* qatar/master:
rv30: return AVERROR(EINVAL) instead of EINVAL
build: add -L flags before existing LDFLAGS
simple_idct: whitespace cosmetics
simple_idct: make repeated code a macro
dsputil: remove huge #if 0 block
simple_idct: change 10-bit add/put stride from pixels to bytes
dsputil: allow 9/10-bit functions for non-h264 codecs
dnxhd: rename some data tables
dnxhdenc: remove inline from function only called through pointer
dnxhdenc: whitespace cosmetics
swscale: mark YUV422P10(LE,BE) as supported for output
configure: add -xc99 to LDFLAGS for Sun CC
Remove unused and non-compiling vestigial g729 decoder
Remove unused code under G729_BITEXACT #ifdef.
mpegvideo: fix invalid picture unreferencing.
dsputil: Remove extra blank line at end.
dsputil: Replace a LONG_MAX check with HAVE_FAST_64BIT.
simple_idct: add 10-bit version
Conflicts:
Makefile
libavcodec/g729data.h
libavcodec/g729dec.c
libavcodec/rv30.c
tests/ref/lavfi/pixdesc
tests/ref/lavfi/pixfmts_copy
tests/ref/lavfi/pixfmts_null
tests/ref/lavfi/pixfmts_scale
tests/ref/lavfi/pixfmts_vflip
Merged-by: Michael Niedermayer <michaelni@gmx.at>
| ... | ... |
@@ -27,8 +27,6 @@ ALLPROGS = $(BASENAMES:%=%$(EXESUF)) |
| 27 | 27 |
ALLPROGS_G = $(BASENAMES:%=%_g$(EXESUF)) |
| 28 | 28 |
ALLMANPAGES = $(BASENAMES:%=%.1) |
| 29 | 29 |
|
| 30 |
-ALLFFLIBS = avcodec avdevice avfilter avformat avutil postproc swscale |
|
| 31 |
- |
|
| 32 | 30 |
FFLIBS-$(CONFIG_AVDEVICE) += avdevice |
| 33 | 31 |
FFLIBS-$(CONFIG_AVFILTER) += avfilter |
| 34 | 32 |
FFLIBS-$(CONFIG_AVFORMAT) += avformat |
| ... | ... |
@@ -20,6 +20,8 @@ $(foreach VAR,$(SILENT),$(eval override $(VAR) = @$($(VAR)))) |
| 20 | 20 |
$(eval INSTALL = @$(call ECHO,INSTALL,$$(^:$(SRC_DIR)/%=%)); $(INSTALL)) |
| 21 | 21 |
endif |
| 22 | 22 |
|
| 23 |
+ALLFFLIBS = avcodec avdevice avfilter avformat avutil postproc swscale |
|
| 24 |
+ |
|
| 23 | 25 |
# NASM requires -I path terminated with / |
| 24 | 26 |
IFLAGS := -I. -I$(SRC_PATH)/ |
| 25 | 27 |
CPPFLAGS := $(IFLAGS) $(CPPFLAGS) |
| ... | ... |
@@ -27,7 +29,7 @@ CFLAGS += $(ECFLAGS) |
| 27 | 27 |
CCFLAGS = $(CFLAGS) |
| 28 | 28 |
YASMFLAGS += $(IFLAGS) -Pconfig.asm |
| 29 | 29 |
HOSTCFLAGS += $(IFLAGS) |
| 30 |
-LDFLAGS += $(ALLFFLIBS:%=-Llib%) |
|
| 30 |
+LDFLAGS := $(ALLFFLIBS:%=-Llib%) $(LDFLAGS) |
|
| 31 | 31 |
|
| 32 | 32 |
define COMPILE |
| 33 | 33 |
$($(1)DEP) |
| ... | ... |
@@ -2049,6 +2049,7 @@ elif $cc -V 2>&1 | grep -q Sun; then |
| 2049 | 2049 |
cc_ident=$($cc -V 2>&1 | head -n1 | cut -d' ' -f 2-) |
| 2050 | 2050 |
DEPEND_CMD='$(DEPCC) $(DEPFLAGS) $< | sed -e "1s,^.*: ,$@: ," -e "\$$!s,\$$, \\\," -e "1!s,^.*: , ," > $(@:.o=.d)' |
| 2051 | 2051 |
DEPFLAGS='$(CPPFLAGS) $(CFLAGS) -xM1' |
| 2052 |
+ add_ldflags -xc99 |
|
| 2052 | 2053 |
speed_cflags='-O5' |
| 2053 | 2054 |
size_cflags='-O5 -xspace' |
| 2054 | 2055 |
filter_cflags=suncc_flags |
| ... | ... |
@@ -105,20 +105,9 @@ int16_t ff_acelp_decode_gain_code( |
| 105 | 105 |
for(i=0; i<ma_pred_order; i++) |
| 106 | 106 |
mr_energy += quant_energy[i] * ma_prediction_coeff[i]; |
| 107 | 107 |
|
| 108 |
-#ifdef G729_BITEXACT |
|
| 109 |
- mr_energy += (((-6165LL * ff_log2(dsp->scalarproduct_int16(fc_v, fc_v, subframe_size, 0))) >> 3) & ~0x3ff); |
|
| 110 |
- |
|
| 111 |
- mr_energy = (5439 * (mr_energy >> 15)) >> 8; // (0.15) = (0.15) * (7.23) |
|
| 112 |
- |
|
| 113 |
- return bidir_sal( |
|
| 114 |
- ((ff_exp2(mr_energy & 0x7fff) + 16) >> 5) * (gain_corr_factor >> 1), |
|
| 115 |
- (mr_energy >> 15) - 25 |
|
| 116 |
- ); |
|
| 117 |
-#else |
|
| 118 | 108 |
mr_energy = gain_corr_factor * exp(M_LN10 / (20 << 23) * mr_energy) / |
| 119 | 109 |
sqrt(dsp->scalarproduct_int16(fc_v, fc_v, subframe_size, 0)); |
| 120 | 110 |
return mr_energy >> 12; |
| 121 |
-#endif |
|
| 122 | 111 |
} |
| 123 | 112 |
|
| 124 | 113 |
float ff_amr_set_fixed_gain(float fixed_gain_factor, float fixed_mean_energy, |
| ... | ... |
@@ -48,26 +48,6 @@ const uint8_t ff_fc_2pulses_9bits_track1_gray[16] = |
| 48 | 48 |
28, 26, |
| 49 | 49 |
}; |
| 50 | 50 |
|
| 51 |
-const uint8_t ff_fc_2pulses_9bits_track2_gray[32] = |
|
| 52 |
-{
|
|
| 53 |
- 0, 2, |
|
| 54 |
- 5, 4, |
|
| 55 |
- 12, 10, |
|
| 56 |
- 7, 9, |
|
| 57 |
- 25, 24, |
|
| 58 |
- 20, 22, |
|
| 59 |
- 14, 15, |
|
| 60 |
- 19, 17, |
|
| 61 |
- 36, 31, |
|
| 62 |
- 21, 26, |
|
| 63 |
- 1, 6, |
|
| 64 |
- 16, 11, |
|
| 65 |
- 27, 29, |
|
| 66 |
- 32, 30, |
|
| 67 |
- 39, 37, |
|
| 68 |
- 34, 35, |
|
| 69 |
-}; |
|
| 70 |
- |
|
| 71 | 51 |
const uint8_t ff_fc_4pulses_8bits_tracks_13[16] = |
| 72 | 52 |
{
|
| 73 | 53 |
0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, |
| ... | ... |
@@ -82,37 +82,6 @@ extern const uint8_t ff_fc_2pulses_9bits_track1[16]; |
| 82 | 82 |
extern const uint8_t ff_fc_2pulses_9bits_track1_gray[16]; |
| 83 | 83 |
|
| 84 | 84 |
/** |
| 85 |
- * Track|Pulse| Positions |
|
| 86 |
- * ----------------------------------------- |
|
| 87 |
- * 2 | 1 | 0, 7, 14, 20, 27, 34, 1, 21 |
|
| 88 |
- * | | 2, 9, 15, 22, 29, 35, 6, 26 |
|
| 89 |
- * | | 4,10, 17, 24, 30, 37, 11, 31 |
|
| 90 |
- * | | 5,12, 19, 25, 32, 39, 16, 36 |
|
| 91 |
- * ----------------------------------------- |
|
| 92 |
- * |
|
| 93 |
- * @remark Track in the table should be read top-to-bottom, left-to-right. |
|
| 94 |
- * |
|
| 95 |
- * @note (EE.1) This table (from the reference code) does not comply with |
|
| 96 |
- * the specification. |
|
| 97 |
- * The specification contains the following table: |
|
| 98 |
- * |
|
| 99 |
- * Track|Pulse| Positions |
|
| 100 |
- * ----------------------------------------- |
|
| 101 |
- * 2 | 1 | 0, 5, 10, 15, 20, 25, 30, 35 |
|
| 102 |
- * | | 1, 6, 11, 16, 21, 26, 31, 36 |
|
| 103 |
- * | | 2, 7, 12, 17, 22, 27, 32, 37 |
|
| 104 |
- * | | 4, 9, 14, 19, 24, 29, 34, 39 |
|
| 105 |
- * |
|
| 106 |
- * ----------------------------------------- |
|
| 107 |
- * |
|
| 108 |
- * @note (EE.2) Reference G.729D code also uses gray decoding for each |
|
| 109 |
- * pulse index before looking up the value in the table. |
|
| 110 |
- * |
|
| 111 |
- * Used in G.729 @@6.4k (with gray coding) |
|
| 112 |
- */ |
|
| 113 |
-extern const uint8_t ff_fc_2pulses_9bits_track2_gray[32]; |
|
| 114 |
- |
|
| 115 |
-/** |
|
| 116 | 85 |
* b60 hamming windowed sinc function coefficients |
| 117 | 86 |
*/ |
| 118 | 87 |
extern const float ff_b60_sinc[61]; |
| ... | ... |
@@ -335,7 +335,7 @@ void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx) |
| 335 | 335 |
put_pixels_clamped_axp_p = c->put_pixels_clamped; |
| 336 | 336 |
add_pixels_clamped_axp_p = c->add_pixels_clamped; |
| 337 | 337 |
|
| 338 |
- if (!avctx->lowres && |
|
| 338 |
+ if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && |
|
| 339 | 339 |
(avctx->idct_algo == FF_IDCT_AUTO || |
| 340 | 340 |
avctx->idct_algo == FF_IDCT_SIMPLEALPHA)) {
|
| 341 | 341 |
c->idct_put = ff_simple_idct_put_axp; |
| ... | ... |
@@ -83,7 +83,7 @@ |
| 83 | 83 |
/** Maximum sharpening factor |
| 84 | 84 |
* |
| 85 | 85 |
* The specification says 0.8, which should be 13107, but the reference C code |
| 86 |
- * uses 13017 instead. (Amusingly the same applies to SHARP_MAX in g729dec.c.) |
|
| 86 |
+ * uses 13017 instead. (Amusingly the same applies to SHARP_MAX in bitexact G.729.) |
|
| 87 | 87 |
*/ |
| 88 | 88 |
#define SHARP_MAX 0.79449462890625 |
| 89 | 89 |
|
| ... | ... |
@@ -80,7 +80,7 @@ void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx) |
| 80 | 80 |
ff_put_pixels_clamped = c->put_pixels_clamped; |
| 81 | 81 |
ff_add_pixels_clamped = c->add_pixels_clamped; |
| 82 | 82 |
|
| 83 |
- if (!avctx->lowres) {
|
|
| 83 |
+ if (!avctx->lowres && avctx->bits_per_raw_sample <= 8) {
|
|
| 84 | 84 |
if(avctx->idct_algo == FF_IDCT_AUTO || |
| 85 | 85 |
avctx->idct_algo == FF_IDCT_ARM){
|
| 86 | 86 |
c->idct_put = j_rev_dct_arm_put; |
| ... | ... |
@@ -29,8 +29,9 @@ void ff_prefetch_arm(void *mem, int stride, int h); |
| 29 | 29 |
|
| 30 | 30 |
void av_cold ff_dsputil_init_armv5te(DSPContext* c, AVCodecContext *avctx) |
| 31 | 31 |
{
|
| 32 |
- if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO || |
|
| 33 |
- avctx->idct_algo == FF_IDCT_SIMPLEARMV5TE)) {
|
|
| 32 |
+ if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && |
|
| 33 |
+ (avctx->idct_algo == FF_IDCT_AUTO || |
|
| 34 |
+ avctx->idct_algo == FF_IDCT_SIMPLEARMV5TE)) {
|
|
| 34 | 35 |
c->idct_put = ff_simple_idct_put_armv5te; |
| 35 | 36 |
c->idct_add = ff_simple_idct_add_armv5te; |
| 36 | 37 |
c->idct = ff_simple_idct_armv5te; |
| ... | ... |
@@ -74,8 +74,9 @@ void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx) |
| 74 | 74 |
{
|
| 75 | 75 |
const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; |
| 76 | 76 |
|
| 77 |
- if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO || |
|
| 78 |
- avctx->idct_algo == FF_IDCT_SIMPLEARMV6)) {
|
|
| 77 |
+ if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && |
|
| 78 |
+ (avctx->idct_algo == FF_IDCT_AUTO || |
|
| 79 |
+ avctx->idct_algo == FF_IDCT_SIMPLEARMV6)) {
|
|
| 79 | 80 |
c->idct_put = ff_simple_idct_put_armv6; |
| 80 | 81 |
c->idct_add = ff_simple_idct_add_armv6; |
| 81 | 82 |
c->idct = ff_simple_idct_armv6; |
| ... | ... |
@@ -177,7 +177,7 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx) |
| 177 | 177 |
{
|
| 178 | 178 |
const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; |
| 179 | 179 |
|
| 180 |
- if (!avctx->lowres) {
|
|
| 180 |
+ if (!avctx->lowres && avctx->bits_per_raw_sample <= 8) {
|
|
| 181 | 181 |
if (avctx->idct_algo == FF_IDCT_AUTO || |
| 182 | 182 |
avctx->idct_algo == FF_IDCT_SIMPLENEON) {
|
| 183 | 183 |
c->idct_put = ff_simple_idct_put_neon; |
| ... | ... |
@@ -256,16 +256,18 @@ void dsputil_init_bfin( DSPContext* c, AVCodecContext *avctx ) |
| 256 | 256 |
if (avctx->dct_algo == FF_DCT_AUTO) |
| 257 | 257 |
c->fdct = ff_bfin_fdct; |
| 258 | 258 |
|
| 259 |
- if (avctx->idct_algo==FF_IDCT_VP3) {
|
|
| 260 |
- c->idct_permutation_type = FF_NO_IDCT_PERM; |
|
| 261 |
- c->idct = ff_bfin_vp3_idct; |
|
| 262 |
- c->idct_add = ff_bfin_vp3_idct_add; |
|
| 263 |
- c->idct_put = ff_bfin_vp3_idct_put; |
|
| 264 |
- } else if (avctx->idct_algo == FF_IDCT_AUTO) {
|
|
| 265 |
- c->idct_permutation_type = FF_NO_IDCT_PERM; |
|
| 266 |
- c->idct = ff_bfin_idct; |
|
| 267 |
- c->idct_add = bfin_idct_add; |
|
| 268 |
- c->idct_put = bfin_idct_put; |
|
| 259 |
+ if (avctx->bits_per_raw_sample <= 8) {
|
|
| 260 |
+ if (avctx->idct_algo == FF_IDCT_VP3) {
|
|
| 261 |
+ c->idct_permutation_type = FF_NO_IDCT_PERM; |
|
| 262 |
+ c->idct = ff_bfin_vp3_idct; |
|
| 263 |
+ c->idct_add = ff_bfin_vp3_idct_add; |
|
| 264 |
+ c->idct_put = ff_bfin_vp3_idct_put; |
|
| 265 |
+ } else if (avctx->idct_algo == FF_IDCT_AUTO) {
|
|
| 266 |
+ c->idct_permutation_type = FF_NO_IDCT_PERM; |
|
| 267 |
+ c->idct = ff_bfin_idct; |
|
| 268 |
+ c->idct_add = bfin_idct_add; |
|
| 269 |
+ c->idct_put = bfin_idct_put; |
|
| 270 |
+ } |
|
| 269 | 271 |
} |
| 270 | 272 |
} |
| 271 | 273 |
|
| ... | ... |
@@ -27,82 +27,6 @@ |
| 27 | 27 |
#include "avcodec.h" |
| 28 | 28 |
#include "celp_math.h" |
| 29 | 29 |
|
| 30 |
-#ifdef G729_BITEXACT |
|
| 31 |
-/** |
|
| 32 |
- * Cosine table: base_cos[i] = (1<<15) * cos(i*PI/64) |
|
| 33 |
- */ |
|
| 34 |
-static const int16_t base_cos[64] = |
|
| 35 |
-{
|
|
| 36 |
- 32767, 32729, 32610, 32413, 32138, 31786, 31357, 30853, |
|
| 37 |
- 30274, 29622, 28899, 28106, 27246, 26320, 25330, 24279, |
|
| 38 |
- 23170, 22006, 20788, 19520, 18205, 16846, 15447, 14010, |
|
| 39 |
- 12540, 11039, 9512, 7962, 6393, 4808, 3212, 1608, |
|
| 40 |
- 0, -1608, -3212, -4808, -6393, -7962, -9512, -11039, |
|
| 41 |
- -12540, -14010, -15447, -16846, -18205, -19520, -20788, -22006, |
|
| 42 |
- -23170, -24279, -25330, -26320, -27246, -28106, -28899, -29622, |
|
| 43 |
- -30274, -30853, -31357, -31786, -32138, -32413, -32610, -32729 |
|
| 44 |
-}; |
|
| 45 |
- |
|
| 46 |
-/** |
|
| 47 |
- * Slope used to compute cos(x) |
|
| 48 |
- * |
|
| 49 |
- * cos(ind*64+offset) = base_cos[ind]+offset*slope_cos[ind] |
|
| 50 |
- * values multiplied by 1<<19 |
|
| 51 |
- */ |
|
| 52 |
-static const int16_t slope_cos[64] = |
|
| 53 |
-{
|
|
| 54 |
- -632, -1893, -3150, -4399, -5638, -6863, -8072, -9261, |
|
| 55 |
- -10428, -11570, -12684, -13767, -14817, -15832, -16808, -17744, |
|
| 56 |
- -18637, -19486, -20287, -21039, -21741, -22390, -22986, -23526, |
|
| 57 |
- -24009, -24435, -24801, -25108, -25354, -25540, -25664, -25726, |
|
| 58 |
- -25726, -25664, -25540, -25354, -25108, -24801, -24435, -24009, |
|
| 59 |
- -23526, -22986, -22390, -21741, -21039, -20287, -19486, -18637, |
|
| 60 |
- -17744, -16808, -15832, -14817, -13767, -12684, -11570, -10428, |
|
| 61 |
- -9261, -8072, -6863, -5638, -4399, -3150, -1893, -632 |
|
| 62 |
-}; |
|
| 63 |
- |
|
| 64 |
-/** |
|
| 65 |
- * Table used to compute exp2(x) |
|
| 66 |
- * |
|
| 67 |
- * tab_exp2[i] = (1<<14) * exp2(i/32) = 2^(i/32) i=0..32 |
|
| 68 |
- */ |
|
| 69 |
-static const uint16_t tab_exp2[33] = |
|
| 70 |
-{
|
|
| 71 |
- 16384, 16743, 17109, 17484, 17867, 18258, 18658, 19066, 19484, 19911, |
|
| 72 |
- 20347, 20792, 21247, 21713, 22188, 22674, 23170, 23678, 24196, 24726, |
|
| 73 |
- 25268, 25821, 26386, 26964, 27554, 28158, 28774, 29405, 30048, 30706, |
|
| 74 |
- 31379, 32066, 32767 |
|
| 75 |
-}; |
|
| 76 |
- |
|
| 77 |
-int16_t ff_cos(uint16_t arg) |
|
| 78 |
-{
|
|
| 79 |
- uint8_t offset= arg; |
|
| 80 |
- uint8_t ind = arg >> 8; |
|
| 81 |
- |
|
| 82 |
- assert(arg < 0x4000); |
|
| 83 |
- |
|
| 84 |
- return FFMAX(base_cos[ind] + ((slope_cos[ind] * offset) >> 12), -0x8000); |
|
| 85 |
-} |
|
| 86 |
- |
|
| 87 |
-int ff_exp2(uint16_t power) |
|
| 88 |
-{
|
|
| 89 |
- uint16_t frac_x0; |
|
| 90 |
- uint16_t frac_dx; |
|
| 91 |
- int result; |
|
| 92 |
- |
|
| 93 |
- assert(power <= 0x7fff); |
|
| 94 |
- |
|
| 95 |
- frac_x0 = power >> 10; |
|
| 96 |
- frac_dx = (power & 0x03ff) << 5; |
|
| 97 |
- |
|
| 98 |
- result = tab_exp2[frac_x0] << 15; |
|
| 99 |
- result += frac_dx * (tab_exp2[frac_x0+1] - tab_exp2[frac_x0]); |
|
| 100 |
- |
|
| 101 |
- return result >> 10; |
|
| 102 |
-} |
|
| 103 |
- |
|
| 104 |
-#else // G729_BITEXACT |
|
| 105 |
- |
|
| 106 | 30 |
/** |
| 107 | 31 |
* Cosine table: base_cos[i] = (1<<15) * cos(i*PI/64) |
| 108 | 32 |
*/ |
| ... | ... |
@@ -154,8 +78,6 @@ int ff_exp2(uint16_t power) |
| 154 | 154 |
return result + ((result*(power&31)*89)>>22); |
| 155 | 155 |
} |
| 156 | 156 |
|
| 157 |
-#endif // else G729_BITEXACT |
|
| 158 |
- |
|
| 159 | 157 |
/** |
| 160 | 158 |
* Table used to compute log2(x) |
| 161 | 159 |
* |
| ... | ... |
@@ -163,17 +85,10 @@ int ff_exp2(uint16_t power) |
| 163 | 163 |
*/ |
| 164 | 164 |
static const uint16_t tab_log2[33] = |
| 165 | 165 |
{
|
| 166 |
-#ifdef G729_BITEXACT |
|
| 167 |
- 0, 1455, 2866, 4236, 5568, 6863, 8124, 9352, |
|
| 168 |
- 10549, 11716, 12855, 13967, 15054, 16117, 17156, 18172, |
|
| 169 |
- 19167, 20142, 21097, 22033, 22951, 23852, 24735, 25603, |
|
| 170 |
- 26455, 27291, 28113, 28922, 29716, 30497, 31266, 32023, 32767, |
|
| 171 |
-#else |
|
| 172 | 166 |
4, 1459, 2870, 4240, 5572, 6867, 8127, 9355, |
| 173 | 167 |
10552, 11719, 12858, 13971, 15057, 16120, 17158, 18175, |
| 174 | 168 |
19170, 20145, 21100, 22036, 22954, 23854, 24738, 25605, |
| 175 | 169 |
26457, 27294, 28116, 28924, 29719, 30500, 31269, 32025, 32769, |
| 176 |
-#endif |
|
| 177 | 170 |
}; |
| 178 | 171 |
|
| 179 | 172 |
int ff_log2(uint32_t value) |
| ... | ... |
@@ -111,7 +111,7 @@ static const struct algo idct_tab[] = {
|
| 111 | 111 |
{ "FAANI", ff_faanidct, NO_PERM },
|
| 112 | 112 |
{ "REF-DBL", ff_ref_idct, NO_PERM },
|
| 113 | 113 |
{ "INT", j_rev_dct, MMX_PERM },
|
| 114 |
- { "SIMPLE-C", ff_simple_idct, NO_PERM },
|
|
| 114 |
+ { "SIMPLE-C", ff_simple_idct_8, NO_PERM },
|
|
| 115 | 115 |
|
| 116 | 116 |
#if HAVE_MMX |
| 117 | 117 |
#if CONFIG_GPL |
| ... | ... |
@@ -41,7 +41,7 @@ static const AVClass class = { "dnxhd", av_default_item_name, options, LIBAVUTIL
|
| 41 | 41 |
|
| 42 | 42 |
#define LAMBDA_FRAC_BITS 10 |
| 43 | 43 |
|
| 44 |
-static av_always_inline void dnxhd_get_pixels_8x4(DCTELEM *restrict block, const uint8_t *pixels, int line_size) |
|
| 44 |
+static void dnxhd_get_pixels_8x4(DCTELEM *restrict block, const uint8_t *pixels, int line_size) |
|
| 45 | 45 |
{
|
| 46 | 46 |
int i; |
| 47 | 47 |
for (i = 0; i < 4; i++) {
|
| ... | ... |
@@ -52,10 +52,10 @@ static av_always_inline void dnxhd_get_pixels_8x4(DCTELEM *restrict block, const |
| 52 | 52 |
pixels += line_size; |
| 53 | 53 |
block += 8; |
| 54 | 54 |
} |
| 55 |
- memcpy(block , block- 8, sizeof(*block)*8); |
|
| 56 |
- memcpy(block+ 8, block-16, sizeof(*block)*8); |
|
| 57 |
- memcpy(block+16, block-24, sizeof(*block)*8); |
|
| 58 |
- memcpy(block+24, block-32, sizeof(*block)*8); |
|
| 55 |
+ memcpy(block, block - 8, sizeof(*block) * 8); |
|
| 56 |
+ memcpy(block + 8, block - 16, sizeof(*block) * 8); |
|
| 57 |
+ memcpy(block + 16, block - 24, sizeof(*block) * 8); |
|
| 58 |
+ memcpy(block + 24, block - 32, sizeof(*block) * 8); |
|
| 59 | 59 |
} |
| 60 | 60 |
|
| 61 | 61 |
static int dnxhd_init_vlc(DNXHDEncContext *ctx) |
| ... | ... |
@@ -64,9 +64,9 @@ static int dnxhd_init_vlc(DNXHDEncContext *ctx) |
| 64 | 64 |
int max_level = 1<<(ctx->cid_table->bit_depth+2); |
| 65 | 65 |
|
| 66 | 66 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->vlc_codes, max_level*4*sizeof(*ctx->vlc_codes), fail); |
| 67 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->vlc_bits , max_level*4*sizeof(*ctx->vlc_bits ), fail); |
|
| 68 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->run_codes, 63*2 , fail); |
|
| 69 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->run_bits , 63 , fail); |
|
| 67 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->vlc_bits, max_level*4*sizeof(*ctx->vlc_bits) , fail); |
|
| 68 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->run_codes, 63*2, fail); |
|
| 69 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->run_bits, 63, fail); |
|
| 70 | 70 |
|
| 71 | 71 |
ctx->vlc_codes += max_level*2; |
| 72 | 72 |
ctx->vlc_bits += max_level*2; |
| ... | ... |
@@ -119,8 +119,8 @@ static int dnxhd_init_qmat(DNXHDEncContext *ctx, int lbias, int cbias) |
| 119 | 119 |
uint16_t weight_matrix[64] = {1,}; // convert_matrix needs uint16_t*
|
| 120 | 120 |
int qscale, i; |
| 121 | 121 |
|
| 122 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_l, (ctx->m.avctx->qmax+1) * 64 * sizeof(int) , fail); |
|
| 123 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_c, (ctx->m.avctx->qmax+1) * 64 * sizeof(int) , fail); |
|
| 122 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_l, (ctx->m.avctx->qmax+1) * 64 * sizeof(int), fail); |
|
| 123 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_c, (ctx->m.avctx->qmax+1) * 64 * sizeof(int), fail); |
|
| 124 | 124 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_l16, (ctx->m.avctx->qmax+1) * 64 * 2 * sizeof(uint16_t), fail); |
| 125 | 125 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->qmatrix_c16, (ctx->m.avctx->qmax+1) * 64 * 2 * sizeof(uint16_t), fail); |
| 126 | 126 |
|
| ... | ... |
@@ -218,7 +218,7 @@ static int dnxhd_encode_init(AVCodecContext *avctx) |
| 218 | 218 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->slice_size, ctx->m.mb_height*sizeof(uint32_t), fail); |
| 219 | 219 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->slice_offs, ctx->m.mb_height*sizeof(uint32_t), fail); |
| 220 | 220 |
FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->mb_bits, ctx->m.mb_num *sizeof(uint16_t), fail); |
| 221 |
- FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->mb_qscale, ctx->m.mb_num *sizeof(uint8_t) , fail); |
|
| 221 |
+ FF_ALLOCZ_OR_GOTO(ctx->m.avctx, ctx->mb_qscale, ctx->m.mb_num *sizeof(uint8_t), fail); |
|
| 222 | 222 |
|
| 223 | 223 |
ctx->frame.key_frame = 1; |
| 224 | 224 |
ctx->frame.pict_type = AV_PICTURE_TYPE_I; |
| ... | ... |
@@ -341,7 +341,7 @@ static av_always_inline int dnxhd_ssd_block(DCTELEM *qblock, DCTELEM *block) |
| 341 | 341 |
int score = 0; |
| 342 | 342 |
int i; |
| 343 | 343 |
for (i = 0; i < 64; i++) |
| 344 |
- score += (block[i]-qblock[i])*(block[i]-qblock[i]); |
|
| 344 |
+ score += (block[i] - qblock[i]) * (block[i] - qblock[i]); |
|
| 345 | 345 |
return score; |
| 346 | 346 |
} |
| 347 | 347 |
|
| ... | ... |
@@ -369,26 +369,28 @@ static av_always_inline void dnxhd_get_blocks(DNXHDEncContext *ctx, int mb_x, in |
| 369 | 369 |
const uint8_t *ptr_v = ctx->thread[0]->src[2] + ((mb_y << 4) * ctx->m.uvlinesize) + (mb_x << 3); |
| 370 | 370 |
DSPContext *dsp = &ctx->m.dsp; |
| 371 | 371 |
|
| 372 |
- dsp->get_pixels(ctx->blocks[0], ptr_y , ctx->m.linesize); |
|
| 372 |
+ dsp->get_pixels(ctx->blocks[0], ptr_y, ctx->m.linesize); |
|
| 373 | 373 |
dsp->get_pixels(ctx->blocks[1], ptr_y + 8, ctx->m.linesize); |
| 374 |
- dsp->get_pixels(ctx->blocks[2], ptr_u , ctx->m.uvlinesize); |
|
| 375 |
- dsp->get_pixels(ctx->blocks[3], ptr_v , ctx->m.uvlinesize); |
|
| 374 |
+ dsp->get_pixels(ctx->blocks[2], ptr_u, ctx->m.uvlinesize); |
|
| 375 |
+ dsp->get_pixels(ctx->blocks[3], ptr_v, ctx->m.uvlinesize); |
|
| 376 | 376 |
|
| 377 | 377 |
if (mb_y+1 == ctx->m.mb_height && ctx->m.avctx->height == 1080) {
|
| 378 | 378 |
if (ctx->interlaced) {
|
| 379 |
- ctx->get_pixels_8x4_sym(ctx->blocks[4], ptr_y + ctx->dct_y_offset , ctx->m.linesize); |
|
| 379 |
+ ctx->get_pixels_8x4_sym(ctx->blocks[4], ptr_y + ctx->dct_y_offset, ctx->m.linesize); |
|
| 380 | 380 |
ctx->get_pixels_8x4_sym(ctx->blocks[5], ptr_y + ctx->dct_y_offset + 8, ctx->m.linesize); |
| 381 |
- ctx->get_pixels_8x4_sym(ctx->blocks[6], ptr_u + ctx->dct_uv_offset , ctx->m.uvlinesize); |
|
| 382 |
- ctx->get_pixels_8x4_sym(ctx->blocks[7], ptr_v + ctx->dct_uv_offset , ctx->m.uvlinesize); |
|
| 381 |
+ ctx->get_pixels_8x4_sym(ctx->blocks[6], ptr_u + ctx->dct_uv_offset, ctx->m.uvlinesize); |
|
| 382 |
+ ctx->get_pixels_8x4_sym(ctx->blocks[7], ptr_v + ctx->dct_uv_offset, ctx->m.uvlinesize); |
|
| 383 | 383 |
} else {
|
| 384 |
- dsp->clear_block(ctx->blocks[4]); dsp->clear_block(ctx->blocks[5]); |
|
| 385 |
- dsp->clear_block(ctx->blocks[6]); dsp->clear_block(ctx->blocks[7]); |
|
| 384 |
+ dsp->clear_block(ctx->blocks[4]); |
|
| 385 |
+ dsp->clear_block(ctx->blocks[5]); |
|
| 386 |
+ dsp->clear_block(ctx->blocks[6]); |
|
| 387 |
+ dsp->clear_block(ctx->blocks[7]); |
|
| 386 | 388 |
} |
| 387 | 389 |
} else {
|
| 388 |
- dsp->get_pixels(ctx->blocks[4], ptr_y + ctx->dct_y_offset , ctx->m.linesize); |
|
| 390 |
+ dsp->get_pixels(ctx->blocks[4], ptr_y + ctx->dct_y_offset, ctx->m.linesize); |
|
| 389 | 391 |
dsp->get_pixels(ctx->blocks[5], ptr_y + ctx->dct_y_offset + 8, ctx->m.linesize); |
| 390 |
- dsp->get_pixels(ctx->blocks[6], ptr_u + ctx->dct_uv_offset , ctx->m.uvlinesize); |
|
| 391 |
- dsp->get_pixels(ctx->blocks[7], ptr_v + ctx->dct_uv_offset , ctx->m.uvlinesize); |
|
| 392 |
+ dsp->get_pixels(ctx->blocks[6], ptr_u + ctx->dct_uv_offset, ctx->m.uvlinesize); |
|
| 393 |
+ dsp->get_pixels(ctx->blocks[7], ptr_v + ctx->dct_uv_offset, ctx->m.uvlinesize); |
|
| 392 | 394 |
} |
| 393 | 395 |
} |
| 394 | 396 |
|
| ... | ... |
@@ -496,14 +498,14 @@ static void dnxhd_setup_threads_slices(DNXHDEncContext *ctx) |
| 496 | 496 |
for (mb_y = 0; mb_y < ctx->m.mb_height; mb_y++) {
|
| 497 | 497 |
int thread_size; |
| 498 | 498 |
ctx->slice_offs[mb_y] = offset; |
| 499 |
- ctx->slice_size[mb_y] = 0; |
|
| 500 |
- for (mb_x = 0; mb_x < ctx->m.mb_width; mb_x++) {
|
|
| 501 |
- unsigned mb = mb_y * ctx->m.mb_width + mb_x; |
|
| 502 |
- ctx->slice_size[mb_y] += ctx->mb_bits[mb]; |
|
| 503 |
- } |
|
| 504 |
- ctx->slice_size[mb_y] = (ctx->slice_size[mb_y]+31)&~31; |
|
| 505 |
- ctx->slice_size[mb_y] >>= 3; |
|
| 506 |
- thread_size = ctx->slice_size[mb_y]; |
|
| 499 |
+ ctx->slice_size[mb_y] = 0; |
|
| 500 |
+ for (mb_x = 0; mb_x < ctx->m.mb_width; mb_x++) {
|
|
| 501 |
+ unsigned mb = mb_y * ctx->m.mb_width + mb_x; |
|
| 502 |
+ ctx->slice_size[mb_y] += ctx->mb_bits[mb]; |
|
| 503 |
+ } |
|
| 504 |
+ ctx->slice_size[mb_y] = (ctx->slice_size[mb_y]+31)&~31; |
|
| 505 |
+ ctx->slice_size[mb_y] >>= 3; |
|
| 506 |
+ thread_size = ctx->slice_size[mb_y]; |
|
| 507 | 507 |
offset += thread_size; |
| 508 | 508 |
} |
| 509 | 509 |
} |
| ... | ... |
@@ -184,7 +184,7 @@ static int pix_norm1_c(uint8_t * pix, int line_size) |
| 184 | 184 |
s += sq[pix[6]]; |
| 185 | 185 |
s += sq[pix[7]]; |
| 186 | 186 |
#else |
| 187 |
-#if LONG_MAX > 2147483647 |
|
| 187 |
+#if HAVE_FAST_64BIT |
|
| 188 | 188 |
register uint64_t x=*(uint64_t*)pix; |
| 189 | 189 |
s += sq[x&0xff]; |
| 190 | 190 |
s += sq[(x>>8)&0xff]; |
| ... | ... |
@@ -2225,7 +2225,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s |
| 2225 | 2225 |
|
| 2226 | 2226 |
s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i); |
| 2227 | 2227 |
s->dct_unquantize_inter(s, temp, 0, s->qscale); |
| 2228 |
- ff_simple_idct(temp); //FIXME |
|
| 2228 |
+ ff_simple_idct_8(temp); //FIXME |
|
| 2229 | 2229 |
|
| 2230 | 2230 |
for(i=0; i<64; i++) |
| 2231 | 2231 |
sum+= (temp[i]-bak[i])*(temp[i]-bak[i]); |
| ... | ... |
@@ -2866,6 +2866,12 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) |
| 2866 | 2866 |
c->idct = j_rev_dct1; |
| 2867 | 2867 |
c->idct_permutation_type= FF_NO_IDCT_PERM; |
| 2868 | 2868 |
}else{
|
| 2869 |
+ if (avctx->bits_per_raw_sample == 10) {
|
|
| 2870 |
+ c->idct_put = ff_simple_idct_put_10; |
|
| 2871 |
+ c->idct_add = ff_simple_idct_add_10; |
|
| 2872 |
+ c->idct = ff_simple_idct_10; |
|
| 2873 |
+ c->idct_permutation_type = FF_NO_IDCT_PERM; |
|
| 2874 |
+ } else {
|
|
| 2869 | 2875 |
if(avctx->idct_algo==FF_IDCT_INT){
|
| 2870 | 2876 |
c->idct_put= ff_jref_idct_put; |
| 2871 | 2877 |
c->idct_add= ff_jref_idct_add; |
| ... | ... |
@@ -2896,11 +2902,12 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) |
| 2896 | 2896 |
c->idct_put = ff_bink_idct_put_c; |
| 2897 | 2897 |
c->idct_permutation_type = FF_NO_IDCT_PERM; |
| 2898 | 2898 |
}else{ //accurate/default
|
| 2899 |
- c->idct_put= ff_simple_idct_put; |
|
| 2900 |
- c->idct_add= ff_simple_idct_add; |
|
| 2901 |
- c->idct = ff_simple_idct; |
|
| 2899 |
+ c->idct_put = ff_simple_idct_put_8; |
|
| 2900 |
+ c->idct_add = ff_simple_idct_add_8; |
|
| 2901 |
+ c->idct = ff_simple_idct_8; |
|
| 2902 | 2902 |
c->idct_permutation_type= FF_NO_IDCT_PERM; |
| 2903 | 2903 |
} |
| 2904 |
+ } |
|
| 2904 | 2905 |
} |
| 2905 | 2906 |
|
| 2906 | 2907 |
c->get_pixels = get_pixels_c; |
| ... | ... |
@@ -3169,21 +3176,18 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) |
| 3169 | 3169 |
dspfunc2(avg_h264_qpel, 1, 8, depth);\ |
| 3170 | 3170 |
dspfunc2(avg_h264_qpel, 2, 4, depth); |
| 3171 | 3171 |
|
| 3172 |
- if (avctx->codec_id != CODEC_ID_H264 || avctx->bits_per_raw_sample == 8) {
|
|
| 3173 |
- BIT_DEPTH_FUNCS(8) |
|
| 3174 |
- } else {
|
|
| 3175 |
- switch (avctx->bits_per_raw_sample) {
|
|
| 3176 |
- case 9: |
|
| 3177 |
- BIT_DEPTH_FUNCS(9) |
|
| 3178 |
- break; |
|
| 3179 |
- case 10: |
|
| 3180 |
- BIT_DEPTH_FUNCS(10) |
|
| 3181 |
- break; |
|
| 3182 |
- default: |
|
| 3183 |
- av_log(avctx, AV_LOG_DEBUG, "Unsupported bit depth: %d\n", avctx->bits_per_raw_sample); |
|
| 3184 |
- BIT_DEPTH_FUNCS(8) |
|
| 3185 |
- break; |
|
| 3186 |
- } |
|
| 3172 |
+ switch (avctx->bits_per_raw_sample) {
|
|
| 3173 |
+ case 9: |
|
| 3174 |
+ BIT_DEPTH_FUNCS(9); |
|
| 3175 |
+ break; |
|
| 3176 |
+ case 10: |
|
| 3177 |
+ BIT_DEPTH_FUNCS(10); |
|
| 3178 |
+ break; |
|
| 3179 |
+ default: |
|
| 3180 |
+ av_log(avctx, AV_LOG_DEBUG, "Unsupported bit depth: %d\n", avctx->bits_per_raw_sample); |
|
| 3181 |
+ case 8: |
|
| 3182 |
+ BIT_DEPTH_FUNCS(8); |
|
| 3183 |
+ break; |
|
| 3187 | 3184 |
} |
| 3188 | 3185 |
|
| 3189 | 3186 |
|
| ... | ... |
@@ -3243,4 +3247,3 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) |
| 3243 | 3243 |
av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n"); |
| 3244 | 3244 |
} |
| 3245 | 3245 |
} |
| 3246 |
- |
| ... | ... |
@@ -230,150 +230,6 @@ static void FUNCC(add_pixels4)(uint8_t *restrict p_pixels, DCTELEM *p_block, int |
| 230 | 230 |
} |
| 231 | 231 |
} |
| 232 | 232 |
|
| 233 |
-#if 0 |
|
| 234 |
- |
|
| 235 |
-#define PIXOP2(OPNAME, OP) \ |
|
| 236 |
-static void OPNAME ## _pixels(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 237 |
-{\
|
|
| 238 |
- int i;\ |
|
| 239 |
- for(i=0; i<h; i++){\
|
|
| 240 |
- OP(*((uint64_t*)block), AV_RN64(pixels));\ |
|
| 241 |
- pixels+=line_size;\ |
|
| 242 |
- block +=line_size;\ |
|
| 243 |
- }\ |
|
| 244 |
-}\ |
|
| 245 |
-\ |
|
| 246 |
-static void OPNAME ## _no_rnd_pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 247 |
-{\
|
|
| 248 |
- int i;\ |
|
| 249 |
- for(i=0; i<h; i++){\
|
|
| 250 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 251 |
- const uint64_t b= AV_RN64(pixels+1);\ |
|
| 252 |
- OP(*((uint64_t*)block), (a&b) + (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\ |
|
| 253 |
- pixels+=line_size;\ |
|
| 254 |
- block +=line_size;\ |
|
| 255 |
- }\ |
|
| 256 |
-}\ |
|
| 257 |
-\ |
|
| 258 |
-static void OPNAME ## _pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 259 |
-{\
|
|
| 260 |
- int i;\ |
|
| 261 |
- for(i=0; i<h; i++){\
|
|
| 262 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 263 |
- const uint64_t b= AV_RN64(pixels+1);\ |
|
| 264 |
- OP(*((uint64_t*)block), (a|b) - (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\ |
|
| 265 |
- pixels+=line_size;\ |
|
| 266 |
- block +=line_size;\ |
|
| 267 |
- }\ |
|
| 268 |
-}\ |
|
| 269 |
-\ |
|
| 270 |
-static void OPNAME ## _no_rnd_pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 271 |
-{\
|
|
| 272 |
- int i;\ |
|
| 273 |
- for(i=0; i<h; i++){\
|
|
| 274 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 275 |
- const uint64_t b= AV_RN64(pixels+line_size);\ |
|
| 276 |
- OP(*((uint64_t*)block), (a&b) + (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\ |
|
| 277 |
- pixels+=line_size;\ |
|
| 278 |
- block +=line_size;\ |
|
| 279 |
- }\ |
|
| 280 |
-}\ |
|
| 281 |
-\ |
|
| 282 |
-static void OPNAME ## _pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 283 |
-{\
|
|
| 284 |
- int i;\ |
|
| 285 |
- for(i=0; i<h; i++){\
|
|
| 286 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 287 |
- const uint64_t b= AV_RN64(pixels+line_size);\ |
|
| 288 |
- OP(*((uint64_t*)block), (a|b) - (((a^b)&0xFEFEFEFEFEFEFEFEULL)>>1));\ |
|
| 289 |
- pixels+=line_size;\ |
|
| 290 |
- block +=line_size;\ |
|
| 291 |
- }\ |
|
| 292 |
-}\ |
|
| 293 |
-\ |
|
| 294 |
-static void OPNAME ## _pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 295 |
-{\
|
|
| 296 |
- int i;\ |
|
| 297 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 298 |
- const uint64_t b= AV_RN64(pixels+1);\ |
|
| 299 |
- uint64_t l0= (a&0x0303030303030303ULL)\ |
|
| 300 |
- + (b&0x0303030303030303ULL)\ |
|
| 301 |
- + 0x0202020202020202ULL;\ |
|
| 302 |
- uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 303 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 304 |
- uint64_t l1,h1;\ |
|
| 305 |
-\ |
|
| 306 |
- pixels+=line_size;\ |
|
| 307 |
- for(i=0; i<h; i+=2){\
|
|
| 308 |
- uint64_t a= AV_RN64(pixels );\ |
|
| 309 |
- uint64_t b= AV_RN64(pixels+1);\ |
|
| 310 |
- l1= (a&0x0303030303030303ULL)\ |
|
| 311 |
- + (b&0x0303030303030303ULL);\ |
|
| 312 |
- h1= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 313 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 314 |
- OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ |
|
| 315 |
- pixels+=line_size;\ |
|
| 316 |
- block +=line_size;\ |
|
| 317 |
- a= AV_RN64(pixels );\ |
|
| 318 |
- b= AV_RN64(pixels+1);\ |
|
| 319 |
- l0= (a&0x0303030303030303ULL)\ |
|
| 320 |
- + (b&0x0303030303030303ULL)\ |
|
| 321 |
- + 0x0202020202020202ULL;\ |
|
| 322 |
- h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 323 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 324 |
- OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ |
|
| 325 |
- pixels+=line_size;\ |
|
| 326 |
- block +=line_size;\ |
|
| 327 |
- }\ |
|
| 328 |
-}\ |
|
| 329 |
-\ |
|
| 330 |
-static void OPNAME ## _no_rnd_pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ |
|
| 331 |
-{\
|
|
| 332 |
- int i;\ |
|
| 333 |
- const uint64_t a= AV_RN64(pixels );\ |
|
| 334 |
- const uint64_t b= AV_RN64(pixels+1);\ |
|
| 335 |
- uint64_t l0= (a&0x0303030303030303ULL)\ |
|
| 336 |
- + (b&0x0303030303030303ULL)\ |
|
| 337 |
- + 0x0101010101010101ULL;\ |
|
| 338 |
- uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 339 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 340 |
- uint64_t l1,h1;\ |
|
| 341 |
-\ |
|
| 342 |
- pixels+=line_size;\ |
|
| 343 |
- for(i=0; i<h; i+=2){\
|
|
| 344 |
- uint64_t a= AV_RN64(pixels );\ |
|
| 345 |
- uint64_t b= AV_RN64(pixels+1);\ |
|
| 346 |
- l1= (a&0x0303030303030303ULL)\ |
|
| 347 |
- + (b&0x0303030303030303ULL);\ |
|
| 348 |
- h1= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 349 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 350 |
- OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ |
|
| 351 |
- pixels+=line_size;\ |
|
| 352 |
- block +=line_size;\ |
|
| 353 |
- a= AV_RN64(pixels );\ |
|
| 354 |
- b= AV_RN64(pixels+1);\ |
|
| 355 |
- l0= (a&0x0303030303030303ULL)\ |
|
| 356 |
- + (b&0x0303030303030303ULL)\ |
|
| 357 |
- + 0x0101010101010101ULL;\ |
|
| 358 |
- h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ |
|
| 359 |
- + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ |
|
| 360 |
- OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ |
|
| 361 |
- pixels+=line_size;\ |
|
| 362 |
- block +=line_size;\ |
|
| 363 |
- }\ |
|
| 364 |
-}\ |
|
| 365 |
-\ |
|
| 366 |
-CALL_2X_PIXELS(OPNAME ## _pixels16_c , OPNAME ## _pixels_c , 8*sizeof(pixel))\ |
|
| 367 |
-CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels_x2_c , 8*sizeof(pixel))\ |
|
| 368 |
-CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels_y2_c , 8*sizeof(pixel))\ |
|
| 369 |
-CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels_xy2_c, 8*sizeof(pixel))\ |
|
| 370 |
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels_x2_c , 8*sizeof(pixel))\ |
|
| 371 |
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels_y2_c , 8*sizeof(pixel))\ |
|
| 372 |
-CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels_xy2_c, 8*sizeof(pixel)) |
|
| 373 |
- |
|
| 374 |
-#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEFEFEFEFEULL)>>1) ) |
|
| 375 |
-#else // 64 bit variant |
|
| 376 |
- |
|
| 377 | 233 |
#define PIXOP2(OPNAME, OP) \ |
| 378 | 234 |
static void FUNCC(OPNAME ## _pixels2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
|
| 379 | 235 |
int i;\ |
| ... | ... |
@@ -749,7 +605,6 @@ CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pi |
| 749 | 749 |
CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\ |
| 750 | 750 |
|
| 751 | 751 |
#define op_avg(a, b) a = rnd_avg_pixel4(a, b) |
| 752 |
-#endif |
|
| 753 | 752 |
#define op_put(a, b) a = b |
| 754 | 753 |
|
| 755 | 754 |
PIXOP2(avg, op_avg) |
| 756 | 755 |
deleted file mode 100644 |
| ... | ... |
@@ -1,278 +0,0 @@ |
| 1 |
-/* |
|
| 2 |
- * data for G.729 decoder |
|
| 3 |
- * Copyright (c) 2007 Vladimir Voroshilov |
|
| 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 |
- |
|
| 22 |
-#ifndef AVCODEC_G729DATA_H |
|
| 23 |
-#define AVCODEC_G729DATA_H |
|
| 24 |
- |
|
| 25 |
-#include <stdint.h> |
|
| 26 |
- |
|
| 27 |
-#define MA_NP 4 ///< Moving Average (MA) prediction order |
|
| 28 |
- |
|
| 29 |
-#define VQ_1ST_BITS 7 ///< first stage vector of quantizer (size in bits) |
|
| 30 |
-#define VQ_2ND_BITS 5 ///< second stage vector of quantizer (size in bits) |
|
| 31 |
- |
|
| 32 |
-#define GC_1ST_IDX_BITS_8K 3 ///< gain codebook (first stage) index, 8k mode (size in bits) |
|
| 33 |
-#define GC_2ND_IDX_BITS_8K 4 ///< gain codebook (second stage) index, 8k mode (size in bits) |
|
| 34 |
- |
|
| 35 |
-#define GC_1ST_IDX_BITS_6K4 3 ///< gain codebook (first stage) index, 6.4k mode (size in bits) |
|
| 36 |
-#define GC_2ND_IDX_BITS_6K4 3 ///< gain codebook (second stage) index, 6.4k mode (size in bits) |
|
| 37 |
- |
|
| 38 |
-/** |
|
| 39 |
- * first stage LSP codebook |
|
| 40 |
- * (10-dimensional, with 128 entries (3.24 of G.729) |
|
| 41 |
- */ |
|
| 42 |
-static const int16_t cb_lsp_1st[1<<VQ_1ST_BITS][10] = { /* (2.13) */
|
|
| 43 |
- { 1486, 2168, 3751, 9074, 12134, 13944, 17983, 19173, 21190, 21820},
|
|
| 44 |
- { 1730, 2640, 3450, 4870, 6126, 7876, 15644, 17817, 20294, 21902},
|
|
| 45 |
- { 1568, 2256, 3088, 4874, 11063, 13393, 18307, 19293, 21109, 21741},
|
|
| 46 |
- { 1733, 2512, 3357, 4708, 6977, 10296, 17024, 17956, 19145, 20350},
|
|
| 47 |
- { 1744, 2436, 3308, 8731, 10432, 12007, 15614, 16639, 21359, 21913},
|
|
| 48 |
- { 1786, 2369, 3372, 4521, 6795, 12963, 17674, 18988, 20855, 21640},
|
|
| 49 |
- { 1631, 2433, 3361, 6328, 10709, 12013, 13277, 13904, 19441, 21088},
|
|
| 50 |
- { 1489, 2364, 3291, 6250, 9227, 10403, 13843, 15278, 17721, 21451},
|
|
| 51 |
- { 1869, 2533, 3475, 4365, 9152, 14513, 15908, 17022, 20611, 21411},
|
|
| 52 |
- { 2070, 3025, 4333, 5854, 7805, 9231, 10597, 16047, 20109, 21834},
|
|
| 53 |
- { 1910, 2673, 3419, 4261, 11168, 15111, 16577, 17591, 19310, 20265},
|
|
| 54 |
- { 1141, 1815, 2624, 4623, 6495, 9588, 13968, 16428, 19351, 21286},
|
|
| 55 |
- { 2192, 3171, 4707, 5808, 10904, 12500, 14162, 15664, 21124, 21789},
|
|
| 56 |
- { 1286, 1907, 2548, 3453, 9574, 11964, 15978, 17344, 19691, 22495},
|
|
| 57 |
- { 1921, 2720, 4604, 6684, 11503, 12992, 14350, 15262, 16997, 20791},
|
|
| 58 |
- { 2052, 2759, 3897, 5246, 6638, 10267, 15834, 16814, 18149, 21675},
|
|
| 59 |
- { 1798, 2497, 5617, 11449, 13189, 14711, 17050, 18195, 20307, 21182},
|
|
| 60 |
- { 1009, 1647, 2889, 5709, 9541, 12354, 15231, 18494, 20966, 22033},
|
|
| 61 |
- { 3016, 3794, 5406, 7469, 12488, 13984, 15328, 16334, 19952, 20791},
|
|
| 62 |
- { 2203, 3040, 3796, 5442, 11987, 13512, 14931, 16370, 17856, 18803},
|
|
| 63 |
- { 2912, 4292, 7988, 9572, 11562, 13244, 14556, 16529, 20004, 21073},
|
|
| 64 |
- { 2861, 3607, 5923, 7034, 9234, 12054, 13729, 18056, 20262, 20974},
|
|
| 65 |
- { 3069, 4311, 5967, 7367, 11482, 12699, 14309, 16233, 18333, 19172},
|
|
| 66 |
- { 2434, 3661, 4866, 5798, 10383, 11722, 13049, 15668, 18862, 19831},
|
|
| 67 |
- { 2020, 2605, 3860, 9241, 13275, 14644, 16010, 17099, 19268, 20251},
|
|
| 68 |
- { 1877, 2809, 3590, 4707, 11056, 12441, 15622, 17168, 18761, 19907},
|
|
| 69 |
- { 2107, 2873, 3673, 5799, 13579, 14687, 15938, 17077, 18890, 19831},
|
|
| 70 |
- { 1612, 2284, 2944, 3572, 8219, 13959, 15924, 17239, 18592, 20117},
|
|
| 71 |
- { 2420, 3156, 6542, 10215, 12061, 13534, 15305, 16452, 18717, 19880},
|
|
| 72 |
- { 1667, 2612, 3534, 5237, 10513, 11696, 12940, 16798, 18058, 19378},
|
|
| 73 |
- { 2388, 3017, 4839, 9333, 11413, 12730, 15024, 16248, 17449, 18677},
|
|
| 74 |
- { 1875, 2786, 4231, 6320, 8694, 10149, 11785, 17013, 18608, 19960},
|
|
| 75 |
- { 679, 1411, 4654, 8006, 11446, 13249, 15763, 18127, 20361, 21567},
|
|
| 76 |
- { 1838, 2596, 3578, 4608, 5650, 11274, 14355, 15886, 20579, 21754},
|
|
| 77 |
- { 1303, 1955, 2395, 3322, 12023, 13764, 15883, 18077, 20180, 21232},
|
|
| 78 |
- { 1438, 2102, 2663, 3462, 8328, 10362, 13763, 17248, 19732, 22344},
|
|
| 79 |
- { 860, 1904, 6098, 7775, 9815, 12007, 14821, 16709, 19787, 21132},
|
|
| 80 |
- { 1673, 2723, 3704, 6125, 7668, 9447, 13683, 14443, 20538, 21731},
|
|
| 81 |
- { 1246, 1849, 2902, 4508, 7221, 12710, 14835, 16314, 19335, 22720},
|
|
| 82 |
- { 1525, 2260, 3862, 5659, 7342, 11748, 13370, 14442, 18044, 21334},
|
|
| 83 |
- { 1196, 1846, 3104, 7063, 10972, 12905, 14814, 17037, 19922, 22636},
|
|
| 84 |
- { 2147, 3106, 4475, 6511, 8227, 9765, 10984, 12161, 18971, 21300},
|
|
| 85 |
- { 1585, 2405, 2994, 4036, 11481, 13177, 14519, 15431, 19967, 21275},
|
|
| 86 |
- { 1778, 2688, 3614, 4680, 9465, 11064, 12473, 16320, 19742, 20800},
|
|
| 87 |
- { 1862, 2586, 3492, 6719, 11708, 13012, 14364, 16128, 19610, 20425},
|
|
| 88 |
- { 1395, 2156, 2669, 3386, 10607, 12125, 13614, 16705, 18976, 21367},
|
|
| 89 |
- { 1444, 2117, 3286, 6233, 9423, 12981, 14998, 15853, 17188, 21857},
|
|
| 90 |
- { 2004, 2895, 3783, 4897, 6168, 7297, 12609, 16445, 19297, 21465},
|
|
| 91 |
- { 1495, 2863, 6360, 8100, 11399, 14271, 15902, 17711, 20479, 22061},
|
|
| 92 |
- { 2484, 3114, 5718, 7097, 8400, 12616, 14073, 14847, 20535, 21396},
|
|
| 93 |
- { 2424, 3277, 5296, 6284, 11290, 12903, 16022, 17508, 19333, 20283},
|
|
| 94 |
- { 2565, 3778, 5360, 6989, 8782, 10428, 14390, 15742, 17770, 21734},
|
|
| 95 |
- { 2727, 3384, 6613, 9254, 10542, 12236, 14651, 15687, 20074, 21102},
|
|
| 96 |
- { 1916, 2953, 6274, 8088, 9710, 10925, 12392, 16434, 20010, 21183},
|
|
| 97 |
- { 3384, 4366, 5349, 7667, 11180, 12605, 13921, 15324, 19901, 20754},
|
|
| 98 |
- { 3075, 4283, 5951, 7619, 9604, 11010, 12384, 14006, 20658, 21497},
|
|
| 99 |
- { 1751, 2455, 5147, 9966, 11621, 13176, 14739, 16470, 20788, 21756},
|
|
| 100 |
- { 1442, 2188, 3330, 6813, 8929, 12135, 14476, 15306, 19635, 20544},
|
|
| 101 |
- { 2294, 2895, 4070, 8035, 12233, 13416, 14762, 17367, 18952, 19688},
|
|
| 102 |
- { 1937, 2659, 4602, 6697, 9071, 12863, 14197, 15230, 16047, 18877},
|
|
| 103 |
- { 2071, 2663, 4216, 9445, 10887, 12292, 13949, 14909, 19236, 20341},
|
|
| 104 |
- { 1740, 2491, 3488, 8138, 9656, 11153, 13206, 14688, 20896, 21907},
|
|
| 105 |
- { 2199, 2881, 4675, 8527, 10051, 11408, 14435, 15463, 17190, 20597},
|
|
| 106 |
- { 1943, 2988, 4177, 6039, 7478, 8536, 14181, 15551, 17622, 21579},
|
|
| 107 |
- { 1825, 3175, 7062, 9818, 12824, 15450, 18330, 19856, 21830, 22412},
|
|
| 108 |
- { 2464, 3046, 4822, 5977, 7696, 15398, 16730, 17646, 20588, 21320},
|
|
| 109 |
- { 2550, 3393, 5305, 6920, 10235, 14083, 18143, 19195, 20681, 21336},
|
|
| 110 |
- { 3003, 3799, 5321, 6437, 7919, 11643, 15810, 16846, 18119, 18980},
|
|
| 111 |
- { 3455, 4157, 6838, 8199, 9877, 12314, 15905, 16826, 19949, 20892},
|
|
| 112 |
- { 3052, 3769, 4891, 5810, 6977, 10126, 14788, 15990, 19773, 20904},
|
|
| 113 |
- { 3671, 4356, 5827, 6997, 8460, 12084, 14154, 14939, 19247, 20423},
|
|
| 114 |
- { 2716, 3684, 5246, 6686, 8463, 10001, 12394, 14131, 16150, 19776},
|
|
| 115 |
- { 1945, 2638, 4130, 7995, 14338, 15576, 17057, 18206, 20225, 20997},
|
|
| 116 |
- { 2304, 2928, 4122, 4824, 5640, 13139, 15825, 16938, 20108, 21054},
|
|
| 117 |
- { 1800, 2516, 3350, 5219, 13406, 15948, 17618, 18540, 20531, 21252},
|
|
| 118 |
- { 1436, 2224, 2753, 4546, 9657, 11245, 15177, 16317, 17489, 19135},
|
|
| 119 |
- { 2319, 2899, 4980, 6936, 8404, 13489, 15554, 16281, 20270, 20911},
|
|
| 120 |
- { 2187, 2919, 4610, 5875, 7390, 12556, 14033, 16794, 20998, 21769},
|
|
| 121 |
- { 2235, 2923, 5121, 6259, 8099, 13589, 15340, 16340, 17927, 20159},
|
|
| 122 |
- { 1765, 2638, 3751, 5730, 7883, 10108, 13633, 15419, 16808, 18574},
|
|
| 123 |
- { 3460, 5741, 9596, 11742, 14413, 16080, 18173, 19090, 20845, 21601},
|
|
| 124 |
- { 3735, 4426, 6199, 7363, 9250, 14489, 16035, 17026, 19873, 20876},
|
|
| 125 |
- { 3521, 4778, 6887, 8680, 12717, 14322, 15950, 18050, 20166, 21145},
|
|
| 126 |
- { 2141, 2968, 6865, 8051, 10010, 13159, 14813, 15861, 17528, 18655},
|
|
| 127 |
- { 4148, 6128, 9028, 10871, 12686, 14005, 15976, 17208, 19587, 20595},
|
|
| 128 |
- { 4403, 5367, 6634, 8371, 10163, 11599, 14963, 16331, 17982, 18768},
|
|
| 129 |
- { 4091, 5386, 6852, 8770, 11563, 13290, 15728, 16930, 19056, 20102},
|
|
| 130 |
- { 2746, 3625, 5299, 7504, 10262, 11432, 13172, 15490, 16875, 17514},
|
|
| 131 |
- { 2248, 3556, 8539, 10590, 12665, 14696, 16515, 17824, 20268, 21247},
|
|
| 132 |
- { 1279, 1960, 3920, 7793, 10153, 14753, 16646, 18139, 20679, 21466},
|
|
| 133 |
- { 2440, 3475, 6737, 8654, 12190, 14588, 17119, 17925, 19110, 19979},
|
|
| 134 |
- { 1879, 2514, 4497, 7572, 10017, 14948, 16141, 16897, 18397, 19376},
|
|
| 135 |
- { 2804, 3688, 7490, 10086, 11218, 12711, 16307, 17470, 20077, 21126},
|
|
| 136 |
- { 2023, 2682, 3873, 8268, 10255, 11645, 15187, 17102, 18965, 19788},
|
|
| 137 |
- { 2823, 3605, 5815, 8595, 10085, 11469, 16568, 17462, 18754, 19876},
|
|
| 138 |
- { 2851, 3681, 5280, 7648, 9173, 10338, 14961, 16148, 17559, 18474},
|
|
| 139 |
- { 1348, 2645, 5826, 8785, 10620, 12831, 16255, 18319, 21133, 22586},
|
|
| 140 |
- { 2141, 3036, 4293, 6082, 7593, 10629, 17158, 18033, 21466, 22084},
|
|
| 141 |
- { 1608, 2375, 3384, 6878, 9970, 11227, 16928, 17650, 20185, 21120},
|
|
| 142 |
- { 2774, 3616, 5014, 6557, 7788, 8959, 17068, 18302, 19537, 20542},
|
|
| 143 |
- { 1934, 4813, 6204, 7212, 8979, 11665, 15989, 17811, 20426, 21703},
|
|
| 144 |
- { 2288, 3507, 5037, 6841, 8278, 9638, 15066, 16481, 21653, 22214},
|
|
| 145 |
- { 2951, 3771, 4878, 7578, 9016, 10298, 14490, 15242, 20223, 20990},
|
|
| 146 |
- { 3256, 4791, 6601, 7521, 8644, 9707, 13398, 16078, 19102, 20249},
|
|
| 147 |
- { 1827, 2614, 3486, 6039, 12149, 13823, 16191, 17282, 21423, 22041},
|
|
| 148 |
- { 1000, 1704, 3002, 6335, 8471, 10500, 14878, 16979, 20026, 22427},
|
|
| 149 |
- { 1646, 2286, 3109, 7245, 11493, 12791, 16824, 17667, 18981, 20222},
|
|
| 150 |
- { 1708, 2501, 3315, 6737, 8729, 9924, 16089, 17097, 18374, 19917},
|
|
| 151 |
- { 2623, 3510, 4478, 5645, 9862, 11115, 15219, 18067, 19583, 20382},
|
|
| 152 |
- { 2518, 3434, 4728, 6388, 8082, 9285, 13162, 18383, 19819, 20552},
|
|
| 153 |
- { 1726, 2383, 4090, 6303, 7805, 12845, 14612, 17608, 19269, 20181},
|
|
| 154 |
- { 2860, 3735, 4838, 6044, 7254, 8402, 14031, 16381, 18037, 19410},
|
|
| 155 |
- { 4247, 5993, 7952, 9792, 12342, 14653, 17527, 18774, 20831, 21699},
|
|
| 156 |
- { 3502, 4051, 5680, 6805, 8146, 11945, 16649, 17444, 20390, 21564},
|
|
| 157 |
- { 3151, 4893, 5899, 7198, 11418, 13073, 15124, 17673, 20520, 21861},
|
|
| 158 |
- { 3960, 4848, 5926, 7259, 8811, 10529, 15661, 16560, 18196, 20183},
|
|
| 159 |
- { 4499, 6604, 8036, 9251, 10804, 12627, 15880, 17512, 20020, 21046},
|
|
| 160 |
- { 4251, 5541, 6654, 8318, 9900, 11686, 15100, 17093, 20572, 21687},
|
|
| 161 |
- { 3769, 5327, 7865, 9360, 10684, 11818, 13660, 15366, 18733, 19882},
|
|
| 162 |
- { 3083, 3969, 6248, 8121, 9798, 10994, 12393, 13686, 17888, 19105},
|
|
| 163 |
- { 2731, 4670, 7063, 9201, 11346, 13735, 16875, 18797, 20787, 22360},
|
|
| 164 |
- { 1187, 2227, 4737, 7214, 9622, 12633, 15404, 17968, 20262, 23533},
|
|
| 165 |
- { 1911, 2477, 3915, 10098, 11616, 12955, 16223, 17138, 19270, 20729},
|
|
| 166 |
- { 1764, 2519, 3887, 6944, 9150, 12590, 16258, 16984, 17924, 18435},
|
|
| 167 |
- { 1400, 3674, 7131, 8718, 10688, 12508, 15708, 17711, 19720, 21068},
|
|
| 168 |
- { 2322, 3073, 4287, 8108, 9407, 10628, 15862, 16693, 19714, 21474},
|
|
| 169 |
- { 2630, 3339, 4758, 8360, 10274, 11333, 12880, 17374, 19221, 19936},
|
|
| 170 |
- { 1721, 2577, 5553, 7195, 8651, 10686, 15069, 16953, 18703, 19929}
|
|
| 171 |
-}; |
|
| 172 |
- |
|
| 173 |
-/** |
|
| 174 |
- * second stage LSP codebook, high and low parts |
|
| 175 |
- (both 5-dimensional, with 32 entries (3.2.4 of G.729) |
|
| 176 |
- */ |
|
| 177 |
-static const int16_t cb_lsp_2nd[1<<VQ_2ND_BITS][10] = { /* (2.13) */
|
|
| 178 |
- { -435, -815, -742, 1033, -518, 582, -1201, 829, 86, 385},
|
|
| 179 |
- { -833, -891, 463, -8, -1251, 1450, 72, -231, 864, 661},
|
|
| 180 |
- {-1021, 231, -306, 321, -220, -163, -526, -754, -1633, 267},
|
|
| 181 |
- { 57, -198, -339, -33, -1468, 573, 796, -169, -631, 816},
|
|
| 182 |
- { 171, -350, 294, 1660, 453, 519, 291, 159, -640, -1296},
|
|
| 183 |
- { -701, -842, -58, 950, 892, 1549, 715, 527, -714, -193},
|
|
| 184 |
- { 584, 31, -289, 356, -333, -457, 612, -283, -1381, -741},
|
|
| 185 |
- { -109, -808, 231, 77, -87, -344, 1341, 1087, -654, -569},
|
|
| 186 |
- { -859, 1236, 550, 854, 714, -543, -1752, -195, -98, -276},
|
|
| 187 |
- { -877, -954, -1248, -299, 212, -235, -728, 949, 1517, 895},
|
|
| 188 |
- { -77, 344, -620, 763, 413, 502, -362, -960, -483, 1386},
|
|
| 189 |
- { -314, -307, -256, -1260, -429, 450, -466, -108, 1010, 2223},
|
|
| 190 |
- { 711, 693, 521, 650, 1305, -28, -378, 744, -1005, 240},
|
|
| 191 |
- { -112, -271, -500, 946, 1733, 271, -15, 909, -259, 1688},
|
|
| 192 |
- { 575, -10, -468, -199, 1101, -1011, 581, -53, -747, 878},
|
|
| 193 |
- { 145, -285, -1280, -398, 36, -498, -1377, 18, -444, 1483},
|
|
| 194 |
- {-1133, -835, 1350, 1284, -95, 1015, -222, 443, 372, -354},
|
|
| 195 |
- {-1459, -1237, 416, -213, 466, 669, 659, 1640, 932, 534},
|
|
| 196 |
- { -15, 66, 468, 1019, -748, 1385, -182, -907, -721, -262},
|
|
| 197 |
- { -338, 148, 1445, 75, -760, 569, 1247, 337, 416, -121},
|
|
| 198 |
- { 389, 239, 1568, 981, 113, 369, -1003, -507, -587, -904},
|
|
| 199 |
- { -312, -98, 949, 31, 1104, 72, -141, 1465, 63, -785},
|
|
| 200 |
- { 1127, 584, 835, 277, -1159, 208, 301, -882, 117, -404},
|
|
| 201 |
- { 539, -114, 856, -493, 223, -912, 623, -76, 276, -440},
|
|
| 202 |
- { 2197, 2337, 1268, 670, 304, -267, -525, 140, 882, -139},
|
|
| 203 |
- {-1596, 550, 801, -456, -56, -697, 865, 1060, 413, 446},
|
|
| 204 |
- { 1154, 593, -77, 1237, -31, 581, -1037, -895, 669, 297},
|
|
| 205 |
- { 397, 558, 203, -797, -919, 3, 692, -292, 1050, 782},
|
|
| 206 |
- { 334, 1475, 632, -80, 48, -1061, -484, 362, -597, -852},
|
|
| 207 |
- { -545, -330, -429, -680, 1133, -1182, -744, 1340, 262, 63},
|
|
| 208 |
- { 1320, 827, -398, -576, 341, -774, -483, -1247, -70, 98},
|
|
| 209 |
- { -163, 674, -11, -886, 531, -1125, -265, -242, 724, 934}
|
|
| 210 |
-}; |
|
| 211 |
- |
|
| 212 |
-/** |
|
| 213 |
- * gain codebook (first stage), 8k mode (3.9.2 of G.729) |
|
| 214 |
- */ |
|
| 215 |
-static const int16_t cb_gain_1st_8k[1<<GC_1ST_IDX_BITS_8K][2] = { /*(0.14) (2.13) */
|
|
| 216 |
- { 3242 , 9949 },
|
|
| 217 |
- { 1551 , 2425 },
|
|
| 218 |
- { 2678 , 27162 },
|
|
| 219 |
- { 1921 , 9291 },
|
|
| 220 |
- { 1831 , 5022 },
|
|
| 221 |
- { 1 , 1516 },
|
|
| 222 |
- { 356 , 14756 },
|
|
| 223 |
- { 57 , 5404 },
|
|
| 224 |
-}; |
|
| 225 |
- |
|
| 226 |
-/** |
|
| 227 |
- * gain codebook (second stage), 8k mode (3.9.2 of G.729) |
|
| 228 |
- */ |
|
| 229 |
-static const int16_t cb_gain_2nd_8k[1<<GC_2ND_IDX_BITS_8K][2] = { /*(1.14) (1.13) */
|
|
| 230 |
- { 5142 , 592 },
|
|
| 231 |
- { 17299 , 1861 },
|
|
| 232 |
- { 6160 , 2395 },
|
|
| 233 |
- { 16112 , 3392 },
|
|
| 234 |
- { 826 , 2005 },
|
|
| 235 |
- { 18973 , 5935 },
|
|
| 236 |
- { 1994 , 0 },
|
|
| 237 |
- { 15434 , 237 },
|
|
| 238 |
- { 10573 , 2966 },
|
|
| 239 |
- { 15132 , 4914 },
|
|
| 240 |
- { 11569 , 1196 },
|
|
| 241 |
- { 14194 , 1630 },
|
|
| 242 |
- { 8091 , 4861 },
|
|
| 243 |
- { 15161 , 14276 },
|
|
| 244 |
- { 9120 , 525 },
|
|
| 245 |
- { 13260 , 3256 },
|
|
| 246 |
-}; |
|
| 247 |
- |
|
| 248 |
-/** |
|
| 249 |
- * 4th order Moving Average (MA) Predictor codebook (3.2.4 of G.729) |
|
| 250 |
- */ |
|
| 251 |
-static const int16_t cb_ma_predictor[2][MA_NP][10] = { /* (0.15) */
|
|
| 252 |
- {
|
|
| 253 |
- { 8421, 9109, 9175, 8965, 9034, 9057, 8765, 8775, 9106, 8673},
|
|
| 254 |
- { 7018, 7189, 7638, 7307, 7444, 7379, 7038, 6956, 6930, 6868},
|
|
| 255 |
- { 5472, 4990, 5134, 5177, 5246, 5141, 5206, 5095, 4830, 5147},
|
|
| 256 |
- { 4056, 3031, 2614, 3024, 2916, 2713, 3309, 3237, 2857, 3473}
|
|
| 257 |
- }, |
|
| 258 |
- {
|
|
| 259 |
- { 7733, 7880, 8188, 8175, 8247, 8490, 8637, 8601, 8359, 7569},
|
|
| 260 |
- { 4210, 3031, 2552, 3473, 3876, 3853, 4184, 4154, 3909, 3968},
|
|
| 261 |
- { 3214, 1930, 1313, 2143, 2493, 2385, 2755, 2706, 2542, 2919},
|
|
| 262 |
- { 3024, 1592, 940, 1631, 1723, 1579, 2034, 2084, 1913, 2601}
|
|
| 263 |
- } |
|
| 264 |
-}; |
|
| 265 |
- |
|
| 266 |
-static const int16_t cb_ma_predictor_sum[2][10] = { /* (0.15) */
|
|
| 267 |
- { 7798, 8447, 8205, 8293, 8126, 8477, 8447, 8703, 9043, 8604},
|
|
| 268 |
- {14585, 18333, 19772, 17344, 16426, 16459, 15155, 15220, 16043, 15708}
|
|
| 269 |
-}; |
|
| 270 |
- |
|
| 271 |
-/** |
|
| 272 |
- * initial LSP coefficients belongs to virtual frame preceding the |
|
| 273 |
- * first frame of the stream |
|
| 274 |
- */ |
|
| 275 |
-static const int16_t lsp_init[10]= { /* (0.15) */
|
|
| 276 |
- 30000, 26000, 21000, 15000, 8000, 0, -8000,-15000,-21000,-26000 |
|
| 277 |
-}; |
|
| 278 |
-#endif /* AVCODEC_G729DATA_H */ |
| 279 | 1 |
deleted file mode 100644 |
| ... | ... |
@@ -1,331 +0,0 @@ |
| 1 |
-/* |
|
| 2 |
- * G.729 decoder |
|
| 3 |
- * Copyright (c) 2008 Vladimir Voroshilov |
|
| 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 <stdlib.h> |
|
| 22 |
-#include <inttypes.h> |
|
| 23 |
-#include <limits.h> |
|
| 24 |
-#include <stdio.h> |
|
| 25 |
-#include <string.h> |
|
| 26 |
-#include <math.h> |
|
| 27 |
-#include <assert.h> |
|
| 28 |
- |
|
| 29 |
-#include "avcodec.h" |
|
| 30 |
-#include "libavutil/avutil.h" |
|
| 31 |
-#include "get_bits.h" |
|
| 32 |
- |
|
| 33 |
-#include "lsp.h" |
|
| 34 |
-#include "celp_math.h" |
|
| 35 |
-#include "acelp_filters.h" |
|
| 36 |
-#include "acelp_pitch_delay.h" |
|
| 37 |
-#include "acelp_vectors.h" |
|
| 38 |
-#include "g729data.h" |
|
| 39 |
- |
|
| 40 |
-/** |
|
| 41 |
- * minimum quantized LSF value (3.2.4) |
|
| 42 |
- * 0.005 in Q13 |
|
| 43 |
- */ |
|
| 44 |
-#define LSFQ_MIN 40 |
|
| 45 |
- |
|
| 46 |
-/** |
|
| 47 |
- * maximum quantized LSF value (3.2.4) |
|
| 48 |
- * 3.135 in Q13 |
|
| 49 |
- */ |
|
| 50 |
-#define LSFQ_MAX 25681 |
|
| 51 |
- |
|
| 52 |
-/** |
|
| 53 |
- * minimum LSF distance (3.2.4) |
|
| 54 |
- * 0.0391 in Q13 |
|
| 55 |
- */ |
|
| 56 |
-#define LSFQ_DIFF_MIN 321 |
|
| 57 |
- |
|
| 58 |
-/** |
|
| 59 |
- * minimum gain pitch value (3.8, Equation 47) |
|
| 60 |
- * 0.2 in (1.14) |
|
| 61 |
- */ |
|
| 62 |
-#define SHARP_MIN 3277 |
|
| 63 |
- |
|
| 64 |
-/** |
|
| 65 |
- * maximum gain pitch value (3.8, Equation 47) |
|
| 66 |
- * (EE) This does not comply with the specification. |
|
| 67 |
- * Specification says about 0.8, which should be |
|
| 68 |
- * 13107 in (1.14), but reference C code uses |
|
| 69 |
- * 13017 (equals to 0.7945) instead of it. |
|
| 70 |
- */ |
|
| 71 |
-#define SHARP_MAX 13017 |
|
| 72 |
- |
|
| 73 |
-/** |
|
| 74 |
- * subframe size |
|
| 75 |
- */ |
|
| 76 |
-#define SUBFRAME_SIZE 40 |
|
| 77 |
- |
|
| 78 |
- |
|
| 79 |
-typedef struct {
|
|
| 80 |
- uint8_t ac_index_bits[2]; ///< adaptive codebook index for second subframe (size in bits) |
|
| 81 |
- uint8_t parity_bit; ///< parity bit for pitch delay |
|
| 82 |
- uint8_t gc_1st_index_bits; ///< gain codebook (first stage) index (size in bits) |
|
| 83 |
- uint8_t gc_2nd_index_bits; ///< gain codebook (second stage) index (size in bits) |
|
| 84 |
- uint8_t fc_signs_bits; ///< number of pulses in fixed-codebook vector |
|
| 85 |
- uint8_t fc_indexes_bits; ///< size (in bits) of fixed-codebook index entry |
|
| 86 |
-} G729FormatDescription; |
|
| 87 |
- |
|
| 88 |
-typedef struct {
|
|
| 89 |
- int pitch_delay_int_prev; ///< integer part of previous subframe's pitch delay (4.1.3) |
|
| 90 |
- |
|
| 91 |
- /// (2.13) LSP quantizer outputs |
|
| 92 |
- int16_t past_quantizer_output_buf[MA_NP + 1][10]; |
|
| 93 |
- int16_t* past_quantizer_outputs[MA_NP + 1]; |
|
| 94 |
- |
|
| 95 |
- int16_t lsfq[10]; ///< (2.13) quantized LSF coefficients from previous frame |
|
| 96 |
- int16_t lsp_buf[2][10]; ///< (0.15) LSP coefficients (previous and current frames) (3.2.5) |
|
| 97 |
- int16_t *lsp[2]; ///< pointers to lsp_buf |
|
| 98 |
-} G729Context; |
|
| 99 |
- |
|
| 100 |
-static const G729FormatDescription format_g729_8k = {
|
|
| 101 |
- .ac_index_bits = {8,5},
|
|
| 102 |
- .parity_bit = 1, |
|
| 103 |
- .gc_1st_index_bits = GC_1ST_IDX_BITS_8K, |
|
| 104 |
- .gc_2nd_index_bits = GC_2ND_IDX_BITS_8K, |
|
| 105 |
- .fc_signs_bits = 4, |
|
| 106 |
- .fc_indexes_bits = 13, |
|
| 107 |
-}; |
|
| 108 |
- |
|
| 109 |
-static const G729FormatDescription format_g729d_6k4 = {
|
|
| 110 |
- .ac_index_bits = {8,4},
|
|
| 111 |
- .parity_bit = 0, |
|
| 112 |
- .gc_1st_index_bits = GC_1ST_IDX_BITS_6K4, |
|
| 113 |
- .gc_2nd_index_bits = GC_2ND_IDX_BITS_6K4, |
|
| 114 |
- .fc_signs_bits = 2, |
|
| 115 |
- .fc_indexes_bits = 9, |
|
| 116 |
-}; |
|
| 117 |
- |
|
| 118 |
-/** |
|
| 119 |
- * @brief pseudo random number generator |
|
| 120 |
- */ |
|
| 121 |
-static inline uint16_t g729_prng(uint16_t value) |
|
| 122 |
-{
|
|
| 123 |
- return 31821 * value + 13849; |
|
| 124 |
-} |
|
| 125 |
- |
|
| 126 |
-/** |
|
| 127 |
- * Get parity bit of bit 2..7 |
|
| 128 |
- */ |
|
| 129 |
-static inline int get_parity(uint8_t value) |
|
| 130 |
-{
|
|
| 131 |
- return (0x6996966996696996ULL >> (value >> 2)) & 1; |
|
| 132 |
-} |
|
| 133 |
- |
|
| 134 |
-static void lsf_decode(int16_t* lsfq, int16_t* past_quantizer_outputs[MA_NP + 1], |
|
| 135 |
- int16_t ma_predictor, |
|
| 136 |
- int16_t vq_1st, int16_t vq_2nd_low, int16_t vq_2nd_high) |
|
| 137 |
-{
|
|
| 138 |
- int i,j; |
|
| 139 |
- static const uint8_t min_distance[2]={10, 5}; //(2.13)
|
|
| 140 |
- int16_t* quantizer_output = past_quantizer_outputs[MA_NP]; |
|
| 141 |
- |
|
| 142 |
- for (i = 0; i < 5; i++) {
|
|
| 143 |
- quantizer_output[i] = cb_lsp_1st[vq_1st][i ] + cb_lsp_2nd[vq_2nd_low ][i ]; |
|
| 144 |
- quantizer_output[i + 5] = cb_lsp_1st[vq_1st][i + 5] + cb_lsp_2nd[vq_2nd_high][i + 5]; |
|
| 145 |
- } |
|
| 146 |
- |
|
| 147 |
- for (j = 0; j < 2; j++) {
|
|
| 148 |
- for (i = 1; i < 10; i++) {
|
|
| 149 |
- int diff = (quantizer_output[i - 1] - quantizer_output[i] + min_distance[j]) >> 1; |
|
| 150 |
- if (diff > 0) {
|
|
| 151 |
- quantizer_output[i - 1] -= diff; |
|
| 152 |
- quantizer_output[i ] += diff; |
|
| 153 |
- } |
|
| 154 |
- } |
|
| 155 |
- } |
|
| 156 |
- |
|
| 157 |
- for (i = 0; i < 10; i++) {
|
|
| 158 |
- int sum = quantizer_output[i] * cb_ma_predictor_sum[ma_predictor][i]; |
|
| 159 |
- for (j = 0; j < MA_NP; j++) |
|
| 160 |
- sum += past_quantizer_outputs[j][i] * cb_ma_predictor[ma_predictor][j][i]; |
|
| 161 |
- |
|
| 162 |
- lsfq[i] = sum >> 15; |
|
| 163 |
- } |
|
| 164 |
- |
|
| 165 |
- /* Rotate past_quantizer_outputs. */ |
|
| 166 |
- memmove(past_quantizer_outputs + 1, past_quantizer_outputs, MA_NP * sizeof(int16_t*)); |
|
| 167 |
- past_quantizer_outputs[0] = quantizer_output; |
|
| 168 |
- |
|
| 169 |
- ff_acelp_reorder_lsf(lsfq, LSFQ_DIFF_MIN, LSFQ_MIN, LSFQ_MAX, 10); |
|
| 170 |
-} |
|
| 171 |
- |
|
| 172 |
-static av_cold int decoder_init(AVCodecContext * avctx) |
|
| 173 |
-{
|
|
| 174 |
- G729Context* ctx = avctx->priv_data; |
|
| 175 |
- int i,k; |
|
| 176 |
- |
|
| 177 |
- if (avctx->channels != 1) {
|
|
| 178 |
- av_log(avctx, AV_LOG_ERROR, "Only mono sound is supported (requested channels: %d).\n", avctx->channels); |
|
| 179 |
- return AVERROR(EINVAL); |
|
| 180 |
- } |
|
| 181 |
- |
|
| 182 |
- /* Both 8kbit/s and 6.4kbit/s modes uses two subframes per frame. */ |
|
| 183 |
- avctx->frame_size = SUBFRAME_SIZE << 1; |
|
| 184 |
- |
|
| 185 |
- for (k = 0; k < MA_NP + 1; k++) {
|
|
| 186 |
- ctx->past_quantizer_outputs[k] = ctx->past_quantizer_output_buf[k]; |
|
| 187 |
- for (i = 1; i < 11; i++) |
|
| 188 |
- ctx->past_quantizer_outputs[k][i - 1] = (18717 * i) >> 3; |
|
| 189 |
- } |
|
| 190 |
- |
|
| 191 |
- ctx->lsp[0] = ctx->lsp_buf[0]; |
|
| 192 |
- ctx->lsp[1] = ctx->lsp_buf[1]; |
|
| 193 |
- memcpy(ctx->lsp[0], lsp_init, 10 * sizeof(int16_t)); |
|
| 194 |
- |
|
| 195 |
- return 0; |
|
| 196 |
-} |
|
| 197 |
- |
|
| 198 |
-static int decode_frame(AVCodecContext *avctx, void *data, int *data_size, |
|
| 199 |
- AVPacket *avpkt) |
|
| 200 |
-{
|
|
| 201 |
- const uint8_t *buf = avpkt->data; |
|
| 202 |
- int buf_size = avpkt->size; |
|
| 203 |
- int16_t *out_frame = data; |
|
| 204 |
- GetBitContext gb; |
|
| 205 |
- G729FormatDescription format; |
|
| 206 |
- int frame_erasure = 0; ///< frame erasure detected during decoding |
|
| 207 |
- int bad_pitch = 0; ///< parity check failed |
|
| 208 |
- int i; |
|
| 209 |
- G729Context *ctx = avctx->priv_data; |
|
| 210 |
- int16_t lp[2][11]; // (3.12) |
|
| 211 |
- uint8_t ma_predictor; ///< switched MA predictor of LSP quantizer |
|
| 212 |
- uint8_t quantizer_1st; ///< first stage vector of quantizer |
|
| 213 |
- uint8_t quantizer_2nd_lo; ///< second stage lower vector of quantizer (size in bits) |
|
| 214 |
- uint8_t quantizer_2nd_hi; ///< second stage higher vector of quantizer (size in bits) |
|
| 215 |
- |
|
| 216 |
- int pitch_delay_int; // pitch delay, integer part |
|
| 217 |
- int pitch_delay_3x; // pitch delay, multiplied by 3 |
|
| 218 |
- |
|
| 219 |
- if (*data_size < SUBFRAME_SIZE << 2) {
|
|
| 220 |
- av_log(avctx, AV_LOG_ERROR, "Error processing packet: output buffer too small\n"); |
|
| 221 |
- return AVERROR(EIO); |
|
| 222 |
- } |
|
| 223 |
- |
|
| 224 |
- if (buf_size == 10) {
|
|
| 225 |
- format = format_g729_8k; |
|
| 226 |
- av_log(avctx, AV_LOG_DEBUG, "Packet type: %s\n", "G.729 @ 8kbit/s"); |
|
| 227 |
- } else if (buf_size == 8) {
|
|
| 228 |
- format = format_g729d_6k4; |
|
| 229 |
- av_log(avctx, AV_LOG_DEBUG, "Packet type: %s\n", "G.729D @ 6.4kbit/s"); |
|
| 230 |
- } else {
|
|
| 231 |
- av_log(avctx, AV_LOG_ERROR, "Packet size %d is unknown.\n", buf_size); |
|
| 232 |
- return AVERROR_INVALIDDATA; |
|
| 233 |
- } |
|
| 234 |
- |
|
| 235 |
- for (i=0; i < buf_size; i++) |
|
| 236 |
- frame_erasure |= buf[i]; |
|
| 237 |
- frame_erasure = !frame_erasure; |
|
| 238 |
- |
|
| 239 |
- init_get_bits(&gb, buf, buf_size); |
|
| 240 |
- |
|
| 241 |
- ma_predictor = get_bits(&gb, 1); |
|
| 242 |
- quantizer_1st = get_bits(&gb, VQ_1ST_BITS); |
|
| 243 |
- quantizer_2nd_lo = get_bits(&gb, VQ_2ND_BITS); |
|
| 244 |
- quantizer_2nd_hi = get_bits(&gb, VQ_2ND_BITS); |
|
| 245 |
- |
|
| 246 |
- lsf_decode(ctx->lsfq, ctx->past_quantizer_outputs, |
|
| 247 |
- ma_predictor, |
|
| 248 |
- quantizer_1st, quantizer_2nd_lo, quantizer_2nd_hi); |
|
| 249 |
- |
|
| 250 |
- ff_acelp_lsf2lsp(ctx->lsp[1], ctx->lsfq, 10); |
|
| 251 |
- |
|
| 252 |
- ff_acelp_lp_decode(&lp[0][0], &lp[1][0], ctx->lsp[1], ctx->lsp[0], 10); |
|
| 253 |
- |
|
| 254 |
- FFSWAP(int16_t*, ctx->lsp[1], ctx->lsp[0]); |
|
| 255 |
- |
|
| 256 |
- for (i = 0; i < 2; i++) {
|
|
| 257 |
- uint8_t ac_index; ///< adaptive codebook index |
|
| 258 |
- uint8_t pulses_signs; ///< fixed-codebook vector pulse signs |
|
| 259 |
- int fc_indexes; ///< fixed-codebook indexes |
|
| 260 |
- uint8_t gc_1st_index; ///< gain codebook (first stage) index |
|
| 261 |
- uint8_t gc_2nd_index; ///< gain codebook (second stage) index |
|
| 262 |
- |
|
| 263 |
- ac_index = get_bits(&gb, format.ac_index_bits[i]); |
|
| 264 |
- if(!i && format.parity_bit) |
|
| 265 |
- bad_pitch = get_parity(ac_index) == get_bits1(&gb); |
|
| 266 |
- fc_indexes = get_bits(&gb, format.fc_indexes_bits); |
|
| 267 |
- pulses_signs = get_bits(&gb, format.fc_signs_bits); |
|
| 268 |
- gc_1st_index = get_bits(&gb, format.gc_1st_index_bits); |
|
| 269 |
- gc_2nd_index = get_bits(&gb, format.gc_2nd_index_bits); |
|
| 270 |
- |
|
| 271 |
- if(!i) {
|
|
| 272 |
- if (bad_pitch) |
|
| 273 |
- pitch_delay_3x = 3 * ctx->pitch_delay_int_prev; |
|
| 274 |
- else |
|
| 275 |
- pitch_delay_3x = ff_acelp_decode_8bit_to_1st_delay3(ac_index); |
|
| 276 |
- } else {
|
|
| 277 |
- int pitch_delay_min = av_clip(ctx->pitch_delay_int_prev - 5, |
|
| 278 |
- PITCH_DELAY_MIN, PITCH_DELAY_MAX - 9); |
|
| 279 |
- |
|
| 280 |
- if(packet_type == FORMAT_G729D_6K4) |
|
| 281 |
- pitch_delay_3x = ff_acelp_decode_4bit_to_2nd_delay3(ac_index, pitch_delay_min); |
|
| 282 |
- else |
|
| 283 |
- pitch_delay_3x = ff_acelp_decode_5_6_bit_to_2nd_delay3(ac_index, pitch_delay_min); |
|
| 284 |
- } |
|
| 285 |
- |
|
| 286 |
- /* Round pitch delay to nearest (used everywhere except ff_acelp_interpolate). */ |
|
| 287 |
- pitch_delay_int = (pitch_delay_3x + 1) / 3; |
|
| 288 |
- |
|
| 289 |
- ff_acelp_weighted_vector_sum(fc + pitch_delay_int, |
|
| 290 |
- fc + pitch_delay_int, |
|
| 291 |
- fc, 1 << 14, |
|
| 292 |
- av_clip(ctx->gain_pitch, SHARP_MIN, SHARP_MAX), |
|
| 293 |
- 0, 14, |
|
| 294 |
- SUBFRAME_SIZE - pitch_delay_int); |
|
| 295 |
- |
|
| 296 |
- if (frame_erasure) {
|
|
| 297 |
- ctx->gain_pitch = (29491 * ctx->gain_pitch) >> 15; // 0.90 (0.15) |
|
| 298 |
- ctx->gain_code = ( 2007 * ctx->gain_code ) >> 11; // 0.98 (0.11) |
|
| 299 |
- |
|
| 300 |
- gain_corr_factor = 0; |
|
| 301 |
- } else {
|
|
| 302 |
- ctx->gain_pitch = cb_gain_1st_8k[gc_1st_index][0] + |
|
| 303 |
- cb_gain_2nd_8k[gc_2nd_index][0]; |
|
| 304 |
- gain_corr_factor = cb_gain_1st_8k[gc_1st_index][1] + |
|
| 305 |
- cb_gain_2nd_8k[gc_2nd_index][1]; |
|
| 306 |
- |
|
| 307 |
- ff_acelp_weighted_vector_sum(ctx->exc + i * SUBFRAME_SIZE, |
|
| 308 |
- ctx->exc + i * SUBFRAME_SIZE, fc, |
|
| 309 |
- (!voicing && frame_erasure) ? 0 : ctx->gain_pitch, |
|
| 310 |
- ( voicing && frame_erasure) ? 0 : ctx->gain_code, |
|
| 311 |
- 1 << 13, 14, SUBFRAME_SIZE); |
|
| 312 |
- |
|
| 313 |
- ctx->pitch_delay_int_prev = pitch_delay_int; |
|
| 314 |
- } |
|
| 315 |
- |
|
| 316 |
- *data_size = SUBFRAME_SIZE << 2; |
|
| 317 |
- return buf_size; |
|
| 318 |
-} |
|
| 319 |
- |
|
| 320 |
-AVCodec ff_g729_decoder = |
|
| 321 |
-{
|
|
| 322 |
- "g729", |
|
| 323 |
- AVMEDIA_TYPE_AUDIO, |
|
| 324 |
- CODEC_ID_G729, |
|
| 325 |
- sizeof(G729Context), |
|
| 326 |
- decoder_init, |
|
| 327 |
- NULL, |
|
| 328 |
- NULL, |
|
| 329 |
- decode_frame, |
|
| 330 |
- .long_name = NULL_IF_CONFIG_SMALL("G.729"),
|
|
| 331 |
-}; |
| ... | ... |
@@ -150,11 +150,7 @@ void ff_acelp_lp_decode(int16_t* lp_1st, int16_t* lp_2nd, const int16_t* lsp_2nd |
| 150 | 150 |
|
| 151 | 151 |
/* LSP values for first subframe (3.2.5 of G.729, Equation 24)*/ |
| 152 | 152 |
for(i=0; i<lp_order; i++) |
| 153 |
-#ifdef G729_BITEXACT |
|
| 154 |
- lsp_1st[i] = (lsp_2nd[i] >> 1) + (lsp_prev[i] >> 1); |
|
| 155 |
-#else |
|
| 156 | 153 |
lsp_1st[i] = (lsp_2nd[i] + lsp_prev[i]) >> 1; |
| 157 |
-#endif |
|
| 158 | 154 |
|
| 159 | 155 |
ff_acelp_lsp2lpc(lp_1st, lsp_1st, lp_order >> 1); |
| 160 | 156 |
|
| ... | ... |
@@ -318,7 +318,7 @@ int ff_alloc_picture(MpegEncContext *s, Picture *pic, int shared){
|
| 318 | 318 |
s->prev_pict_types[0]= s->dropable ? AV_PICTURE_TYPE_B : s->pict_type; |
| 319 | 319 |
if (pic->f.age < PREV_PICT_TYPES_BUFFER_SIZE && s->prev_pict_types[pic->f.age] == AV_PICTURE_TYPE_B) |
| 320 | 320 |
pic->f.age = INT_MAX; // Skipped MBs in B-frames are quite rare in MPEG-1/2 and it is a bit tricky to skip them anyway. |
| 321 |
- pic->owner2 = NULL; |
|
| 321 |
+ pic->owner2 = s; |
|
| 322 | 322 |
|
| 323 | 323 |
return 0; |
| 324 | 324 |
fail: //for the FF_ALLOCZ_OR_GOTO macro |
| ... | ... |
@@ -1036,14 +1036,16 @@ int MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx) |
| 1036 | 1036 |
/* mark&release old frames */ |
| 1037 | 1037 |
if (s->pict_type != AV_PICTURE_TYPE_B && s->last_picture_ptr && s->last_picture_ptr != s->next_picture_ptr && s->last_picture_ptr->f.data[0]) {
|
| 1038 | 1038 |
if(s->out_format != FMT_H264 || s->codec_id == CODEC_ID_SVQ3){
|
| 1039 |
- free_frame_buffer(s, s->last_picture_ptr); |
|
| 1039 |
+ if (s->last_picture_ptr->owner2 == s) |
|
| 1040 |
+ free_frame_buffer(s, s->last_picture_ptr); |
|
| 1040 | 1041 |
|
| 1041 | 1042 |
/* release forgotten pictures */ |
| 1042 | 1043 |
/* if(mpeg124/h263) */ |
| 1043 | 1044 |
if(!s->encoding){
|
| 1044 | 1045 |
for(i=0; i<s->picture_count; i++){
|
| 1045 |
- if (s->picture[i].f.data[0] && &s->picture[i] != s->next_picture_ptr && s->picture[i].f.reference) {
|
|
| 1046 |
- av_log(avctx, AV_LOG_ERROR, "releasing zombie picture\n"); |
|
| 1046 |
+ if (s->picture[i].owner2 == s && s->picture[i].f.data[0] && &s->picture[i] != s->next_picture_ptr && s->picture[i].f.reference) {
|
|
| 1047 |
+ if (!(avctx->active_thread_type & FF_THREAD_FRAME)) |
|
| 1048 |
+ av_log(avctx, AV_LOG_ERROR, "releasing zombie picture\n"); |
|
| 1047 | 1049 |
free_frame_buffer(s, &s->picture[i]); |
| 1048 | 1050 |
} |
| 1049 | 1051 |
} |
| ... | ... |
@@ -178,7 +178,7 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx) |
| 178 | 178 |
} |
| 179 | 179 |
#endif //CONFIG_ENCODERS |
| 180 | 180 |
|
| 181 |
- if (avctx->lowres==0) {
|
|
| 181 |
+ if (avctx->lowres == 0 && avctx->bits_per_raw_sample <= 8) {
|
|
| 182 | 182 |
if ((avctx->idct_algo == FF_IDCT_AUTO) || |
| 183 | 183 |
(avctx->idct_algo == FF_IDCT_ALTIVEC)) {
|
| 184 | 184 |
c->idct_put = idct_put_altivec; |
| ... | ... |
@@ -156,7 +156,8 @@ void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx) |
| 156 | 156 |
|
| 157 | 157 |
c->get_pixels = get_pixels_mmi; |
| 158 | 158 |
|
| 159 |
- if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2){
|
|
| 159 |
+ if (avctx->bits_per_raw_sample <= 8 && |
|
| 160 |
+ (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2)) {
|
|
| 160 | 161 |
c->idct_put= ff_mmi_idct_put; |
| 161 | 162 |
c->idct_add= ff_mmi_idct_add; |
| 162 | 163 |
c->idct = ff_mmi_idct; |
| ... | ... |
@@ -97,7 +97,8 @@ void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx) |
| 97 | 97 |
|
| 98 | 98 |
if (!high_bit_depth) |
| 99 | 99 |
c->clear_blocks = clear_blocks_sh4; |
| 100 |
- if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SH4){
|
|
| 100 |
+ if (avctx->bits_per_raw_sample <= 8 && |
|
| 101 |
+ (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SH4)) {
|
|
| 101 | 102 |
c->idct_put = idct_put; |
| 102 | 103 |
c->idct_add = idct_add; |
| 103 | 104 |
c->idct = idct_sh4; |
| ... | ... |
@@ -25,339 +25,19 @@ |
| 25 | 25 |
* simpleidct in C. |
| 26 | 26 |
*/ |
| 27 | 27 |
|
| 28 |
-/* |
|
| 29 |
- based upon some outcommented c code from mpeg2dec (idct_mmx.c |
|
| 30 |
- written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>) |
|
| 31 |
- */ |
|
| 32 |
- |
|
| 33 | 28 |
#include "libavutil/intreadwrite.h" |
| 34 | 29 |
#include "avcodec.h" |
| 35 | 30 |
#include "dsputil.h" |
| 36 | 31 |
#include "mathops.h" |
| 37 | 32 |
#include "simple_idct.h" |
| 38 | 33 |
|
| 39 |
-#define W1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 40 |
-#define W2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 41 |
-#define W3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 42 |
-#define W4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 43 |
-#define W5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 44 |
-#define W6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 45 |
-#define W7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 46 |
-#define ROW_SHIFT 11 |
|
| 47 |
-#define COL_SHIFT 20 // 6 |
|
| 48 |
- |
|
| 49 |
-static inline void idctRowCondDC (DCTELEM * row) |
|
| 50 |
-{
|
|
| 51 |
- int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 52 |
- |
|
| 53 |
-#if HAVE_FAST_64BIT |
|
| 54 |
-#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN) |
|
| 55 |
- if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) {
|
|
| 56 |
- uint64_t temp = (row[0] << 3) & 0xffff; |
|
| 57 |
- temp += temp << 16; |
|
| 58 |
- temp += temp << 32; |
|
| 59 |
- ((uint64_t *)row)[0] = temp; |
|
| 60 |
- ((uint64_t *)row)[1] = temp; |
|
| 61 |
- return; |
|
| 62 |
- } |
|
| 63 |
-#else |
|
| 64 |
- if (!(((uint32_t*)row)[1] | |
|
| 65 |
- ((uint32_t*)row)[2] | |
|
| 66 |
- ((uint32_t*)row)[3] | |
|
| 67 |
- row[1])) {
|
|
| 68 |
- uint32_t temp = (row[0] << 3) & 0xffff; |
|
| 69 |
- temp += temp << 16; |
|
| 70 |
- ((uint32_t*)row)[0]=((uint32_t*)row)[1] = |
|
| 71 |
- ((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp; |
|
| 72 |
- return; |
|
| 73 |
- } |
|
| 74 |
-#endif |
|
| 75 |
- |
|
| 76 |
- a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); |
|
| 77 |
- a1 = a0; |
|
| 78 |
- a2 = a0; |
|
| 79 |
- a3 = a0; |
|
| 80 |
- |
|
| 81 |
- /* no need to optimize : gcc does it */ |
|
| 82 |
- a0 += W2 * row[2]; |
|
| 83 |
- a1 += W6 * row[2]; |
|
| 84 |
- a2 -= W6 * row[2]; |
|
| 85 |
- a3 -= W2 * row[2]; |
|
| 86 |
- |
|
| 87 |
- b0 = MUL16(W1, row[1]); |
|
| 88 |
- MAC16(b0, W3, row[3]); |
|
| 89 |
- b1 = MUL16(W3, row[1]); |
|
| 90 |
- MAC16(b1, -W7, row[3]); |
|
| 91 |
- b2 = MUL16(W5, row[1]); |
|
| 92 |
- MAC16(b2, -W1, row[3]); |
|
| 93 |
- b3 = MUL16(W7, row[1]); |
|
| 94 |
- MAC16(b3, -W5, row[3]); |
|
| 95 |
- |
|
| 96 |
- if (AV_RN64A(row + 4)) {
|
|
| 97 |
- a0 += W4*row[4] + W6*row[6]; |
|
| 98 |
- a1 += - W4*row[4] - W2*row[6]; |
|
| 99 |
- a2 += - W4*row[4] + W2*row[6]; |
|
| 100 |
- a3 += W4*row[4] - W6*row[6]; |
|
| 101 |
- |
|
| 102 |
- MAC16(b0, W5, row[5]); |
|
| 103 |
- MAC16(b0, W7, row[7]); |
|
| 104 |
- |
|
| 105 |
- MAC16(b1, -W1, row[5]); |
|
| 106 |
- MAC16(b1, -W5, row[7]); |
|
| 107 |
- |
|
| 108 |
- MAC16(b2, W7, row[5]); |
|
| 109 |
- MAC16(b2, W3, row[7]); |
|
| 110 |
- |
|
| 111 |
- MAC16(b3, W3, row[5]); |
|
| 112 |
- MAC16(b3, -W1, row[7]); |
|
| 113 |
- } |
|
| 114 |
- |
|
| 115 |
- row[0] = (a0 + b0) >> ROW_SHIFT; |
|
| 116 |
- row[7] = (a0 - b0) >> ROW_SHIFT; |
|
| 117 |
- row[1] = (a1 + b1) >> ROW_SHIFT; |
|
| 118 |
- row[6] = (a1 - b1) >> ROW_SHIFT; |
|
| 119 |
- row[2] = (a2 + b2) >> ROW_SHIFT; |
|
| 120 |
- row[5] = (a2 - b2) >> ROW_SHIFT; |
|
| 121 |
- row[3] = (a3 + b3) >> ROW_SHIFT; |
|
| 122 |
- row[4] = (a3 - b3) >> ROW_SHIFT; |
|
| 123 |
-} |
|
| 124 |
- |
|
| 125 |
-static inline void idctSparseColPut (uint8_t *dest, int line_size, |
|
| 126 |
- DCTELEM * col) |
|
| 127 |
-{
|
|
| 128 |
- int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 129 |
- uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |
|
| 130 |
- |
|
| 131 |
- /* XXX: I did that only to give same values as previous code */ |
|
| 132 |
- a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); |
|
| 133 |
- a1 = a0; |
|
| 134 |
- a2 = a0; |
|
| 135 |
- a3 = a0; |
|
| 136 |
- |
|
| 137 |
- a0 += + W2*col[8*2]; |
|
| 138 |
- a1 += + W6*col[8*2]; |
|
| 139 |
- a2 += - W6*col[8*2]; |
|
| 140 |
- a3 += - W2*col[8*2]; |
|
| 141 |
- |
|
| 142 |
- b0 = MUL16(W1, col[8*1]); |
|
| 143 |
- b1 = MUL16(W3, col[8*1]); |
|
| 144 |
- b2 = MUL16(W5, col[8*1]); |
|
| 145 |
- b3 = MUL16(W7, col[8*1]); |
|
| 146 |
- |
|
| 147 |
- MAC16(b0, + W3, col[8*3]); |
|
| 148 |
- MAC16(b1, - W7, col[8*3]); |
|
| 149 |
- MAC16(b2, - W1, col[8*3]); |
|
| 150 |
- MAC16(b3, - W5, col[8*3]); |
|
| 151 |
- |
|
| 152 |
- if(col[8*4]){
|
|
| 153 |
- a0 += + W4*col[8*4]; |
|
| 154 |
- a1 += - W4*col[8*4]; |
|
| 155 |
- a2 += - W4*col[8*4]; |
|
| 156 |
- a3 += + W4*col[8*4]; |
|
| 157 |
- } |
|
| 158 |
- |
|
| 159 |
- if (col[8*5]) {
|
|
| 160 |
- MAC16(b0, + W5, col[8*5]); |
|
| 161 |
- MAC16(b1, - W1, col[8*5]); |
|
| 162 |
- MAC16(b2, + W7, col[8*5]); |
|
| 163 |
- MAC16(b3, + W3, col[8*5]); |
|
| 164 |
- } |
|
| 165 |
- |
|
| 166 |
- if(col[8*6]){
|
|
| 167 |
- a0 += + W6*col[8*6]; |
|
| 168 |
- a1 += - W2*col[8*6]; |
|
| 169 |
- a2 += + W2*col[8*6]; |
|
| 170 |
- a3 += - W6*col[8*6]; |
|
| 171 |
- } |
|
| 172 |
- |
|
| 173 |
- if (col[8*7]) {
|
|
| 174 |
- MAC16(b0, + W7, col[8*7]); |
|
| 175 |
- MAC16(b1, - W5, col[8*7]); |
|
| 176 |
- MAC16(b2, + W3, col[8*7]); |
|
| 177 |
- MAC16(b3, - W1, col[8*7]); |
|
| 178 |
- } |
|
| 179 |
- |
|
| 180 |
- dest[0] = cm[(a0 + b0) >> COL_SHIFT]; |
|
| 181 |
- dest += line_size; |
|
| 182 |
- dest[0] = cm[(a1 + b1) >> COL_SHIFT]; |
|
| 183 |
- dest += line_size; |
|
| 184 |
- dest[0] = cm[(a2 + b2) >> COL_SHIFT]; |
|
| 185 |
- dest += line_size; |
|
| 186 |
- dest[0] = cm[(a3 + b3) >> COL_SHIFT]; |
|
| 187 |
- dest += line_size; |
|
| 188 |
- dest[0] = cm[(a3 - b3) >> COL_SHIFT]; |
|
| 189 |
- dest += line_size; |
|
| 190 |
- dest[0] = cm[(a2 - b2) >> COL_SHIFT]; |
|
| 191 |
- dest += line_size; |
|
| 192 |
- dest[0] = cm[(a1 - b1) >> COL_SHIFT]; |
|
| 193 |
- dest += line_size; |
|
| 194 |
- dest[0] = cm[(a0 - b0) >> COL_SHIFT]; |
|
| 195 |
-} |
|
| 196 |
- |
|
| 197 |
-static inline void idctSparseColAdd (uint8_t *dest, int line_size, |
|
| 198 |
- DCTELEM * col) |
|
| 199 |
-{
|
|
| 200 |
- int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 201 |
- uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |
|
| 202 |
- |
|
| 203 |
- /* XXX: I did that only to give same values as previous code */ |
|
| 204 |
- a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); |
|
| 205 |
- a1 = a0; |
|
| 206 |
- a2 = a0; |
|
| 207 |
- a3 = a0; |
|
| 208 |
- |
|
| 209 |
- a0 += + W2*col[8*2]; |
|
| 210 |
- a1 += + W6*col[8*2]; |
|
| 211 |
- a2 += - W6*col[8*2]; |
|
| 212 |
- a3 += - W2*col[8*2]; |
|
| 213 |
- |
|
| 214 |
- b0 = MUL16(W1, col[8*1]); |
|
| 215 |
- b1 = MUL16(W3, col[8*1]); |
|
| 216 |
- b2 = MUL16(W5, col[8*1]); |
|
| 217 |
- b3 = MUL16(W7, col[8*1]); |
|
| 34 |
+#define BIT_DEPTH 8 |
|
| 35 |
+#include "simple_idct_template.c" |
|
| 36 |
+#undef BIT_DEPTH |
|
| 218 | 37 |
|
| 219 |
- MAC16(b0, + W3, col[8*3]); |
|
| 220 |
- MAC16(b1, - W7, col[8*3]); |
|
| 221 |
- MAC16(b2, - W1, col[8*3]); |
|
| 222 |
- MAC16(b3, - W5, col[8*3]); |
|
| 223 |
- |
|
| 224 |
- if(col[8*4]){
|
|
| 225 |
- a0 += + W4*col[8*4]; |
|
| 226 |
- a1 += - W4*col[8*4]; |
|
| 227 |
- a2 += - W4*col[8*4]; |
|
| 228 |
- a3 += + W4*col[8*4]; |
|
| 229 |
- } |
|
| 230 |
- |
|
| 231 |
- if (col[8*5]) {
|
|
| 232 |
- MAC16(b0, + W5, col[8*5]); |
|
| 233 |
- MAC16(b1, - W1, col[8*5]); |
|
| 234 |
- MAC16(b2, + W7, col[8*5]); |
|
| 235 |
- MAC16(b3, + W3, col[8*5]); |
|
| 236 |
- } |
|
| 237 |
- |
|
| 238 |
- if(col[8*6]){
|
|
| 239 |
- a0 += + W6*col[8*6]; |
|
| 240 |
- a1 += - W2*col[8*6]; |
|
| 241 |
- a2 += + W2*col[8*6]; |
|
| 242 |
- a3 += - W6*col[8*6]; |
|
| 243 |
- } |
|
| 244 |
- |
|
| 245 |
- if (col[8*7]) {
|
|
| 246 |
- MAC16(b0, + W7, col[8*7]); |
|
| 247 |
- MAC16(b1, - W5, col[8*7]); |
|
| 248 |
- MAC16(b2, + W3, col[8*7]); |
|
| 249 |
- MAC16(b3, - W1, col[8*7]); |
|
| 250 |
- } |
|
| 251 |
- |
|
| 252 |
- dest[0] = cm[dest[0] + ((a0 + b0) >> COL_SHIFT)]; |
|
| 253 |
- dest += line_size; |
|
| 254 |
- dest[0] = cm[dest[0] + ((a1 + b1) >> COL_SHIFT)]; |
|
| 255 |
- dest += line_size; |
|
| 256 |
- dest[0] = cm[dest[0] + ((a2 + b2) >> COL_SHIFT)]; |
|
| 257 |
- dest += line_size; |
|
| 258 |
- dest[0] = cm[dest[0] + ((a3 + b3) >> COL_SHIFT)]; |
|
| 259 |
- dest += line_size; |
|
| 260 |
- dest[0] = cm[dest[0] + ((a3 - b3) >> COL_SHIFT)]; |
|
| 261 |
- dest += line_size; |
|
| 262 |
- dest[0] = cm[dest[0] + ((a2 - b2) >> COL_SHIFT)]; |
|
| 263 |
- dest += line_size; |
|
| 264 |
- dest[0] = cm[dest[0] + ((a1 - b1) >> COL_SHIFT)]; |
|
| 265 |
- dest += line_size; |
|
| 266 |
- dest[0] = cm[dest[0] + ((a0 - b0) >> COL_SHIFT)]; |
|
| 267 |
-} |
|
| 268 |
- |
|
| 269 |
-static inline void idctSparseCol (DCTELEM * col) |
|
| 270 |
-{
|
|
| 271 |
- int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 272 |
- |
|
| 273 |
- /* XXX: I did that only to give same values as previous code */ |
|
| 274 |
- a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); |
|
| 275 |
- a1 = a0; |
|
| 276 |
- a2 = a0; |
|
| 277 |
- a3 = a0; |
|
| 278 |
- |
|
| 279 |
- a0 += + W2*col[8*2]; |
|
| 280 |
- a1 += + W6*col[8*2]; |
|
| 281 |
- a2 += - W6*col[8*2]; |
|
| 282 |
- a3 += - W2*col[8*2]; |
|
| 283 |
- |
|
| 284 |
- b0 = MUL16(W1, col[8*1]); |
|
| 285 |
- b1 = MUL16(W3, col[8*1]); |
|
| 286 |
- b2 = MUL16(W5, col[8*1]); |
|
| 287 |
- b3 = MUL16(W7, col[8*1]); |
|
| 288 |
- |
|
| 289 |
- MAC16(b0, + W3, col[8*3]); |
|
| 290 |
- MAC16(b1, - W7, col[8*3]); |
|
| 291 |
- MAC16(b2, - W1, col[8*3]); |
|
| 292 |
- MAC16(b3, - W5, col[8*3]); |
|
| 293 |
- |
|
| 294 |
- if(col[8*4]){
|
|
| 295 |
- a0 += + W4*col[8*4]; |
|
| 296 |
- a1 += - W4*col[8*4]; |
|
| 297 |
- a2 += - W4*col[8*4]; |
|
| 298 |
- a3 += + W4*col[8*4]; |
|
| 299 |
- } |
|
| 300 |
- |
|
| 301 |
- if (col[8*5]) {
|
|
| 302 |
- MAC16(b0, + W5, col[8*5]); |
|
| 303 |
- MAC16(b1, - W1, col[8*5]); |
|
| 304 |
- MAC16(b2, + W7, col[8*5]); |
|
| 305 |
- MAC16(b3, + W3, col[8*5]); |
|
| 306 |
- } |
|
| 307 |
- |
|
| 308 |
- if(col[8*6]){
|
|
| 309 |
- a0 += + W6*col[8*6]; |
|
| 310 |
- a1 += - W2*col[8*6]; |
|
| 311 |
- a2 += + W2*col[8*6]; |
|
| 312 |
- a3 += - W6*col[8*6]; |
|
| 313 |
- } |
|
| 314 |
- |
|
| 315 |
- if (col[8*7]) {
|
|
| 316 |
- MAC16(b0, + W7, col[8*7]); |
|
| 317 |
- MAC16(b1, - W5, col[8*7]); |
|
| 318 |
- MAC16(b2, + W3, col[8*7]); |
|
| 319 |
- MAC16(b3, - W1, col[8*7]); |
|
| 320 |
- } |
|
| 321 |
- |
|
| 322 |
- col[0 ] = ((a0 + b0) >> COL_SHIFT); |
|
| 323 |
- col[8 ] = ((a1 + b1) >> COL_SHIFT); |
|
| 324 |
- col[16] = ((a2 + b2) >> COL_SHIFT); |
|
| 325 |
- col[24] = ((a3 + b3) >> COL_SHIFT); |
|
| 326 |
- col[32] = ((a3 - b3) >> COL_SHIFT); |
|
| 327 |
- col[40] = ((a2 - b2) >> COL_SHIFT); |
|
| 328 |
- col[48] = ((a1 - b1) >> COL_SHIFT); |
|
| 329 |
- col[56] = ((a0 - b0) >> COL_SHIFT); |
|
| 330 |
-} |
|
| 331 |
- |
|
| 332 |
-void ff_simple_idct_put(uint8_t *dest, int line_size, DCTELEM *block) |
|
| 333 |
-{
|
|
| 334 |
- int i; |
|
| 335 |
- for(i=0; i<8; i++) |
|
| 336 |
- idctRowCondDC(block + i*8); |
|
| 337 |
- |
|
| 338 |
- for(i=0; i<8; i++) |
|
| 339 |
- idctSparseColPut(dest + i, line_size, block + i); |
|
| 340 |
-} |
|
| 341 |
- |
|
| 342 |
-void ff_simple_idct_add(uint8_t *dest, int line_size, DCTELEM *block) |
|
| 343 |
-{
|
|
| 344 |
- int i; |
|
| 345 |
- for(i=0; i<8; i++) |
|
| 346 |
- idctRowCondDC(block + i*8); |
|
| 347 |
- |
|
| 348 |
- for(i=0; i<8; i++) |
|
| 349 |
- idctSparseColAdd(dest + i, line_size, block + i); |
|
| 350 |
-} |
|
| 351 |
- |
|
| 352 |
-void ff_simple_idct(DCTELEM *block) |
|
| 353 |
-{
|
|
| 354 |
- int i; |
|
| 355 |
- for(i=0; i<8; i++) |
|
| 356 |
- idctRowCondDC(block + i*8); |
|
| 357 |
- |
|
| 358 |
- for(i=0; i<8; i++) |
|
| 359 |
- idctSparseCol(block + i); |
|
| 360 |
-} |
|
| 38 |
+#define BIT_DEPTH 10 |
|
| 39 |
+#include "simple_idct_template.c" |
|
| 40 |
+#undef BIT_DEPTH |
|
| 361 | 41 |
|
| 362 | 42 |
/* 2x4x8 idct */ |
| 363 | 43 |
|
| ... | ... |
@@ -428,7 +108,7 @@ void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block) |
| 428 | 428 |
|
| 429 | 429 |
/* IDCT8 on each line */ |
| 430 | 430 |
for(i=0; i<8; i++) {
|
| 431 |
- idctRowCondDC(block + i*8); |
|
| 431 |
+ idctRowCondDC_8(block + i*8); |
|
| 432 | 432 |
} |
| 433 | 433 |
|
| 434 | 434 |
/* IDCT4 and store */ |
| ... | ... |
@@ -503,7 +183,7 @@ void ff_simple_idct84_add(uint8_t *dest, int line_size, DCTELEM *block) |
| 503 | 503 |
|
| 504 | 504 |
/* IDCT8 on each line */ |
| 505 | 505 |
for(i=0; i<4; i++) {
|
| 506 |
- idctRowCondDC(block + i*8); |
|
| 506 |
+ idctRowCondDC_8(block + i*8); |
|
| 507 | 507 |
} |
| 508 | 508 |
|
| 509 | 509 |
/* IDCT4 and store */ |
| ... | ... |
@@ -523,7 +203,7 @@ void ff_simple_idct48_add(uint8_t *dest, int line_size, DCTELEM *block) |
| 523 | 523 |
|
| 524 | 524 |
/* IDCT8 and store */ |
| 525 | 525 |
for(i=0; i<4; i++){
|
| 526 |
- idctSparseColAdd(dest + i, line_size, block + i); |
|
| 526 |
+ idctSparseColAdd_8(dest + i, line_size, block + i); |
|
| 527 | 527 |
} |
| 528 | 528 |
} |
| 529 | 529 |
|
| ... | ... |
@@ -31,12 +31,17 @@ |
| 31 | 31 |
#include <stdint.h> |
| 32 | 32 |
#include "dsputil.h" |
| 33 | 33 |
|
| 34 |
-void ff_simple_idct_put(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 35 |
-void ff_simple_idct_add(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 34 |
+void ff_simple_idct_put_8(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 35 |
+void ff_simple_idct_add_8(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 36 |
+void ff_simple_idct_8(DCTELEM *block); |
|
| 37 |
+ |
|
| 38 |
+void ff_simple_idct_put_10(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 39 |
+void ff_simple_idct_add_10(uint8_t *dest, int line_size, DCTELEM *block); |
|
| 40 |
+void ff_simple_idct_10(DCTELEM *block); |
|
| 41 |
+ |
|
| 36 | 42 |
void ff_simple_idct_mmx(int16_t *block); |
| 37 | 43 |
void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block); |
| 38 | 44 |
void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block); |
| 39 |
-void ff_simple_idct(DCTELEM *block); |
|
| 40 | 45 |
|
| 41 | 46 |
void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block); |
| 42 | 47 |
|
| 43 | 48 |
new file mode 100644 |
| ... | ... |
@@ -0,0 +1,316 @@ |
| 0 |
+/* |
|
| 1 |
+ * Simple IDCT |
|
| 2 |
+ * |
|
| 3 |
+ * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at> |
|
| 4 |
+ * |
|
| 5 |
+ * This file is part of Libav. |
|
| 6 |
+ * |
|
| 7 |
+ * Libav 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 |
+ * Libav 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 Libav; if not, write to the Free Software |
|
| 19 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
| 20 |
+ */ |
|
| 21 |
+ |
|
| 22 |
+/** |
|
| 23 |
+ * @file |
|
| 24 |
+ * simpleidct in C. |
|
| 25 |
+ */ |
|
| 26 |
+ |
|
| 27 |
+/* |
|
| 28 |
+ based upon some outcommented c code from mpeg2dec (idct_mmx.c |
|
| 29 |
+ written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>) |
|
| 30 |
+ */ |
|
| 31 |
+ |
|
| 32 |
+#include "bit_depth_template.c" |
|
| 33 |
+ |
|
| 34 |
+#undef W1 |
|
| 35 |
+#undef W2 |
|
| 36 |
+#undef W3 |
|
| 37 |
+#undef W4 |
|
| 38 |
+#undef W5 |
|
| 39 |
+#undef W6 |
|
| 40 |
+#undef W7 |
|
| 41 |
+#undef ROW_SHIFT |
|
| 42 |
+#undef COL_SHIFT |
|
| 43 |
+#undef DC_SHIFT |
|
| 44 |
+#undef MUL |
|
| 45 |
+#undef MAC |
|
| 46 |
+ |
|
| 47 |
+#if BIT_DEPTH == 8 |
|
| 48 |
+ |
|
| 49 |
+#define W1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 50 |
+#define W2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 51 |
+#define W3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 52 |
+#define W4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 53 |
+#define W5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 54 |
+#define W6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 55 |
+#define W7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
| 56 |
+ |
|
| 57 |
+#define ROW_SHIFT 11 |
|
| 58 |
+#define COL_SHIFT 20 |
|
| 59 |
+#define DC_SHIFT 3 |
|
| 60 |
+ |
|
| 61 |
+#define MUL(a, b) MUL16(a, b) |
|
| 62 |
+#define MAC(a, b, c) MAC16(a, b, c) |
|
| 63 |
+ |
|
| 64 |
+#elif BIT_DEPTH == 10 |
|
| 65 |
+ |
|
| 66 |
+#define W1 90901 |
|
| 67 |
+#define W2 85627 |
|
| 68 |
+#define W3 77062 |
|
| 69 |
+#define W4 65535 |
|
| 70 |
+#define W5 51491 |
|
| 71 |
+#define W6 35468 |
|
| 72 |
+#define W7 18081 |
|
| 73 |
+ |
|
| 74 |
+#define ROW_SHIFT 15 |
|
| 75 |
+#define COL_SHIFT 20 |
|
| 76 |
+#define DC_SHIFT 1 |
|
| 77 |
+ |
|
| 78 |
+#define MUL(a, b) ((a) * (b)) |
|
| 79 |
+#define MAC(a, b, c) ((a) += (b) * (c)) |
|
| 80 |
+ |
|
| 81 |
+#else |
|
| 82 |
+ |
|
| 83 |
+#error "Unsupported bitdepth" |
|
| 84 |
+ |
|
| 85 |
+#endif |
|
| 86 |
+ |
|
| 87 |
+static inline void FUNC(idctRowCondDC)(DCTELEM *row) |
|
| 88 |
+{
|
|
| 89 |
+ int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 90 |
+ |
|
| 91 |
+#if HAVE_FAST_64BIT |
|
| 92 |
+#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN) |
|
| 93 |
+ if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) {
|
|
| 94 |
+ uint64_t temp = (row[0] << DC_SHIFT) & 0xffff; |
|
| 95 |
+ temp += temp << 16; |
|
| 96 |
+ temp += temp << 32; |
|
| 97 |
+ ((uint64_t *)row)[0] = temp; |
|
| 98 |
+ ((uint64_t *)row)[1] = temp; |
|
| 99 |
+ return; |
|
| 100 |
+ } |
|
| 101 |
+#else |
|
| 102 |
+ if (!(((uint32_t*)row)[1] | |
|
| 103 |
+ ((uint32_t*)row)[2] | |
|
| 104 |
+ ((uint32_t*)row)[3] | |
|
| 105 |
+ row[1])) {
|
|
| 106 |
+ uint32_t temp = (row[0] << DC_SHIFT) & 0xffff; |
|
| 107 |
+ temp += temp << 16; |
|
| 108 |
+ ((uint32_t*)row)[0]=((uint32_t*)row)[1] = |
|
| 109 |
+ ((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp; |
|
| 110 |
+ return; |
|
| 111 |
+ } |
|
| 112 |
+#endif |
|
| 113 |
+ |
|
| 114 |
+ a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); |
|
| 115 |
+ a1 = a0; |
|
| 116 |
+ a2 = a0; |
|
| 117 |
+ a3 = a0; |
|
| 118 |
+ |
|
| 119 |
+ a0 += W2 * row[2]; |
|
| 120 |
+ a1 += W6 * row[2]; |
|
| 121 |
+ a2 -= W6 * row[2]; |
|
| 122 |
+ a3 -= W2 * row[2]; |
|
| 123 |
+ |
|
| 124 |
+ b0 = MUL(W1, row[1]); |
|
| 125 |
+ MAC(b0, W3, row[3]); |
|
| 126 |
+ b1 = MUL(W3, row[1]); |
|
| 127 |
+ MAC(b1, -W7, row[3]); |
|
| 128 |
+ b2 = MUL(W5, row[1]); |
|
| 129 |
+ MAC(b2, -W1, row[3]); |
|
| 130 |
+ b3 = MUL(W7, row[1]); |
|
| 131 |
+ MAC(b3, -W5, row[3]); |
|
| 132 |
+ |
|
| 133 |
+ if (AV_RN64A(row + 4)) {
|
|
| 134 |
+ a0 += W4*row[4] + W6*row[6]; |
|
| 135 |
+ a1 += - W4*row[4] - W2*row[6]; |
|
| 136 |
+ a2 += - W4*row[4] + W2*row[6]; |
|
| 137 |
+ a3 += W4*row[4] - W6*row[6]; |
|
| 138 |
+ |
|
| 139 |
+ MAC(b0, W5, row[5]); |
|
| 140 |
+ MAC(b0, W7, row[7]); |
|
| 141 |
+ |
|
| 142 |
+ MAC(b1, -W1, row[5]); |
|
| 143 |
+ MAC(b1, -W5, row[7]); |
|
| 144 |
+ |
|
| 145 |
+ MAC(b2, W7, row[5]); |
|
| 146 |
+ MAC(b2, W3, row[7]); |
|
| 147 |
+ |
|
| 148 |
+ MAC(b3, W3, row[5]); |
|
| 149 |
+ MAC(b3, -W1, row[7]); |
|
| 150 |
+ } |
|
| 151 |
+ |
|
| 152 |
+ row[0] = (a0 + b0) >> ROW_SHIFT; |
|
| 153 |
+ row[7] = (a0 - b0) >> ROW_SHIFT; |
|
| 154 |
+ row[1] = (a1 + b1) >> ROW_SHIFT; |
|
| 155 |
+ row[6] = (a1 - b1) >> ROW_SHIFT; |
|
| 156 |
+ row[2] = (a2 + b2) >> ROW_SHIFT; |
|
| 157 |
+ row[5] = (a2 - b2) >> ROW_SHIFT; |
|
| 158 |
+ row[3] = (a3 + b3) >> ROW_SHIFT; |
|
| 159 |
+ row[4] = (a3 - b3) >> ROW_SHIFT; |
|
| 160 |
+} |
|
| 161 |
+ |
|
| 162 |
+#define IDCT_COLS do { \
|
|
| 163 |
+ a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \ |
|
| 164 |
+ a1 = a0; \ |
|
| 165 |
+ a2 = a0; \ |
|
| 166 |
+ a3 = a0; \ |
|
| 167 |
+ \ |
|
| 168 |
+ a0 += W2*col[8*2]; \ |
|
| 169 |
+ a1 += W6*col[8*2]; \ |
|
| 170 |
+ a2 += -W6*col[8*2]; \ |
|
| 171 |
+ a3 += -W2*col[8*2]; \ |
|
| 172 |
+ \ |
|
| 173 |
+ b0 = MUL(W1, col[8*1]); \ |
|
| 174 |
+ b1 = MUL(W3, col[8*1]); \ |
|
| 175 |
+ b2 = MUL(W5, col[8*1]); \ |
|
| 176 |
+ b3 = MUL(W7, col[8*1]); \ |
|
| 177 |
+ \ |
|
| 178 |
+ MAC(b0, W3, col[8*3]); \ |
|
| 179 |
+ MAC(b1, -W7, col[8*3]); \ |
|
| 180 |
+ MAC(b2, -W1, col[8*3]); \ |
|
| 181 |
+ MAC(b3, -W5, col[8*3]); \ |
|
| 182 |
+ \ |
|
| 183 |
+ if (col[8*4]) { \
|
|
| 184 |
+ a0 += W4*col[8*4]; \ |
|
| 185 |
+ a1 += -W4*col[8*4]; \ |
|
| 186 |
+ a2 += -W4*col[8*4]; \ |
|
| 187 |
+ a3 += W4*col[8*4]; \ |
|
| 188 |
+ } \ |
|
| 189 |
+ \ |
|
| 190 |
+ if (col[8*5]) { \
|
|
| 191 |
+ MAC(b0, W5, col[8*5]); \ |
|
| 192 |
+ MAC(b1, -W1, col[8*5]); \ |
|
| 193 |
+ MAC(b2, W7, col[8*5]); \ |
|
| 194 |
+ MAC(b3, W3, col[8*5]); \ |
|
| 195 |
+ } \ |
|
| 196 |
+ \ |
|
| 197 |
+ if (col[8*6]) { \
|
|
| 198 |
+ a0 += W6*col[8*6]; \ |
|
| 199 |
+ a1 += -W2*col[8*6]; \ |
|
| 200 |
+ a2 += W2*col[8*6]; \ |
|
| 201 |
+ a3 += -W6*col[8*6]; \ |
|
| 202 |
+ } \ |
|
| 203 |
+ \ |
|
| 204 |
+ if (col[8*7]) { \
|
|
| 205 |
+ MAC(b0, W7, col[8*7]); \ |
|
| 206 |
+ MAC(b1, -W5, col[8*7]); \ |
|
| 207 |
+ MAC(b2, W3, col[8*7]); \ |
|
| 208 |
+ MAC(b3, -W1, col[8*7]); \ |
|
| 209 |
+ } \ |
|
| 210 |
+ } while (0) |
|
| 211 |
+ |
|
| 212 |
+static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size, |
|
| 213 |
+ DCTELEM *col) |
|
| 214 |
+{
|
|
| 215 |
+ int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 216 |
+ INIT_CLIP; |
|
| 217 |
+ |
|
| 218 |
+ IDCT_COLS; |
|
| 219 |
+ |
|
| 220 |
+ dest[0] = CLIP((a0 + b0) >> COL_SHIFT); |
|
| 221 |
+ dest += line_size; |
|
| 222 |
+ dest[0] = CLIP((a1 + b1) >> COL_SHIFT); |
|
| 223 |
+ dest += line_size; |
|
| 224 |
+ dest[0] = CLIP((a2 + b2) >> COL_SHIFT); |
|
| 225 |
+ dest += line_size; |
|
| 226 |
+ dest[0] = CLIP((a3 + b3) >> COL_SHIFT); |
|
| 227 |
+ dest += line_size; |
|
| 228 |
+ dest[0] = CLIP((a3 - b3) >> COL_SHIFT); |
|
| 229 |
+ dest += line_size; |
|
| 230 |
+ dest[0] = CLIP((a2 - b2) >> COL_SHIFT); |
|
| 231 |
+ dest += line_size; |
|
| 232 |
+ dest[0] = CLIP((a1 - b1) >> COL_SHIFT); |
|
| 233 |
+ dest += line_size; |
|
| 234 |
+ dest[0] = CLIP((a0 - b0) >> COL_SHIFT); |
|
| 235 |
+} |
|
| 236 |
+ |
|
| 237 |
+static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size, |
|
| 238 |
+ DCTELEM *col) |
|
| 239 |
+{
|
|
| 240 |
+ int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 241 |
+ INIT_CLIP; |
|
| 242 |
+ |
|
| 243 |
+ IDCT_COLS; |
|
| 244 |
+ |
|
| 245 |
+ dest[0] = CLIP(dest[0] + ((a0 + b0) >> COL_SHIFT)); |
|
| 246 |
+ dest += line_size; |
|
| 247 |
+ dest[0] = CLIP(dest[0] + ((a1 + b1) >> COL_SHIFT)); |
|
| 248 |
+ dest += line_size; |
|
| 249 |
+ dest[0] = CLIP(dest[0] + ((a2 + b2) >> COL_SHIFT)); |
|
| 250 |
+ dest += line_size; |
|
| 251 |
+ dest[0] = CLIP(dest[0] + ((a3 + b3) >> COL_SHIFT)); |
|
| 252 |
+ dest += line_size; |
|
| 253 |
+ dest[0] = CLIP(dest[0] + ((a3 - b3) >> COL_SHIFT)); |
|
| 254 |
+ dest += line_size; |
|
| 255 |
+ dest[0] = CLIP(dest[0] + ((a2 - b2) >> COL_SHIFT)); |
|
| 256 |
+ dest += line_size; |
|
| 257 |
+ dest[0] = CLIP(dest[0] + ((a1 - b1) >> COL_SHIFT)); |
|
| 258 |
+ dest += line_size; |
|
| 259 |
+ dest[0] = CLIP(dest[0] + ((a0 - b0) >> COL_SHIFT)); |
|
| 260 |
+} |
|
| 261 |
+ |
|
| 262 |
+static inline void FUNC(idctSparseCol)(DCTELEM *col) |
|
| 263 |
+{
|
|
| 264 |
+ int a0, a1, a2, a3, b0, b1, b2, b3; |
|
| 265 |
+ |
|
| 266 |
+ IDCT_COLS; |
|
| 267 |
+ |
|
| 268 |
+ col[0 ] = ((a0 + b0) >> COL_SHIFT); |
|
| 269 |
+ col[8 ] = ((a1 + b1) >> COL_SHIFT); |
|
| 270 |
+ col[16] = ((a2 + b2) >> COL_SHIFT); |
|
| 271 |
+ col[24] = ((a3 + b3) >> COL_SHIFT); |
|
| 272 |
+ col[32] = ((a3 - b3) >> COL_SHIFT); |
|
| 273 |
+ col[40] = ((a2 - b2) >> COL_SHIFT); |
|
| 274 |
+ col[48] = ((a1 - b1) >> COL_SHIFT); |
|
| 275 |
+ col[56] = ((a0 - b0) >> COL_SHIFT); |
|
| 276 |
+} |
|
| 277 |
+ |
|
| 278 |
+void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, DCTELEM *block) |
|
| 279 |
+{
|
|
| 280 |
+ pixel *dest = (pixel *)dest_; |
|
| 281 |
+ int i; |
|
| 282 |
+ |
|
| 283 |
+ line_size /= sizeof(pixel); |
|
| 284 |
+ |
|
| 285 |
+ for (i = 0; i < 8; i++) |
|
| 286 |
+ FUNC(idctRowCondDC)(block + i*8); |
|
| 287 |
+ |
|
| 288 |
+ for (i = 0; i < 8; i++) |
|
| 289 |
+ FUNC(idctSparseColPut)(dest + i, line_size, block + i); |
|
| 290 |
+} |
|
| 291 |
+ |
|
| 292 |
+void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, DCTELEM *block) |
|
| 293 |
+{
|
|
| 294 |
+ pixel *dest = (pixel *)dest_; |
|
| 295 |
+ int i; |
|
| 296 |
+ |
|
| 297 |
+ line_size /= sizeof(pixel); |
|
| 298 |
+ |
|
| 299 |
+ for (i = 0; i < 8; i++) |
|
| 300 |
+ FUNC(idctRowCondDC)(block + i*8); |
|
| 301 |
+ |
|
| 302 |
+ for (i = 0; i < 8; i++) |
|
| 303 |
+ FUNC(idctSparseColAdd)(dest + i, line_size, block + i); |
|
| 304 |
+} |
|
| 305 |
+ |
|
| 306 |
+void FUNC(ff_simple_idct)(DCTELEM *block) |
|
| 307 |
+{
|
|
| 308 |
+ int i; |
|
| 309 |
+ |
|
| 310 |
+ for (i = 0; i < 8; i++) |
|
| 311 |
+ FUNC(idctRowCondDC)(block + i*8); |
|
| 312 |
+ |
|
| 313 |
+ for (i = 0; i < 8; i++) |
|
| 314 |
+ FUNC(idctSparseCol)(block + i); |
|
| 315 |
+} |
| ... | ... |
@@ -3956,7 +3956,8 @@ void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx) |
| 3956 | 3956 |
const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; |
| 3957 | 3957 |
|
| 3958 | 3958 |
if (accel & ACCEL_SPARC_VIS) {
|
| 3959 |
- if(avctx->idct_algo==FF_IDCT_SIMPLEVIS){
|
|
| 3959 |
+ if (avctx->bits_per_raw_sample <= 8 && |
|
| 3960 |
+ avctx->idct_algo == FF_IDCT_SIMPLEVIS) {
|
|
| 3960 | 3961 |
c->idct_put = ff_simple_idct_put_vis; |
| 3961 | 3962 |
c->idct_add = ff_simple_idct_add_vis; |
| 3962 | 3963 |
c->idct = ff_simple_idct_vis; |
| ... | ... |
@@ -337,11 +337,11 @@ int vc1_decode_sequence_header(AVCodecContext *avctx, VC1Context *v, GetBitConte |
| 337 | 337 |
v->res_fasttx = get_bits1(gb); |
| 338 | 338 |
if (!v->res_fasttx) |
| 339 | 339 |
{
|
| 340 |
- v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct; |
|
| 340 |
+ v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct_8; |
|
| 341 | 341 |
v->vc1dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add; |
| 342 | 342 |
v->vc1dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add; |
| 343 | 343 |
v->vc1dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add; |
| 344 |
- v->vc1dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add; |
|
| 344 |
+ v->vc1dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add_8; |
|
| 345 | 345 |
v->vc1dsp.vc1_inv_trans_8x4_dc = ff_simple_idct84_add; |
| 346 | 346 |
v->vc1dsp.vc1_inv_trans_4x8_dc = ff_simple_idct48_add; |
| 347 | 347 |
v->vc1dsp.vc1_inv_trans_4x4_dc = ff_simple_idct44_add; |
| ... | ... |
@@ -2369,7 +2369,7 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx) |
| 2369 | 2369 |
if (mm_flags & AV_CPU_FLAG_MMX) {
|
| 2370 | 2370 |
const int idct_algo= avctx->idct_algo; |
| 2371 | 2371 |
|
| 2372 |
- if(avctx->lowres==0){
|
|
| 2372 |
+ if (avctx->lowres == 0 && avctx->bits_per_raw_sample <= 8) {
|
|
| 2373 | 2373 |
if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SIMPLEMMX){
|
| 2374 | 2374 |
c->idct_put= ff_simple_idct_put_mmx; |
| 2375 | 2375 |
c->idct_add= ff_simple_idct_add_mmx; |
| ... | ... |
@@ -182,11 +182,13 @@ int sws_isSupportedInput(enum PixelFormat pix_fmt) |
| 182 | 182 |
|| (x)==PIX_FMT_YUV420P9LE \ |
| 183 | 183 |
|| (x)==PIX_FMT_YUV420P10LE \ |
| 184 | 184 |
|| (x)==PIX_FMT_YUV420P16LE \ |
| 185 |
+ || (x)==PIX_FMT_YUV422P10LE \ |
|
| 185 | 186 |
|| (x)==PIX_FMT_YUV422P16LE \ |
| 186 | 187 |
|| (x)==PIX_FMT_YUV444P16LE \ |
| 187 | 188 |
|| (x)==PIX_FMT_YUV422P10BE \ |
| 188 | 189 |
|| (x)==PIX_FMT_YUV420P9BE \ |
| 189 | 190 |
|| (x)==PIX_FMT_YUV420P10BE \ |
| 191 |
+ || (x)==PIX_FMT_YUV422P10BE \ |
|
| 190 | 192 |
|| (x)==PIX_FMT_YUV420P16BE \ |
| 191 | 193 |
|| (x)==PIX_FMT_YUV422P16BE \ |
| 192 | 194 |
|| (x)==PIX_FMT_YUV444P16BE \ |