%xmm1, %%xmm0 \n") \
2401 "addps %%xmm1, %%xmm2 \n" \
2402 "addps %%xmm3, %%xmm0 \n" \
2403 "addps %%xmm4, %%xmm2 \n" \
2404 mono("addps %%xmm2, %%xmm0 \n") \
2405 "movaps %%xmm0, (%0, %1) \n" \
2406 stereo("movaps %%xmm2, 0x400(%0, %1) \n") \
2407 "add $16, %0 \n" \
2408 "jl 1b \n" \
2409 : "+&r"(i) \
2410 : "r"(samples[0] + len), "r"(matrix) \
2411 : XMM_CLOBBERS("%xmm0", "%xmm1", "%xmm2", "%xmm3", \
2412 "%xmm4", "%xmm5", "%xmm6", "%xmm7",) \
2413 "memory" \
2414 );
2415
2416#define MIX_MISC(stereo) \
2417 __asm__ volatile ( \
2418 "1: \n" \
2419 "movaps (%3, %0), %%xmm0 \n" \
2420 stereo("movaps %%xmm0, %%xmm1 \n") \
2421 "mulps %%xmm4, %%xmm0 \n" \
2422 stereo("mulps %%xmm5, %%xmm1 \n") \
2423 "lea 1024(%3, %0), %1 \n" \
2424 "mov %5, %2 \n" \
2425 "2: \n" \
2426 "movaps (%1), %%xmm2 \n" \
2427 stereo("movaps %%xmm2, %%xmm3 \n") \
2428 "mulps (%4, %2), %%xmm2 \n" \
2429 stereo("mulps 16(%4, %2), %%xmm3 \n") \
2430 "addps %%xmm2, %%xmm0 \n" \
2431 stereo("addps %%xmm3, %%xmm1 \n") \
2432 "add $1024, %1 \n" \
2433 "add $32, %2 \n" \
2434 "jl 2b \n" \
2435 "movaps %%xmm0, (%3, %0) \n" \
2436 stereo("movaps %%xmm1, 1024(%3, %0) \n") \
2437 "add $16, %0 \n" \
2438 "jl 1b \n" \
2439 : "+&r"(i), "=&r"(j), "=&r"(k) \
2440 : "r"(samples[0] + len), "r"(matrix_simd + in_ch), \
2441 "g"((intptr_t) - 32 * (in_ch - 1)) \
2442 : "memory" \
2443 );
2444
2445static void ac3_downmix_sse(float (*samples)[256], float (*matrix)[2],
2446 int out_ch, int in_ch, int len)
2447{
2448 int (*matrix_cmp)[2] = (int(*)[2])matrix;
2449 intptr_t i, j, k;
2450
2451 i = -len * sizeof(float);
2452 if (in_ch == 5 && out_ch == 2 &&
2453 !(matrix_cmp[0][1] | matrix_cmp[2][0] |
2454 matrix_cmp[3][1] | matrix_cmp[4][0] |
2455 (matrix_cmp[1][0] ^ matrix_cmp[1][1]) |
2456 (matrix_cmp[0][0] ^ matrix_cmp[2][1]))) {
2457 MIX5(IF0, IF1);
2458 } else if (in_ch == 5 && out_ch == 1 &&
2459 matrix_cmp[0][0] == matrix_cmp[2][0] &&
2460 matrix_cmp[3][0] == matrix_cmp[4][0]) {
2461 MIX5(IF1, IF0);
2462 } else {
2463 DECLARE_ALIGNED(16, float, matrix_simd)[AC3_MAX_CHANNELS][2][4];
2464 j = 2 * in_ch * sizeof(float);
2465 __asm__ volatile (
2466 "1: \n"
2467 "sub $8, %0 \n"
2468 "movss (%2, %0), %%xmm4 \n"
2469 "movss 4(%2, %0), %%xmm5 \n"
2470 "shufps $0, %%xmm4, %%xmm4 \n"
2471 "shufps $0, %%xmm5, %%xmm5 \n"
2472 "movaps %%xmm4, (%1, %0, 4) \n"
2473 "movaps %%xmm5, 16(%1, %0, 4) \n"
2474 "jg 1b \n"
2475 : "+&r"(j)
2476 : "r"(matrix_simd), "r"(matrix)
2477 : "memory"
2478 );
2479 if (out_ch == 2) {
2480 MIX_MISC(IF1);
2481 } else {
2482 MIX_MISC(IF0);
2483 }
2484 }
2485}
2486
2487#if HAVE_6REGS2377#if HAVE_6REGS
2488static void vector_fmul_window_3dnowext(float *dst, const float *src0,2378static void vector_fmul_window_3dnowext(float *dst, const float *src0,
2489 const float *src1, const float *win,2379 const float *src1, const float *win,
@@ -2958,7 +2848,6 @@ static void dsputil_init_sse(DSPContext *c, AVCodecContext *avctx, int mm_flags) · 468 unmodified lines
2958 }2848 }
2959 2849
2960 c->vorbis_inverse_coupling = vorbis_inverse_coupling_sse;2850 c->vorbis_inverse_coupling = vorbis_inverse_coupling_sse;
2961 c->ac3_downmix = ac3_downmix_sse;
2962 2851
2963#if HAVE_6REGS2852#if HAVE_6REGS
2964 c->vector_fmul_window = vector_fmul_window_sse;2853 c->vector_fmul_window = vector_fmul_window_sse;
libavformat/os_support.h+4−0
@@ -48,7 +48,11 @@ · 47 unmodified lines
48#endif /* defined(__MINGW32__) && !defined(__MINGW32CE__) */48#endif /* defined(__MINGW32__) && !defined(__MINGW32CE__) */
49 49
50#ifdef _WIN3250#ifdef _WIN32
51#if HAVE_DIRECT_H
51#include <direct.h>52#include <direct.h>
53#elif HAVE_IO_H
54#include <io.h>
55#endif
52#define mkdir(a, b) _mkdir(a)56#define mkdir(a, b) _mkdir(a)
53#else57#else
54#include <sys/stat.h>58#include <sys/stat.h>
libavresample/audio_mix_matrix.c+1−0
@@ -54,6 +54,7 @@ · 53 unmodified lines
54#define WIDE_RIGHT 3254#define WIDE_RIGHT 32
55#define SURROUND_DIRECT_LEFT 3355#define SURROUND_DIRECT_LEFT 33
56#define SURROUND_DIRECT_RIGHT 3456#define SURROUND_DIRECT_RIGHT 34
57#define LOW_FREQUENCY_2 35
57 58
58#define SQRT3_2 1.22474487139158904909 /* sqrt(3/2) */59#define SQRT3_2 1.22474487139158904909 /* sqrt(3/2) */
59 60
libavutil/audioconvert.c+1−0
@@ -59,6 +59,7 @@ static const struct channel_name channel_names[] = { · 58 unmodified lines
59 [32] = { "WR", "wide right" },59 [32] = { "WR", "wide right" },
60 [33] = { "SDL", "surround direct left" },60 [33] = { "SDL", "surround direct left" },
61 [34] = { "SDR", "surround direct right" },61 [34] = { "SDR", "surround direct right" },
62 [35] = { "LFE2", "low frequency 2" },
62};63};
63 64
64static const char *get_channel_name(int channel_id)65static const char *get_channel_name(int channel_id)
libavutil/audioconvert.h+1−0
@@ -62,6 +62,7 @@ · 61 unmodified lines
62#define AV_CH_WIDE_RIGHT 0x0000000100000000ULL62#define AV_CH_WIDE_RIGHT 0x0000000100000000ULL
63#define AV_CH_SURROUND_DIRECT_LEFT 0x0000000200000000ULL63#define AV_CH_SURROUND_DIRECT_LEFT 0x0000000200000000ULL
64#define AV_CH_SURROUND_DIRECT_RIGHT 0x0000000400000000ULL64#define AV_CH_SURROUND_DIRECT_RIGHT 0x0000000400000000ULL
65#define AV_CH_LOW_FREQUENCY_2 0x0000000800000000ULL
65 66
66/** Channel mask value used for AVCodecContext.request_channel_layout67/** Channel mask value used for AVCodecContext.request_channel_layout
67 to indicate that the user requests the channel order of the decoder output68 to indicate that the user requests the channel order of the decoder output
libavutil/version.h+1−1
@@ -39,7 +39,7 @@ · 38 unmodified lines
39 */39 */
40 40
41#define LIBAVUTIL_VERSION_MAJOR 5141#define LIBAVUTIL_VERSION_MAJOR 51
42#define LIBAVUTIL_VERSION_MINOR 7242#define LIBAVUTIL_VERSION_MINOR 73
43#define LIBAVUTIL_VERSION_MICRO 10043#define LIBAVUTIL_VERSION_MICRO 100
44 44
45#define LIBAVUTIL_VERSION_INT AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \45#define LIBAVUTIL_VERSION_INT AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \
libswscale/swscale.c+0−2
@@ -628,8 +628,6 @@ static int swScale(SwsContext *c, const uint8_t *src[], · 627 unmodified lines
628 }628 }
629 }629 }
630 } else {630 } else {
631 assert(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize * 2);
632 assert(chrUSrcPtr + vChrFilterSize - 1 < chrUPixBuf + vChrBufSize * 2);
633 if (c->yuv2packed1 && vLumFilterSize == 1 &&631 if (c->yuv2packed1 && vLumFilterSize == 1 &&
634 vChrFilterSize <= 2) { // unscaled RGB632 vChrFilterSize <= 2) { // unscaled RGB
635 int chrAlpha = vChrFilterSize == 1 ? 0 : vChrFilter[2 * dstY + 1];633 int chrAlpha = vChrFilterSize == 1 ? 0 : vChrFilter[2 * dstY + 1];

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t