ffmpeg/FFmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2024-09-25 15:50:48 +00:00
Code Issues Packages Projects Releases Wiki Activity
40,293 Commits 37 Branches 398 Tags 539 MiB
c5fca0174d
Commit Graph

905 Commits

Author SHA1 Message Date
Diego Biurrun
82dd1026cf x86: dsputil: Move hpeldsp-related declarations to a separate header 2014-03-22 06:17:29 -07:00
Diego Biurrun
6655c933a8 x86: dsputil: Move fpel declarations to a separate header 2014-03-22 06:17:29 -07:00
Diego Biurrun
322a1dda97 dsputil: Refactor duplicated CALL_2X_PIXELS / PIXELS16 macros 2014-03-22 06:17:29 -07:00
Diego Biurrun
600b854ad8 imgconvert: Move ff_deinterlace_line_*_mmx declarations out of dsputil 2014-03-22 06:17:29 -07:00
Diego Biurrun
1a8d0cf77e x86: dsputil: Move inline assembly macros to a separate header 2014-03-22 06:17:29 -07:00
Diego Biurrun
82bb304801 dsputil: Use correct type in me_cmp_func function pointer 2014-03-20 05:03:23 -07:00
Diego Biurrun
0e083d7e43 build: Group general components separate from de/encoders in arch Makefiles 
This is in line with how the top-level libavcodec Makefile is structured.
 2014-03-20 05:03:23 -07:00
Diego Biurrun
5169e68895 dsputil: Propagate bit depth information to all (sub)init functions 
This avoids recalculating the value over and over again.
 2014-03-20 05:03:23 -07:00
Diego Biurrun
db3f61a04f x86: dsputil_init: Drop some unnecessary parentheses 2014-03-13 08:15:51 -07:00
Diego Biurrun
441b093915 x86: dsputil_init: K&R formatting cosmetics 2014-03-13 08:15:51 -07:00
Diego Biurrun
4cb4680c10 x86: dsputil_x86.h: K&R formatting cosmetics 2014-03-13 08:15:51 -07:00
Diego Biurrun
f8bbebecfd x86: motion_est: K&R formatting cosmetics 2014-03-13 08:15:51 -07:00
Diego Biurrun
a36947c167 dsputilenc_mmx: K&R formatting cosmetics 2014-03-13 08:15:51 -07:00
Diego Biurrun
38675229a8 dsputil_mmx: K&R formatting cosmetics 2014-03-13 08:15:51 -07:00
Diego Biurrun
6a8b35dc88 dsputilenc_mmx: Merge two assignment blocks with identical conditions 2014-03-13 08:15:51 -07:00
Diego Biurrun
55519926ef x86: Make function prototype comments in assembly code consistent 
This helps grepping for functions, among other things.
 2014-03-13 05:50:29 -07:00
Diego Biurrun
edd1f833fa x86: h264_idct_10_bit: Use proper type in function prototype comments 2014-03-13 05:50:29 -07:00
Diego Biurrun
831a118078 Update dsputil- and SIMD-related comments to match reality more closely 2014-03-13 05:50:29 -07:00
Diego Biurrun
17608f6ee3 x86: Add some more missing headers 2014-03-13 05:50:28 -07:00
Diego Biurrun
08dba0e1c3 x86: mpegvideoenc: Remove some remnants of the long-gone libmpeg2 IDCT 2014-03-13 05:50:28 -07:00
Diego Biurrun
3bfdee00cd x86: dcadsp: Fix linking with yasm and optimizations disabled 
Some optimized functions reference optimized symbols, so the functions
must be explicitly disabled when those symbols are unavailable.
 2014-03-05 23:16:21 +01:00
Diego Biurrun
3741aa37c2 x86: cabac: Use correct #includes to make header compile standalone 2014-03-05 13:32:25 +01:00
Christophe Gisquet
4cb6964244 dcadec: simplify decoding of VQ high frequencies 
The vector dequantization has a test in a loop preventing effective SIMD
implementation. By moving it out of the loop, this loop can be DSPized.

Therefore, modify the current DSP implementation. In particular, the
DSP implementation no longer has to handle null loop sizes.

The decode_hf implementations have following timings:

For x86 Arrandale:
        C  SSE SSE2 SSE4
win32: 260 162  119  104
win64: 242 N/A   89   72

The arm NEON optimizations follow in a later patch as external asm. The
now unused check for the y modifier in arm inline asm is removed from
configure.
 2014-02-28 13:03:22 +01:00
Christophe Gisquet
08e3ea60ff x86: synth filter float: implement SSE2 version 
Timings for Arrandale:
          C    SSE
win32:  2108   334
win64:  1152   322

Factorizing the inner loop with a call/jmp is a >15 cycles cost, even with
the jmp destination being aligned.

Unrolling for ARCH_X86_64 is a 20 cycles gain.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
 2014-02-28 13:00:48 +01:00
Christophe Gisquet
ad507d7907 x86: dcadsp: implement SSE lfe_dir 
Results for Arrandale/Windows:
32: 1670 -> 316
64:  728 -> 298

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
 2014-02-28 13:00:47 +01:00
Diego Biurrun
b23650491f prores: Use consistent names for DSP arch initialization functions 2014-02-28 10:34:55 +01:00
Diego Biurrun
017a06a9ee x86: dsputil: Use correct file name as multiple inclusion guard 2014-02-20 04:16:15 -08:00
Diego Biurrun
b23bc95920 x86: dca: Add missing multiple inclusion guards 2014-02-19 10:19:15 +01:00
Janne Grunau
5c1c6e8226 dca: include dcadsp.h in {arm,x86}/dca.h for checkheaders 2014-02-08 13:38:36 +01:00
Janne Grunau
0cffd6fff5 x86: use the inline int8x8_fmul_int32 only if inline SSE2 is availbale 
Fixes compilation with MSVC. Also does not rely on on earlier config.h
include but include it directly.
 2014-02-08 12:10:56 +01:00
Christophe Gisquet
5b59a9fc61 x86: dcadsp: implement int8x8_fmul_int32 
For the callable function (as opposed to the inline one):
         C  SSE  SSE2  SSE4
Win32:  47   42   29    26
Win64:  30   33   25    23
The SSE version is neither compiled nor set for ARCH_X86_64, as the
inlinable function takes over.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
 2014-02-07 22:52:40 +01:00
Ronald S. Bultje
9ee9c679a7 x86: videodsp: Fix a bug in a %if statement where we used '%%' instead of '&&'. 
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
 2014-01-30 15:33:23 +01:00
Ronald S. Bultje
51daafb02e x86: videodsp: Properly mark sse2 instructions in emulated_edge_mc as such. 
Should fix crashes or corrupt output on pre-SSE2 CPUs when they were
using SSE2-code (e.g. AMD Athlon XP 2400+ or Intel Pentium III) in
hfix or hvar single-edge (left/right) extension functions.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
 2014-01-30 15:30:01 +01:00
Diego Biurrun
aab40bbfd5 x86: dsputil: Simplify xvmc deprecation conditional 2014-01-15 15:23:46 +01:00
Diego Biurrun
46bacb5cc6 x86: Consistently use cpu flag detection macros in places that still miss it 2014-01-14 00:04:58 +01:00
Diego Biurrun
4c642d8d98 x86: hpeldsp: Add missing av_cold attribute to init function 2014-01-09 15:09:07 +01:00
Diego Biurrun
b0be1ae792 x86: avcodec: Add a bunch of missing #includes for av_cold 2014-01-09 15:09:07 +01:00
Anton Khirnov
a03a642d5c h264: do not use 422 functions for monochrome 
Fixes invalid memory access.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org
 2014-01-06 08:25:36 +01:00
Anton Khirnov
dfc50ac85e x86: mpegvideo: move denoise_dct asm to mpegvideoenc 
This function is encoding-only.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
 2013-12-20 17:16:11 +01:00
Diego Biurrun
4958f35a2e dsputil: Move apply_window_int16 to ac3dsp 
The (optimized) functions are used nowhere else.
 2013-12-08 17:57:15 +01:00
Diego Biurrun
3d7c84747d x86: Initialize mmxext after amd3dnow optimizations 
The mmxext optimizations should be at least equally fast if available and
amd3dnow optimizations are being deprecated. Thus the former should
override the latter, not the other way around.
 2013-12-04 18:52:48 +01:00
Diego Biurrun
7ffaa19570 dsputil: x86: Move ff_inv_zigzag_direct16 table init to mpegvideo 
The table is MMX-specific and used nowhere else.
 2013-12-02 04:05:18 +01:00
Diego Biurrun
cf7860db60 x86: dsputil: Suppress deprecation warnings for XvMC bits 
These parts are scheduled for removal on the next version bump.

Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
 2013-11-28 16:04:30 +01:00
Ronald S. Bultje
72ca830f51 lavc: VP9 decoder 
Originally written by Ronald S. Bultje <rsbultje@gmail.com> and
Clément Bœsch <u@pkh.me>

Further contributions by:
Anton Khirnov <anton@khirnov.net>
Diego Biurrun <diego@biurrun.de>
Luca Barbato <lu_zero@gentoo.org>
Martin Storsjö <martin@martin.st>

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
 2013-11-15 10:16:28 +01:00
Ronald S. Bultje
458446acfa lavc: Edge emulation with dst/src linesize 
Allow supporting files for which the image stride is smaller than
the maximum block size + number of subpel mc taps, e.g. a 64x64 VP9
file or a 16x16 VP8 file with -fflags +emu_edge.
 2013-11-15 10:16:27 +01:00
Diego Biurrun
19e30a58fc Deprecate obsolete XvMC hardware decoding support 
XvMC has long ago been superseded by newer acceleration APIs, such as
VDPAU, and few downstreams still support it. Furthermore XvMC is not
implemented within the hwaccel framework, but requires its own specific
code in the MPEG-1/2 decoder, which is a maintenance burden.
 2013-11-13 21:07:45 +01:00
Diego Biurrun
0338c39698 dsputil: Split off H.263 bits into their own H263DSPContext 2013-11-08 12:40:47 +01:00
Diego Biurrun
e2b5b09789 x86: rv40dsp: Use PAVGB instruction macro where appropriate 2013-11-04 21:14:39 +01:00
Mikulas Patocka
694d997afe x86: hpeldsp: Use PAVGB instruction macro where necessary 
Signed-off-by: Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
 2013-11-04 01:29:23 +01:00
Diego Biurrun
1700b4e678 x86: vp8dsp: Split loopfilter code into a separate file 2013-11-01 22:05:20 +01:00
Diego Biurrun
6405ca7d4a x86: h264_idct: Update comments to match 8/10-bit depth optimization split 2013-10-07 21:46:46 +02:00
Henrik Gramner
bbe4a6db44 x86inc: Utilize the shadow space on 64-bit Windows 
Store XMM6 and XMM7 in the shadow space in functions that
clobbers them. This way we don't have to adjust the stack
pointer as often, reducing the number of instructions as
well as code size.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
 2013-10-07 06:25:35 -04:00
Diego Biurrun
ce1e8045e0 x86: fdct: Employ more specific ifdefs 
This avoids building mmxext and sse2 code when disabled by configure.
 2013-10-06 22:02:25 +02:00
Diego Biurrun
2ddb35b911 x86: dsputil: Separate ff_add_hfyu_median_prediction_cmov from dsputil_mmx 
The function does not depend on MMX and compilation without MMX enabled
fails if the function is compiled conditional on MMX availability.
 2013-10-05 19:21:15 +02:00
Diego Biurrun
258414d077 x86: fdct: Initialize optimized fdct implementations in the standard way 2013-10-05 18:20:52 +02:00
Diego Biurrun
0b8b2ae5e9 x86: xviddct: Employ more specific ifdefs 
This avoids building mmxext and sse2 code when disabled by configure.
 2013-10-05 18:14:58 +02:00
Diego Biurrun
6cc133ec58 x86: fdct: Only build fdct code if encoders have been enabled 
fdct is only initialized if encoders are enabled.
 2013-10-04 10:50:44 +02:00
Martin Storsjö
1daea5232f x86: Add an xmm clobbering wrapper for avcodec_encode_video2 
This is required since 187105ff8 when we started trying to
wrap this function as well.

Signed-off-by: Martin Storsjö <martin@martin.st>
 2013-09-16 22:22:41 +03:00
Hendrik Leppkes
a06a5b78e2 mathops/x86: work around inline asm miscompilation with GCC 4.8.1 
The volatile is not required here, and prevents a miscompilation with GCC
4.8.1 when building on x86 with --cpu=i686

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
 2013-09-15 11:15:07 -04:00
Diego Biurrun
e998b56362 x86: avcodec: Consistently structure CPU extension initialization 2013-08-29 13:07:37 +02:00
Diego Biurrun
6369ba3c9c x86: avcodec: Use convenience macros to check for CPU flags 2013-08-29 13:07:37 +02:00
Diego Biurrun
cd52917237 x86: rv40dsp: Move inline assembly optimizations out of YASM init section 2013-08-28 23:59:24 +02:00
Diego Biurrun
a64f6a04ac dsputil: x86: Hide arch-specific initialization details 
Also give consistent names to init functions.
 2013-08-28 23:59:24 +02:00
Diego Biurrun
8506ff97c9 vp56: Mark VP6-only optimizations as such. 
Most of our VP56 optimizations are VP6-only and will stay that way.
So avoid compiling them for VP5-only builds.
 2013-08-23 14:42:19 +02:00
Diego Biurrun
e7b31844f6 x86: Split DCT and FFT initialization into separate files 2013-08-21 20:15:27 +02:00
Diego Biurrun
0b45269c2d x86: h264_idct: Remove incorrect comment 2013-08-21 15:09:58 +02:00
Diego Biurrun
3ac7fa81b2 Consistently use "cpu_flags" as variable/parameter name for CPU flags 2013-07-18 00:31:35 +02:00
Christophe Gisquet
b6293e2798 fmtconvert: Explicitly use int32_t instead of int 
Signed-off-by: Martin Storsjö <martin@martin.st>
 2013-07-17 11:02:47 +03:00
Luca Barbato
2b379a9251 mlpdsp: x86: Respect cpuflags 2013-07-12 04:34:49 +02:00
Jason Garrett-Glaser
d222f6e39e cabac: x86 version of get_cabac_bypass 
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
 2013-07-04 16:06:10 +02:00
Diego Biurrun
186599ffe0 build: cosmetics: Place unconditional before conditional OBJS lines 
Signed-off-by: Martin Storsjö <martin@martin.st>
 2013-05-30 02:17:31 +03:00
Diego Biurrun
004b81c465 mpegvideo: Remove commented-out PARANOID debug cruft 2013-05-15 23:53:42 +02:00
Diego Biurrun
1399931d07 x86: dsputil: Rename dsputil_mmx.h --> dsputil_x86.h 
The header is not (anymore) MMX-specific.
 2013-05-12 22:28:07 +02:00
Diego Biurrun
245b76a108 x86: dsputil: Split inline assembly from init code 
Also remove some pointless comments.
 2013-05-12 22:28:07 +02:00
Diego Biurrun
46bb456853 x86: dsputil: Refactor pixels16 wrapper functions with a macro 2013-05-12 22:28:07 +02:00
Diego Biurrun
f54b55058a configure: Rename cmov processor capability to i686 
The goal is to make the capapility slightly more general and have it
cover the availability of the nopl instruction in addition to cmov.
 2013-05-12 21:23:38 +02:00
Christophe Gisquet
2c299d4165 x86: sbrdsp: implement SSE2 qmf_pre_shuffle 
From 253 to 51 cycles on Arrandale and Win64.
44 cycles on SandyBridge.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
 2013-05-10 09:31:27 +02:00
Diego Biurrun
f243bf7aa2 x86: dsputil: Remove unused argument from QPEL_OP macro 2013-05-08 18:18:58 +02:00
Diego Biurrun
3d40c1ee74 x86: dsputil: Move TRANSPOSE4 macro to the only place it is used 2013-05-08 18:18:23 +02:00
Diego Biurrun
71469f3b63 x86: dsputil: Move constant declarations into separate header 2013-05-08 18:18:23 +02:00
Diego Biurrun
ed880050ed x86: dsputil: Group all assembly constants together in constants.c 2013-05-08 01:04:04 +02:00
Diego Biurrun
8761466760 x86: dsputil: Move ff_pd assembly constants to the only place they are used 2013-05-08 01:04:04 +02:00
Diego Biurrun
1b343cedd7 x86: dsputil: Remove unused ff_pb_3F constant 2013-05-07 18:03:35 +02:00
Diego Biurrun
3334cbec0a x86: dsputil: Remove unused MOVQ_BONE macro 2013-05-07 18:03:35 +02:00
Diego Biurrun
63bac48f73 x86: dsputil: Move rv40-specific functions where they belong 2013-05-07 18:03:35 +02:00
Diego Biurrun
92f8e06ecb x86: dsputil hpeldsp: Move shared template functions into separate object 2013-05-07 18:03:34 +02:00
Diego Biurrun
7edaf4edb5 x86: rnd_template: Eliminate pointless OP_AVG macro indirection 2013-05-07 18:03:34 +02:00
Diego Biurrun
110796739a x86: hpeldsp: Move avg_pixels8_x2_mmx() out of hpeldsp_rnd_template.c 
The function is only instantiated once, so there is no point
in keeping it in a template file.
 2013-05-06 11:02:08 +02:00
Diego Biurrun
dc1b328d0d x86: hpeldsp: Only compile MMX hpeldsp code if MMX is enabled 2013-05-06 11:02:08 +02:00
Diego Biurrun
9e5e76ef9e x86: More specific ifdefs for dsputil/hpeldsp init functions 2013-05-06 11:02:07 +02:00
Diego Biurrun
6fee1b90ce avcodec: Add av_cold attributes to init functions missing them 2013-05-04 21:09:45 +02:00
Diego Biurrun
a5f8873620 silly typo fixes 2013-05-03 18:26:12 +02:00
Christophe Gisquet
5a97469a4f x86: sbrdsp: Implement SSE2 qmf_deint_bfly 
Sandybridge: 47 cycles

Having a loop counter is a 7 cycle gain.
Unrolling is another 7 cycle gain.
Working in reverse scan is another 6 cycles.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
 2013-05-03 18:23:14 +02:00
Diego Biurrun
bf7c3c6b15 x86: dsputil: Move cavs and vc1-specific functions where they belong 2013-05-02 11:45:37 +02:00
Diego Biurrun
9328062321 x86: dsputil: Move avg_pixels16_mmx() out of rnd_template.c 
The function does not do any rounding, so there is no point in
keeping it in a round template file.
 2013-05-02 11:45:37 +02:00
Diego Biurrun
9c112a6158 x86: dsputil: Move avg_pixels8_mmx() out of rnd_template.c 
The function is only instantiated once, so there is no point
in keeping it in a template file.
 2013-05-02 11:45:37 +02:00
Diego Biurrun
9b3a04d306 x86: Move duplicated put_pixels{8|16}_mmx functions into their own file 2013-05-02 11:16:45 +02:00
Diego Biurrun
f2e9d44a57 x86: Drop unnecessary ff_ name prefixes from static functions 2013-04-30 16:02:03 +02:00
Diego Biurrun
643e433bf7 mpegaudiosp: More consistent names for ppc/x86 optimization files 2013-04-30 12:19:43 +02:00
Diego Biurrun
97c56ad796 x86: dsputil: Remove a set of pointless #ifs around function declarations 2013-04-30 01:42:32 +02:00