virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-18 03:19:31 +02:00
Code Issues Releases Activity
94,360 Commits 37 Branches 400 Tags 464 MiB
0af08cb803
Commit Graph

264 Commits

Author SHA1 Message Date
James Almer
b8f1542dcb avfilter/vf_gblur: add missing preprocessor check 
Fixes compilation on x86_32

Signed-off-by: James Almer <jamrial@gmail.com>
 2019-06-12 10:54:59 -03:00
Ruiling Song
83f9da7768 avfilter/vf_gblur: add x86 SIMD optimizations 
The horizontal pass get ~2x performance with the patch
under single thread.

Tested overall performance using the command(avx2 enabled):
./ffmpeg -i 1080p.mp4 -vf gblur -f null /dev/null
./ffmpeg -i 1080p.mp4 -vf gblur=threads=1 -f null /dev/null
For single thread, the fps improves from 43 to 60, about 40%.
For multi-thread, the fps improves from 110 to 130, about 20%.

Signed-off-by: Ruiling Song <ruiling.song@intel.com>
 2019-06-12 08:53:11 +08:00
Paul B Mahol
dcae5ba322 avfilter: add anlmdn filter x86 SIMD optimizations 2019-01-10 21:49:47 +01:00
James Almer
ef67af31ff x86/af_afir: use three operand form forat some instructions 
Fixes compilation with old yasm versions.

Signed-off-by: James Almer <jamrial@gmail.com>
 2019-01-03 23:36:19 -03:00
James Almer
5402c1886b x86/af_afir: add ff_fcmul_add_avx() 
fcmul_add_c: 1228.8
fcmul_add_sse3: 334.3
fcmul_add_avx: 186.3

Tested on a Core i5 4460 @ 3.2GHz

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2019-01-03 10:12:19 -03:00
James Almer
82043dfd2e avfilter/af_afir: split off fcmul_add into a DSP context 
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2019-01-03 10:12:18 -03:00
James Almer
9b5bd665e1 x86/af_afir: fix processing the last element 
ff_fcmul_add_sse3() is now identical to the C version.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2019-01-03 10:12:18 -03:00
James Almer
3913d6f734 x86/scene_sad: fix link errors when HAVE_X86ASM is not defined 
Reviewed-by: Haihao Xiang <haihao.xiang@intel.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2018-11-21 22:26:07 -03:00
Paul B Mahol
c98a32e4ad avfilter/vf_blend: add 10bit support 2018-11-15 14:44:24 +01:00
Philip Langdale
1096614c42 avfilter/vf_bwdif: Use common yadif frame management logic 
After adding field type management to the common yadif logic, we can
remove the duplicate copy of that logic from bwdif.
 2018-11-14 17:41:01 -08:00
Marton Balint
6c2a7a8e9a avfilter/vf_framerate: factorize SAD functions which compute SAD for a whole frame 
Also add SIMD which works on lines because it is faster then calculating it on
8x8 blocks using pixelutils.

Signed-off-by: Marton Balint <cus@passwd.hu>
 2018-11-11 20:30:50 +01:00
Paul B Mahol
0f0d468fbc avfilter/vf_overlay: exclude nv12/nv21 formats from x86 asm check 
They are yet to be supported,

Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2018-05-03 09:22:28 +02:00
Paul B Mahol
6d7c63588c avfilter/vf_overlay: add x86 SIMD 
Specifically for yuv444, yuv422, yuv420 format when main stream has no alpha, and alpha
is straight.

Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2018-05-02 23:58:21 +02:00
Vasile Toncu
9c01cdb94e avfilter/vf_interlace: remove duplicate code with same funcionality 2018-04-23 23:48:30 +02:00
Martin Vignali
f3df42e81d avfilter/x86/vf_blend : add SIMD for 16 bit version of 
grainextract
grainmerge
average
extremity
negation
 2018-04-05 21:46:16 +02:00
Martin Vignali
8eb0bb1108 avfilter/x86/vf_blend : reorganize DIFFERENCE macro to reduce line duplication between 8bit and 16 bit version 2018-04-05 21:46:11 +02:00
Martin Vignali
53a03b5c8c avfilter/x86/vf_blend : add 16 bit version for BLEND_SIMPLE, phoenix, difference for SSE and AVX2 (x86_64) 2018-02-24 21:44:19 +01:00
Martin Vignali
6c6c9d14a8 avfilter/x86/vf_blend : indent 2018-02-24 21:44:16 +01:00
Martin Vignali
7590d58b61 avfilter/x86/vf_blend : reorganize init in order to add 16 bit version 2018-02-24 21:44:13 +01:00
Martin Vignali
3a230ce5fa avfilter/x86/vf_blend : avfilter/x86/vf_blend : add AVX2 version for each func except divide 
and optimize average, grainextract, multiply, screen, grain merge
 2018-01-28 20:21:32 +01:00
Marton Balint
4d95c6d5d7 avfilter/vf_framerate: add SIMD functions for frame blending 
Blend function speedups on x86_64 Core i5 4460:

ffmpeg -f lavfi -i allyuv -vf framerate=60:threads=1 -f null none

C:     447548411 decicycles in Blend,    2048 runs,      0 skips
SSSE3: 130020087 decicycles in Blend,    2048 runs,      0 skips
AVX2:  128508221 decicycles in Blend,    2048 runs,      0 skips

ffmpeg -f lavfi -i allyuv -vf format=yuv420p12,framerate=60:threads=1 -f null none

C:     228932745 decicycles in Blend,    2048 runs,      0 skips
SSE4:  123357781 decicycles in Blend,    2048 runs,      0 skips
AVX2:  121215353 decicycles in Blend,    2048 runs,      0 skips

Signed-off-by: Marton Balint <cus@passwd.hu>
 2018-01-28 18:50:52 +01:00
Martin Vignali
b94cd55155 avfilter/x86/vf_interlace : add AVX2 version 2018-01-11 21:03:19 +01:00
James Almer
8e0e4384b0 Revert "avfilter/vf_interlace : add AVX2 for lowpass_line 8 and 16" 
This reverts commits 1a5865b6dc and
8fb1d63d91.

They made fate interlace tests fail when AVX2 was used.

Signed-off-by: James Almer <jamrial@gmail.com>
 2017-12-19 19:04:25 -03:00
Martin Vignali
3df6e61dad avfilter/x86/vf_hflip : indent 
based on patch by Paul B Mahol
 2017-12-19 21:10:12 +01:00
Martin Vignali
f181648176 avfilter/x86/vf_hflip : add avx2 version for hflip_byte and hflip_short 2017-12-19 21:10:09 +01:00
Martin Vignali
a4a4179e83 avfilter/x86/vf_hflip : merge hflip byte and hflip short to one macro 2017-12-19 21:10:05 +01:00
Martin Vignali
8fb1d63d91 avfilter/vf_tinterlace : add AVX2 func for lowpass_line 8 and 16 2017-12-19 20:59:59 +01:00
Martin Vignali
1a5865b6dc avfilter/vf_interlace : add AVX2 for lowpass_line 8 and 16 2017-12-19 20:59:54 +01:00
Martin Vignali
d31770d9a6 avfilter/vf_interlace : move func init in ff_interlace_init and add depth arg for ff_interlace_init_x86 2017-12-19 20:59:47 +01:00
Martin Vignali
3c6dc27035 avfilter/x86/vf_interlace : avfilter/x86/vf_interlace : fix crash when using unaligned data in low_pass complex 
related to ticket 6491
 2017-12-15 11:28:29 +01:00
Martin Vignali
49dced9fd0 avfilter/x86/vf_interlace : avoid crash when data are unaligned 
ticket 6491
 2017-12-15 11:28:25 +01:00
Martin Vignali
869efbf971 avfilter/x86/vf_threshold : add threshold16 SIMD (SSE4 and AVX2) 2017-12-09 14:47:09 +01:00
James Almer
f2aa0ce5a0 x86/vf_hflip: use xor to zero initialize registers 
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-12-07 19:34:12 -03:00
James Almer
dc33fe1d00 x86/vf_hflip: don't load the width argument twice 
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-12-07 19:34:12 -03:00
James Almer
cc2ba526d4 x86/vf_threshold: make threshold8 functions work on x86_32 
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-12-04 15:46:09 -03:00
Paul B Mahol
5ff0d2acae avfilter/x86/vf_hflip.asm: fix building on x32 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2017-12-04 15:08:43 +01:00
Paul B Mahol
86fda8be3f avfilter: add hflip x86 SIMD 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2017-12-04 09:58:25 +01:00
James Almer
b73304f79e x86vf_threshold/: use the PBLENDVB macro 
Fixes building with yasm

Tested-by: stevenliu
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-12-04 02:22:30 -03:00
Martin Vignali
6e3e696591 avfilter/x86/vf_threshold : cosmetic indent 2017-12-03 19:17:28 +01:00
Martin Vignali
9719d57b34 avfilter/x86/vf_threshold : add avx2 version for threshold 8 2017-12-03 19:17:23 +01:00
Martin Vignali
51345cb1d5 avfilter/x86/vf_threshold : make macro for threshold8 in order to add avx2 version 2017-12-03 19:17:19 +01:00
Paul B Mahol
bbfcb1b7c8 avfilter/vf_threshold: add x86 SIMD 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2017-12-02 14:58:56 +01:00
James Almer
2904db9045 Merge commit '994c4bc10751e39c7ed9f67ffd0c0dea5223daf2' 
* commit '994c4bc10751e39c7ed9f67ffd0c0dea5223daf2':
  x86util: Port all macros to cpuflags

See d5f8a642f6

Merged-by: James Almer <jamrial@gmail.com>
 2017-10-21 12:15:57 -03:00
Thomas Mundt
40bfaa190c avfilter/interlace: add support for 10 and 12 bit 
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Thomas Mundt <tmundt75@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-09-23 16:19:58 -03:00
Thomas Mundt
a7f6bfdc18 avfilter/interlace: prevent over-sharpening with the complex low-pass filter 
The complex vertical low-pass filter slightly over-sharpens the picture. This becomes visible when several transcodings are cascaded and the error potentises, e.g. some generations of HD->SD SD->HD.
To prevent this behaviour the destination pixel must not exceed the source pixel when the average of the pixels above and below is less than the source pixel. And the other way around.

Tested and approved in a visual transcoding cascade test by video professionals.
SSIM/PSNR test with the first generation of an HD->SD file as a reference against the 6th generation(3 x SD->HD HD->SD):
Results without the patch:
SSIM Y:0.956508 (13.615881) U:0.991601 (20.757750) V:0.993004 (21.551382) All:0.974405 (15.918463)
PSNR y:31.838009 u:48.424280 v:48.962711 average:34.759466 min:31.699297 max:40.857847
Results with the patch:
SSIM Y:0.970051 (15.236232) U:0.991883 (20.905857) V:0.993174 (21.658049) All:0.981290 (17.279202)
PSNR y:34.412108 u:48.504454 v:48.969496 average:37.264644 min:34.310637 max:42.373392

Signed-off-by: Thomas Mundt <tmundt75@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2017-09-15 22:40:21 +02:00
Paul B Mahol
f8d0689d3f avfilter/vf_blend: rename addition128 and difference128 to grainmerge and grainextract 2017-08-24 14:45:52 +02:00
James Almer
5688fd77b5 x86/vf_limiter: make limiter functions work on x86_32 
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-07-13 18:17:17 -03:00
Paul B Mahol
01e545d046 avfilter: add limiter filter 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2017-07-08 11:49:54 +02:00
James Almer
d2ef9e6e7f x86/vf_blend: use ABS2 macro 2017-06-27 20:45:55 -03:00
James Almer
0daa1cf073 x86/vf_blend: optimize difference and negation functions 
Process more pixels per loop.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2017-06-27 13:17:23 -03:00