generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
78,150 Commits 24 Branches 241 Tags
Commit Graph

2095 Commits

Author SHA1 Message Date
Ronald S. Bultje
f76423d097 vp9: add x86 simd (sse2/ssse3) for iadst4 10bpp functions. 2015-10-13 11:05:58 -04:00
Ronald S. Bultje
6b579cf547 vp9: add 10bpp simd (mmxext/ssse3) for idct_idct_4x4. 2015-10-13 11:05:58 -04:00
Ronald S. Bultje
1c3be32533 vp9: add 10/12bpp mmxext-optimized iwht_iwht_4x4 function. 2015-10-13 11:05:57 -04:00
Christophe Gisquet
b6594a9605 x86: dct-test: add more idcts 
In particular for 10 and 12 bits.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-13 16:03:04 +02:00
Christophe Gisquet
7ece8b50b1 x86: simple_idct: 12bits versions 
On 12 frames of a 444p 12 bits DNxHR sequence, _put function:
C:         78902 decicycles in idct,  262071 runs,     73 skips
avx:       32478 decicycles in idct,  262045 runs,     99 skips

Difference between the 2:
stddev:    0.39 PSNR:104.47 MAXDIFF:    2

This is unavoidable and due to the scale factors used in the x86
version, which cannot match the C ones.

In addition, the trick of adding an initial bias to the input of a
pass can overflow, as the input coefficients are already 15bits,
which is the maximum this function can handle.

Overall, however, the omse on 12 bits samples goes from 0.16916 to
0.16883. Reducing rowshift by 1 improves to 0.0908, but causes
overflows.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-13 15:34:32 +02:00
Christophe Gisquet
4369b9dc7b x86: simple_idct(_put): 10bits versions 
Modeled from the prores version. Clips to [0;1023] and is bitexact.
Bitexactness requires to add offsets in different places compared to
prores or C, and makes the function approximately 2% slower.

For 16 frames of a DNxHD 4:2:2 10bits test sequence:

C:    60861 decicycles in idct, 1048205 runs,    371 skips
sse2: 27567 decicycles in idct, 1048216 runs,    360 skips
avx:  26272 decicycles in idct, 1048171 runs,    405 skips

The add version is not implemented, so the corresponding dsp
function is set to NULL to make it clear in a code executing it.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-13 13:32:21 +02:00
Christophe Gisquet
e652f69b35 x86: simple_idct10_template: fix overflow in pass 
When the input of a pass has 15 or 16 bits of precision (in particular
the column pass), the addition of a bias to W4 may lead to overflows
in the input to pmaddwd.

This requires postponing the adding of the bias to after the first
butterfly. To do so, the fact that m15, unused although zeroed, is
exploited. In case the pass is safe, an address can be directly used,
and the number of xmm regs can be decreased. Otherwise, the 32bits bias
is loaded into it.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-13 12:51:10 +02:00
Christophe Gisquet
e9a68b0316 x86: prores: templatize 10 bits simple_idct 
This should be reused for a generic simple_idct10 function.
Requires a bit of trickery to declare common constants in C.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-13 01:10:34 +02:00
James Almer
dab5f65b25 x86/takdsp: use arithmetic shift instructions 
p1 and p2 are int32_t.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-10-09 23:52:39 -03:00
Paul B Mahol
35af7add6f avcodec/takdec: add x86 SIMD for rest of decorrelation modes 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
 2015-10-09 21:38:15 +02:00
Ronald S. Bultje
ce78729033 vp9: don't keep a stack pointer if we don't need it. 
This saves one register in a few cases on 32bit builds with unaligned
stack (e.g. MSVC), making the code slightly easier to maintain.

(Can someone please test this on 32bit+msvc and confirm make fate-vp9
and tests/checkasm/checkasm still work after this patch?)
 2015-10-07 08:55:19 -04:00
James Almer
72254b19b8 x86/alacdsp: add simd optimized functions 
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-10-06 20:22:00 -03:00
Ronald S. Bultje
cb912b4521 vp9: fix msvc build by using 6 GPRs on 32bit if stack!=aligned. 2015-10-05 16:51:05 -04:00
Christophe Gisquet
f827a17005 blockdsp: reindent after parameter removal 
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-03 23:34:56 +02:00
Ronald S. Bultje
061b67fb50 vp9: 10/12bpp SIMD (sse2/ssse3/avx) for directional intra prediction. 2015-10-03 14:42:39 -04:00
Ronald S. Bultje
26ece7a511 vp9: 16bpp tm/dc/h/v intra pred simd (mostly sse2) functions. 2015-10-03 14:42:39 -04:00
Ronald S. Bultje
db7786e8ff vp9: sse2/ssse3/avx 16bpp loopfilter x86 simd. 2015-10-03 14:42:39 -04:00
Ganesh Ajjanagadde
0493e42eb2 avcodec/x86/hpeldsp_rnd_template: silence -Wunused-function on --disable-mmx 
This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
Header guards are too brittle and ugly for this case.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-03 14:24:41 +02:00
Christophe Gisquet
562ba4a827 blockdsp: remove high bitdepth parameter 
It is only (mis-)used to set the dsp fucntions clear_block(s). But
these functions always work on 16bits-wide elements, which make
the parameter useless and actually harmful, as it causes all content
on more than 8-bits to not use accelerated functions.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-10-02 04:38:40 +02:00
James Almer
3178931a14 x86/hevc_sao: move 10/12bit functions into a separate file 
Tested-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-30 02:59:55 -03:00
Ganesh Ajjanagadde
308e7484a3 avcodec/x86/rnd_template: silence -Wunused-function on --disable-mmx 
This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
Header guards are too brittle and ugly for this case.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-29 19:37:26 +02:00
Michael Niedermayer
1b82b934a1 avcodec/x86/sbrdsp: Fix using uninitialized upper 32bit of noise 
Fixes crash
Fixes: flicker-1.scout3d21443372922.28.m4a

Found-by: Dale Curtis <dalecurtis@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-29 13:23:25 +02:00
Ganesh Ajjanagadde
07cd8d5676 avcodec/x86/cavsdsp: silence -Wunused-variable on --disable-mmx 
This silences -Wunused-variable when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
The alternative of header guards will make it far too ugly.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-24 04:27:50 +02:00
Ganesh Ajjanagadde
0544c95fd6 avcodec/x86/mpegaudiodsp: silence -Wunused-variable on --disable-mmx 
This silences -Wunused-variable when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
The alternative of header guards will make it far too ugly.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-22 23:45:03 +02:00
Ganesh Ajjanagadde
4f90818ea1 avcodec/x86/rv40dsp_init: silence -Wunused-variable on --disable-mmx 
This silences -Wunused-variable when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
The alternative of header guards will make it far too ugly.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-22 23:45:03 +02:00
James Almer
7086154aaa x86/vp9dsp: fix local header include 
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-21 14:37:32 -03:00
James Almer
91fcb10f08 x86/vp9dsp: add missing header include 
Fixes make checkheaders

Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-21 14:34:08 -03:00
James Almer
4bb6cb4c7d x86/vp9mc: fix string concatenation of fullpel function names 
Fixes compilation with NASM

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-20 12:32:27 -03:00
Ganesh Ajjanagadde
92fabca427 avcodec/x86/hpeldsp_rnd_template: silence -Wunused-function on --disable-mmx 
This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.
Header guards are too brittle and ugly for this case.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-20 04:00:42 +02:00
Ganesh Ajjanagadde
e681baf638 avcodec/x86/mpegvideoenc: silence -Wunused-function on --disable-mmx 
This silences -Wunused-function when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-19 23:26:57 +02:00
Ganesh Ajjanagadde
f0c635f577 avcodec/x86/hpeldsp_init: silence -Wunused-function on --disable-mmx 
This silences some of the -Wunused-function warnings when compiled with --disable-mmx, e.g
http://fate.ffmpeg.org/log.cgi?time=20150919094617&log=compile&slot=x86_64-archlinux-gcc-disable-mmx.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-19 23:10:52 +02:00
James Almer
6f9ba0cb82 x86/vp9dsp: add missing preprocessor guards 
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-19 13:33:53 -03:00
James Almer
e47564828b x86/vp9mc: add missing preprocessor guards 
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-18 15:14:53 -03:00
James Almer
2f9ab15960 x86/vp9: add avx2 subpel MC SIMD for 10/12bpp 
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-18 12:28:55 -03:00
Michael Niedermayer
58fe57d5a0 avcodec/mpeg12enc: Basic support for encoding non even QPs for -non_linear_quant 1 
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-18 02:52:57 +02:00
Michael Niedermayer
2d35757814 avcodec/mpegvideo: Change mpeg2 unquant to work with higher precission qscale 
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-18 02:39:17 +02:00
Ronald S. Bultje
344d519040 vp9: add subpel MC SIMD for 10/12bpp. 2015-09-16 21:11:34 -04:00
Ronald S. Bultje
77f359670f vp9: add fullpel (avg) MC SIMD for 10/12bpp. 2015-09-16 21:11:34 -04:00
Ronald S. Bultje
6354ff0383 vp9: add fullpel (put) MC SIMD for 10/12bpp. 2015-09-16 21:11:34 -04:00
Hendrik Leppkes
7b865c222e Merge commit '5d14cf199990cd378904a2618b5c72c4b02290f6' 
* commit '5d14cf199990cd378904a2618b5c72c4b02290f6':
  mpegvideo: Make sure mpegutils.h is included where needed

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
 2015-09-16 11:23:40 +02:00
Vittorio Giovara
5d14cf1999 mpegvideo: Make sure mpegutils.h is included where needed 2015-09-13 17:34:45 +02:00
James Almer
d5f8a642f6 x86: port PSIGNW to cpuflags 
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
 2015-09-11 23:27:03 -03:00
Ronald S. Bultje
4b66274a86 vp9: save one (PSIGNW) instruction in iadst16_1d sse2/ssse3. 2015-09-11 20:36:51 -04:00
Ronald S. Bultje
fd8b90f5f6 vp9: fix overflow in 8x8 topleft 32x32 idct ssse3 version. 
Also disable the mmx/iwht optimization when the bitexact flag is set.
With synthetically coded coefficients (i.e. these that lead to a
residual well outside the [-255,255] range), our optimizations will
overflow. It doesn't make sense to fix the overflows, since they can
only occur on synthetic input, not on real fwht-generated input. Thus,
add a bitexact flag that disables this optimization.
 2015-09-10 07:51:16 -04:00
Hendrik Leppkes
5d8e836d0e Replace all remaining occurances of step/depth_minus1 and offset_plus1 2015-09-08 17:10:48 +02:00
Ronald S. Bultje
f12093fffd vp9: fix integer overflows in sse2 version of iadst4. 2015-09-06 15:07:19 -04:00
Michael Niedermayer
8d860f9a77 avcodec/x86/w64xmmtest: Fix another build failure 
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
 2015-09-05 22:15:53 +02:00
Ronald S. Bultje
086c9b78d4 vp9: fix rounding error in idct_8x8_ssse3. 2015-09-05 15:50:02 -04:00
Hendrik Leppkes
41194f065c Merge commit 'cad40a3833ad81a352e7657ec6f7d637cea3b798' 
* commit 'cad40a3833ad81a352e7657ec6f7d637cea3b798':
  lavc: Drop deprecated deinterlace module

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>
 2015-09-05 17:06:14 +02:00
Vittorio Giovara
cad40a3833 lavc: Drop deprecated deinterlace module 
Deprecated in 03/2013.
 2015-08-28 16:04:19 +02:00