diff options
author | Michael Niedermayer <michaelni@gmx.at> | 2011-06-03 05:19:30 +0200 |
---|---|---|
committer | Michael Niedermayer <michaelni@gmx.at> | 2011-06-03 05:19:30 +0200 |
commit | 99eb31e263a24bc6c5a7a3f455a2bcb04a60e70e (patch) | |
tree | 83aab2f083fd3d6923d4e76fe0cc0c41ef7aef35 /libavcodec/x86 | |
parent | 9034001b17077e9da5205c4344eb1b88b9882f03 (diff) | |
parent | f190f676bc93a7e80344f2feeb3b9b44604d4717 (diff) | |
download | ffmpeg-streaming-99eb31e263a24bc6c5a7a3f455a2bcb04a60e70e.zip ffmpeg-streaming-99eb31e263a24bc6c5a7a3f455a2bcb04a60e70e.tar.gz |
Merge remote-tracking branch 'qatar/master'
* qatar/master: (25 commits)
Replace custom DEBUG preprocessor trickery by the standard one.
vorbis: Remove non-compiling debug statement.
vorbis: Remove pointless DEBUG #ifdef around debug output macros.
cook: Remove non-compiling debug output.
Remove pointless #ifdefs around function declarations in a header.
Replace #ifdef + av_log() combinations by av_dlog().
Replace custom debug output functions by av_dlog().
cook: Remove unused debug functions.
Remove stray extra arguments from av_dlog() invocations.
targa: fix big-endian build
v4l2: remove one forgotten use of AVFormatParameters.pix_fmt.
vfwcap: add a framerate private option.
v4l2: add a framerate private option.
libdc1394: add a framerate private option.
fbdev: add a framerate private option.
bktr: add a framerate private option.
oma: check avio_read() return value
nutdec: remove unused variable
Remove unused variables
swscale: allocate larger buffer to handle altivec overreads.
...
Conflicts:
ffmpeg.c
libavcodec/dca.c
libavcodec/dirac.c
libavcodec/error_resilience.c
libavcodec/h264.c
libavcodec/mpeg12.c
libavcodec/mpeg4videodec.c
libavcodec/mpegvideo.c
libavcodec/mpegvideo_enc.c
libavcodec/pthread.c
libavcodec/rv10.c
libavcodec/s302m.c
libavcodec/shorten.c
libavcodec/truemotion2.c
libavcodec/utils.c
libavdevice/dv1394.c
libavdevice/fbdev.c
libavdevice/libdc1394.c
libavdevice/v4l2.c
libavformat/4xm.c
libavformat/apetag.c
libavformat/asfdec.c
libavformat/avidec.c
libavformat/mmf.c
libavformat/mpeg.c
libavformat/mpegenc.c
libavformat/mpegts.c
libavformat/oggdec.c
libavformat/oggparseogm.c
libavformat/rl2.c
libavformat/rmdec.c
libavformat/rpl.c
libavformat/rtpdec_latm.c
libavformat/sauce.c
libavformat/sol.c
libswscale/utils.c
tests/ref/vsynth1/error
tests/ref/vsynth2/error
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec/x86')
-rw-r--r-- | libavcodec/x86/h264_idct_10bit.asm | 178 | ||||
-rw-r--r-- | libavcodec/x86/h264_intrapred.asm | 6 | ||||
-rw-r--r-- | libavcodec/x86/h264_intrapred_init.c | 8 |
3 files changed, 84 insertions, 108 deletions
diff --git a/libavcodec/x86/h264_idct_10bit.asm b/libavcodec/x86/h264_idct_10bit.asm index 3d0004e..3f7cf4c 100644 --- a/libavcodec/x86/h264_idct_10bit.asm +++ b/libavcodec/x86/h264_idct_10bit.asm @@ -133,7 +133,7 @@ ADD4x4IDCT avx %macro ADD16_OP 3 cmp byte [r4+%3], 0 jz .skipblock%2 - mov r5d, dword [r1+%2*4] + mov r5d, [r1+%2*4] call add4x4_idct_%1 .skipblock%2: %if %2<15 @@ -159,7 +159,7 @@ cglobal h264_idct_add16_10_%1, 5,6 ADD16_OP %1, 13, 7+3*8 ADD16_OP %1, 14, 6+4*8 ADD16_OP %1, 15, 7+4*8 - RET + REP_RET %endmacro INIT_XMM @@ -201,7 +201,7 @@ IDCT_ADD16_10 avx INIT_MMX cglobal h264_idct_dc_add_10_mmx2,3,3 - movd m0, dword [r1] + movd m0, [r1] paddd m0, [pd_32] psrad m0, 6 lea r1, [r2*3] @@ -215,7 +215,7 @@ cglobal h264_idct_dc_add_10_mmx2,3,3 ;----------------------------------------------------------------------------- %macro IDCT8_DC_ADD 1 cglobal h264_idct8_dc_add_10_%1,3,3,7 - mov r1d, dword [r1] + mov r1d, [r1] add r1, 32 sar r1, 6 movd m0, r1d @@ -240,26 +240,27 @@ IDCT8_DC_ADD avx ;----------------------------------------------------------------------------- %macro AC 2 .ac%2 - mov r5d, dword [r1+(%2+0)*4] + mov r5d, [r1+(%2+0)*4] call add4x4_idct_%1 - mov r5d, dword [r1+(%2+1)*4] + mov r5d, [r1+(%2+1)*4] add r2, 64 call add4x4_idct_%1 add r2, 64 jmp .skipadd%2 %endmacro +%assign last_block 16 %macro ADD16_OP_INTRA 3 - cmp word [r4+%3], 0 + cmp word [r4+%3], 0 jnz .ac%2 - mov r6d, dword [r2+ 0] - or r6d, dword [r2+64] + mov r5d, [r2+ 0] + or r5d, [r2+64] jz .skipblock%2 - mov r5d, dword [r1+(%2+0)*4] + mov r5d, [r1+(%2+0)*4] call idct_dc_add_%1 .skipblock%2: -%if %2<15 - add r2, 128 +%if %2<last_block-2 + add r2, 128 %endif .skipadd%2: %endmacro @@ -287,12 +288,15 @@ cglobal h264_idct_add16intra_10_%1,5,7,8 ADD16_OP_INTRA %1, 10, 4+4*8 ADD16_OP_INTRA %1, 12, 6+3*8 ADD16_OP_INTRA %1, 14, 6+4*8 - RET -%assign i 14 -%rep 8 - AC %1, i -%assign i i-2 -%endrep + REP_RET + AC %1, 8 + AC %1, 10 + AC %1, 12 + AC %1, 14 + AC %1, 0 + AC %1, 2 + AC %1, 4 + AC %1, 6 %endmacro INIT_XMM @@ -302,47 +306,33 @@ INIT_AVX IDCT_ADD16INTRA_10 avx %endif +%assign last_block 24 ;----------------------------------------------------------------------------- ; h264_idct_add8(pixel **dst, const int *block_offset, dctcoef *block, int stride, const uint8_t nnzc[6*8]) ;----------------------------------------------------------------------------- %macro IDCT_ADD8 1 cglobal h264_idct_add8_10_%1,5,7 - mov r5, 16 - add r2, 1024 -%ifdef PIC - lea r11, [scan8_mem] -%endif -%ifdef ARCH_X86_64 - mov r10, r0 -%endif -.nextblock: - movzx r6, byte [scan8+r5] - movzx r6, byte [r4+r6] - or r6d, dword [r2] - test r6, r6 - jz .skipblock %ifdef ARCH_X86_64 - mov r0d, dword [r1+r5*4] - add r0, [r10] -%else - mov r0, r0m - mov r0, [r0] - add r0, dword [r1+r5*4] + mov r10, r0 %endif - IDCT4_ADD_10 r0, r2, r3 -.skipblock: - inc r5 - add r2, 64 - test r5, 3 - jnz .nextblock + add r2, 1024 + mov r0, [r0] + ADD16_OP_INTRA %1, 16, 1+1*8 + ADD16_OP_INTRA %1, 18, 1+2*8 %ifdef ARCH_X86_64 - add r10, gprsize + mov r0, [r10+gprsize] %else - add r0mp, gprsize + mov r0, r0m + mov r0, [r0+gprsize] %endif - test r5, 4 - jnz .nextblock + ADD16_OP_INTRA %1, 20, 1+4*8 + ADD16_OP_INTRA %1, 22, 1+5*8 REP_RET + AC %1, 16 + AC %1, 18 + AC %1, 20 + AC %1, 22 + %endmacro ; IDCT_ADD8 INIT_XMM @@ -356,51 +346,51 @@ IDCT_ADD8 avx ; void h264_idct8_add(pixel *dst, dctcoef *block, int stride) ;----------------------------------------------------------------------------- %macro IDCT8_1D 2 - SWAP 0, 1 - psrad m4, m5, 1 - psrad m1, m0, 1 - paddd m4, m5 - paddd m1, m0 - paddd m4, m7 - paddd m1, m5 - psubd m4, m0 - paddd m1, m3 - - psubd m0, m3 - psubd m5, m3 - paddd m0, m7 - psubd m5, m7 - psrad m3, 1 - psrad m7, 1 - psubd m0, m3 - psubd m5, m7 - - SWAP 1, 7 - psrad m1, m7, 2 - psrad m3, m4, 2 - paddd m3, m0 - psrad m0, 2 - paddd m1, m5 - psrad m5, 2 - psubd m0, m4 - psubd m7, m5 - - SWAP 5, 6 - psrad m4, m2, 1 - psrad m6, m5, 1 - psubd m4, m5 - paddd m6, m2 - - mova m2, %1 - mova m5, %2 - SUMSUB_BA d, 5, 2 - SUMSUB_BA d, 6, 5 - SUMSUB_BA d, 4, 2 - SUMSUB_BA d, 7, 6 - SUMSUB_BA d, 0, 4 - SUMSUB_BA d, 3, 2 - SUMSUB_BA d, 1, 5 - SWAP 7, 6, 4, 5, 2, 3, 1, 0 ; 70315246 -> 01234567 + SWAP 0, 1 + psrad m4, m5, 1 + psrad m1, m0, 1 + paddd m4, m5 + paddd m1, m0 + paddd m4, m7 + paddd m1, m5 + psubd m4, m0 + paddd m1, m3 + + psubd m0, m3 + psubd m5, m3 + paddd m0, m7 + psubd m5, m7 + psrad m3, 1 + psrad m7, 1 + psubd m0, m3 + psubd m5, m7 + + SWAP 1, 7 + psrad m1, m7, 2 + psrad m3, m4, 2 + paddd m3, m0 + psrad m0, 2 + paddd m1, m5 + psrad m5, 2 + psubd m0, m4 + psubd m7, m5 + + SWAP 5, 6 + psrad m4, m2, 1 + psrad m6, m5, 1 + psubd m4, m5 + paddd m6, m2 + + mova m2, %1 + mova m5, %2 + SUMSUB_BA d, 5, 2 + SUMSUB_BA d, 6, 5 + SUMSUB_BA d, 4, 2 + SUMSUB_BA d, 7, 6 + SUMSUB_BA d, 0, 4 + SUMSUB_BA d, 3, 2 + SUMSUB_BA d, 1, 5 + SWAP 7, 6, 4, 5, 2, 3, 1, 0 ; 70315246 -> 01234567 %endmacro %macro IDCT8_1D_FULL 1 @@ -536,7 +526,7 @@ IDCT8_ADD avx %macro IDCT8_ADD4_OP 3 cmp byte [r4+%3], 0 jz .skipblock%2 - mov r0d, dword [r6+%2*4] + mov r0d, [r6+%2*4] add r0, r5 call h264_idct8_add1_10_%1 .skipblock%2: diff --git a/libavcodec/x86/h264_intrapred.asm b/libavcodec/x86/h264_intrapred.asm index 191eb8d..a9da6a1 100644 --- a/libavcodec/x86/h264_intrapred.asm +++ b/libavcodec/x86/h264_intrapred.asm @@ -836,7 +836,6 @@ PRED8x8_H ssse3 ;----------------------------------------------------------------------------- ; void pred8x8_top_dc_mmxext(uint8_t *src, int stride) ;----------------------------------------------------------------------------- -%ifdef CONFIG_GPL cglobal pred8x8_top_dc_mmxext, 2,5 sub r0, r1 movq mm0, [r0] @@ -927,7 +926,6 @@ cglobal pred8x8_dc_mmxext, 2,5 movq [r4+r1*1], m1 movq [r4+r1*2], m1 RET -%endif ;----------------------------------------------------------------------------- ; void pred8x8_dc_rv40(uint8_t *src, int stride) @@ -1083,7 +1081,6 @@ cglobal pred8x8_tm_vp8_ssse3, 2,3,6 ;----------------------------------------------------------------------------- ; void pred8x8l_top_dc(uint8_t *src, int has_topleft, int has_topright, int stride) ;----------------------------------------------------------------------------- -%ifdef CONFIG_GPL %macro PRED8x8L_TOP_DC 1 cglobal pred8x8l_top_dc_%1, 4,4 sub r0, r3 @@ -2476,7 +2473,6 @@ PRED8x8L_HORIZONTAL_DOWN sse2 INIT_MMX %define PALIGNR PALIGNR_SSSE3 PRED8x8L_HORIZONTAL_DOWN ssse3 -%endif ;----------------------------------------------------------------------------- ; void pred4x4_dc_mmxext(uint8_t *src, const uint8_t *topright, int stride) @@ -2608,7 +2604,6 @@ cglobal pred4x4_vertical_vp8_mmxext, 3,3 ;----------------------------------------------------------------------------- ; void pred4x4_down_left_mmxext(uint8_t *src, const uint8_t *topright, int stride) ;----------------------------------------------------------------------------- -%ifdef CONFIG_GPL INIT_MMX cglobal pred4x4_down_left_mmxext, 3,3 sub r0, r2 @@ -2786,4 +2781,3 @@ cglobal pred4x4_down_right_mmxext, 3,3 psrlq m0, 8 movh [r0+r2*1], m0 RET -%endif diff --git a/libavcodec/x86/h264_intrapred_init.c b/libavcodec/x86/h264_intrapred_init.c index fa5b86f..f001242 100644 --- a/libavcodec/x86/h264_intrapred_init.c +++ b/libavcodec/x86/h264_intrapred_init.c @@ -129,7 +129,6 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth h->pred16x16[HOR_PRED8x8 ] = ff_pred16x16_horizontal_mmxext; h->pred16x16[DC_PRED8x8 ] = ff_pred16x16_dc_mmxext; h->pred8x8 [HOR_PRED8x8 ] = ff_pred8x8_horizontal_mmxext; -#if CONFIG_GPL h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_mmxext; h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_mmxext; h->pred8x8l [HOR_PRED ] = ff_pred8x8l_horizontal_mmxext; @@ -142,9 +141,7 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth h->pred4x4 [DIAG_DOWN_RIGHT_PRED] = ff_pred4x4_down_right_mmxext; h->pred4x4 [VERT_RIGHT_PRED ] = ff_pred4x4_vertical_right_mmxext; h->pred4x4 [HOR_DOWN_PRED ] = ff_pred4x4_horizontal_down_mmxext; -#endif h->pred4x4 [DC_PRED ] = ff_pred4x4_dc_mmxext; -#if CONFIG_GPL if (codec_id == CODEC_ID_VP8 || codec_id == CODEC_ID_H264) h->pred4x4 [DIAG_DOWN_LEFT_PRED ] = ff_pred4x4_down_left_mmxext; if (codec_id == CODEC_ID_SVQ3 || codec_id == CODEC_ID_H264) @@ -156,7 +153,6 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth h->pred8x8 [TOP_DC_PRED8x8 ] = ff_pred8x8_top_dc_mmxext; h->pred8x8 [DC_PRED8x8 ] = ff_pred8x8_dc_mmxext; } -#endif if (codec_id == CODEC_ID_VP8) { h->pred16x16[PLANE_PRED8x8] = ff_pred16x16_tm_vp8_mmxext; h->pred8x8 [DC_PRED8x8 ] = ff_pred8x8_dc_rv40_mmxext; @@ -181,13 +177,11 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth if (mm_flags & AV_CPU_FLAG_SSE2) { h->pred16x16[DC_PRED8x8 ] = ff_pred16x16_dc_sse2; -#if CONFIG_GPL h->pred8x8l [DIAG_DOWN_LEFT_PRED ] = ff_pred8x8l_down_left_sse2; h->pred8x8l [DIAG_DOWN_RIGHT_PRED] = ff_pred8x8l_down_right_sse2; h->pred8x8l [VERT_RIGHT_PRED ] = ff_pred8x8l_vertical_right_sse2; h->pred8x8l [VERT_LEFT_PRED ] = ff_pred8x8l_vertical_left_sse2; h->pred8x8l [HOR_DOWN_PRED ] = ff_pred8x8l_horizontal_down_sse2; -#endif if (codec_id == CODEC_ID_VP8) { h->pred16x16[PLANE_PRED8x8] = ff_pred16x16_tm_vp8_sse2; h->pred8x8 [PLANE_PRED8x8] = ff_pred8x8_tm_vp8_sse2; @@ -207,7 +201,6 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth h->pred16x16[HOR_PRED8x8 ] = ff_pred16x16_horizontal_ssse3; h->pred16x16[DC_PRED8x8 ] = ff_pred16x16_dc_ssse3; h->pred8x8 [HOR_PRED8x8 ] = ff_pred8x8_horizontal_ssse3; -#if CONFIG_GPL h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_ssse3; h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_ssse3; h->pred8x8l [HOR_PRED ] = ff_pred8x8l_horizontal_ssse3; @@ -218,7 +211,6 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id, const int bit_depth h->pred8x8l [VERT_LEFT_PRED ] = ff_pred8x8l_vertical_left_ssse3; h->pred8x8l [HOR_UP_PRED ] = ff_pred8x8l_horizontal_up_ssse3; h->pred8x8l [HOR_DOWN_PRED ] = ff_pred8x8l_horizontal_down_ssse3; -#endif if (codec_id == CODEC_ID_VP8) { h->pred8x8 [PLANE_PRED8x8] = ff_pred8x8_tm_vp8_ssse3; h->pred4x4 [TM_VP8_PRED ] = ff_pred4x4_tm_vp8_ssse3; |