Commit Graph

1801 Commits

Author SHA1 Message Date
Michael Niedermayer
3bb2297351 Merge commit 'efd26bedec9a345a5960dbfcbaec888418f2d4e6'
* commit 'efd26bedec9a345a5960dbfcbaec888418f2d4e6':
  build: Add explanatory comments to (optimization) blocks in the Makefiles

Conflicts:
	libavcodec/ppc/Makefile
	libavcodec/x86/Makefile

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-15 20:25:12 +02:00
Michael Niedermayer
c1df467d73 Merge commit '835f798c7d20bca89eb4f3593846251ad0d84e4b'
* commit '835f798c7d20bca89eb4f3593846251ad0d84e4b':
  mpegvideo: cosmetics: Lowercase ugly uppercase MPV_ function name prefixes

Conflicts:
	libavcodec/h261dec.c
	libavcodec/intrax8.c
	libavcodec/mjpegenc.c
	libavcodec/mpeg12dec.c
	libavcodec/mpeg12enc.c
	libavcodec/mpeg4videoenc.c
	libavcodec/mpegvideo.c
	libavcodec/mpegvideo.h
	libavcodec/mpegvideo_enc.c
	libavcodec/rv10.c
	libavcodec/x86/mpegvideoenc.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-15 20:11:56 +02:00
Diego Biurrun
efd26bedec build: Add explanatory comments to (optimization) blocks in the Makefiles 2014-08-15 02:55:21 -07:00
Diego Biurrun
835f798c7d mpegvideo: cosmetics: Lowercase ugly uppercase MPV_ function name prefixes 2014-08-15 01:26:33 -07:00
James Darnley
54a51d3840 lavc/flacenc: partially unroll loop in flac_enc_lpc_16
It now does 12 samples per iteration, up from 4.

From 1.8 to 3.2 times faster again.  3.6 to 5.7 times faster overall.
Runtime is reduced by a further 2 to 18%.  Overall runtime reduced by
4 to 50%.

Same conditions as before apply.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-13 03:09:26 +02:00
James Darnley
0081a14e7d lavc/flacenc: add sse4 version of the 16-bit lpc encoder
From 1.8 to 2.4 times faster.  Runtime is reduced by 2 to 39%.  The
speed-up generally increases with compression_level.

This lpc encoder is not used with levels < 3 so it provides no speed-up
in these cases.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-13 01:14:47 +02:00
Ronald S. Bultje
45bed0ab30 vp9/x86: fix bug in intra_pred_hd_32x32.
Fixes mismatch in first keyframe in sample
ffvp9_fails_where_libvpx.succeeds.webm from ticket 3849. There's still
a second mismatch a few frames into the sample.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-12 13:11:21 +02:00
James Almer
c97870d1a1 x86/dca: remove unused header
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-12 12:46:53 +02:00
James Almer
e20ff251a6 x86/ttadsp: remove an unnecessary mova
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-12 12:29:05 +02:00
Michael Niedermayer
3841f2ae66 Merge commit 'd35b94fbabd8beb5d566c0b5d01688aff62c3b36'
* commit 'd35b94fbabd8beb5d566c0b5d01688aff62c3b36':
  avcodec: Rename xvidmmx IDCT to xvid

Conflicts:
	doc/APIchanges
	libavcodec/version.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-09 12:11:13 +02:00
Michael Niedermayer
0dcebb9f63 Merge commit '84d173d3de97c753234ab0c0b50551d51413d663'
* commit '84d173d3de97c753234ab0c0b50551d51413d663':
  xvididct: Ensure that the scantable permutation is always set correctly

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-08 22:17:04 +02:00
Diego Biurrun
d35b94fbab avcodec: Rename xvidmmx IDCT to xvid
The Xvid IDCT is not MMX-specific.
2014-08-08 11:13:30 -07:00
Diego Biurrun
84d173d3de xvididct: Ensure that the scantable permutation is always set correctly
This fixes cases where the scantable permuation would get overwritten by
the general idctdsp initialization.
2014-08-08 11:13:29 -07:00
Christophe Gisquet
75837e9add x86: sbrdsp/fft: reuse ps_neg constant
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 19:25:08 +02:00
Christophe Gisquet
51dd80e751 x86: diracdsp: reuse constants
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 19:25:02 +02:00
Christophe Gisquet
6622a6cff3 x86: dwt: better share constants
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 19:24:57 +02:00
Christophe Gisquet
71db2d08b1 x86: better share ff_pw_2
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 19:24:49 +02:00
Christophe Gisquet
4e128ab0b1 x86: vpx/h264/hevc/mpeg2: share constants
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 18:36:31 +02:00
Michael Niedermayer
305f72aee7 avcodec: Change get_pixels() to ptrdiff_t linesize
Found-by: ubitux
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 15:50:54 +02:00
Christophe Gisquet
6786848585 hevc_deblock: change tc type
The x86 asm expects int32_t so use that type.

Reviewed-by: Mickaël Raulet <mraulet@insa-rennes.fr>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-06 12:38:26 +02:00
James Almer
de417982e8 x86/vp9lpf: use fewer instructions in SPLATB_MIX
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-05 02:47:54 +02:00
Christophe Gisquet
e8c003edd2 x86: hevc_deblock: remove unnecessary masking
The unpacks/shuffles later on makes it unnecessary.

Before:
1508 decicycles in h, 2096759 runs, 393 skips
2512 decicycles in v, 2095422 runs, 1730 skips

After:
1477 decicycles in h, 2096745 runs, 407 skips
2484 decicycles in v, 2095297 runs, 1855 skips

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-04 17:46:04 +02:00
James Almer
b7863c972c x86/hevc_mc: use fewer instructions in hevc_put_hevc_{uni, bi}_w[24]_{8, 10, 12}
Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-04 14:47:15 +02:00
James Almer
b1a44e6bf5 x86/hevc_mc: remove an unnecessary pxor
Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-04 14:35:08 +02:00
James Almer
d0f56ca071 x86/hevc_deblock: improve 8bit transpose store macros
Up to four instructions less depending on function and instruction set.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-03 04:24:15 +02:00
Michael Niedermayer
f54e01c24e Merge commit 'a786c8259dafeca9744252230b5d78f67810770c'
* commit 'a786c8259dafeca9744252230b5d78f67810770c':
  idct: Split off Xvid IDCT

Conflicts:
	libavcodec/Makefile
	libavcodec/mpeg4videodec.c
	libavcodec/x86/Makefile
	libavcodec/x86/idctdsp_init.c

This split is somewhat restructured leaving the xvid IDCT available
outside mpeg4 if manually selected.

The code also could not be merged unchanged as it conflicted with a
bugfix in FFmpeg

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-01 16:21:52 +02:00
Diego Biurrun
a786c8259d idct: Split off Xvid IDCT
The Xvid IDCT is only required to decode some Xvid-encoded MPEG-4 files,
so there is no point in having it as an unconditional part of idctdsp.
2014-08-01 01:25:18 -07:00
James Almer
62baf5b853 x86/hevc_deblock: use existing x86util transpose macro in chroma_{10, 12}
Cosmetic change. No measurable difference in speed.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-31 22:56:21 +02:00
Christophe Gisquet
a507623bad x86: hevc_mc: fix register count usage
A macro was using a fixed register, causing too many GPRs to be
declared as used.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 22:50:50 +02:00
James Almer
73c4f63ba5 x86/hevc_deblock: add add ff_hevc_[hv]_loop_filter_luma_{8, 10, 12}_avx
~5% faster than SSSE3

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 14:04:59 +02:00
James Almer
88ba821f23 x86/hevc_deblock: improve luma functions register allocation
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 13:38:05 +02:00
James Almer
c74b08c5c6 x86/hevc_deblock: remove some unnecessary instructions
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 13:27:44 +02:00
James Almer
4f91bb0ff0 x86/hevc_deblock: use psignw instead of pmullw where possible
It's slightly faster

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 03:42:29 +02:00
Michael Niedermayer
a91c5ed008 Merge commit '4f8cf0dc4ef6110174056df7edd9dc2f2a988b6d'
* commit '4f8cf0dc4ef6110174056df7edd9dc2f2a988b6d':
  x86: build: Restore ordering of OBJS lines

Conflicts:
	libavcodec/x86/Makefile

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-29 00:34:53 +02:00
Diego Biurrun
4f8cf0dc4e x86: build: Restore ordering of OBJS lines 2014-07-28 13:19:04 -07:00
James Almer
664e9e4331 x86/hevc_deblock: load less data in hevc_h_loop_filter_luma_8
Reading 8 bytes is enough.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-28 21:55:22 +02:00
James Almer
f137876182 x86/hevc_idct: add a colon to labels
This fixes a warning spam when using NASM

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-28 21:43:32 +02:00
Christophe Gisquet
81943a10b5 x86: hevc_mc: load less data in epel filters
Before:
5679 decicycles in epel_bi, 2059976 runs, 37176 skips
3468 decicycles in epel_uni, 1040886 runs, 7690 skips

After:
5323 decicycles in epel_bi, 2059493 runs, 37659 skips
3262 decicycles in epel_uni, 1040871 runs, 7705 skips

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-27 18:34:39 +02:00
Christophe Gisquet
36284ae981 x86: hevc_mc: replace one lea by add
Should have been in 036f11bdb5.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-27 17:42:56 +02:00
James Almer
bfb3b2b7a6 x86/hevc_idct: add 12bit idct_dc
Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-27 00:30:56 +02:00
Michael Niedermayer
d4a9e89b27 avcodec/x86/hevcdsp_init: make license header consistent
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-27 00:28:44 +02:00
Michael Niedermayer
706f81a2c2 Merge commit '1a880b2fb8456ce68eefe5902bac95fea1e6a72d'
* commit '1a880b2fb8456ce68eefe5902bac95fea1e6a72d':
  hevc: SSE2 and SSSE3 loop filters

Conflicts:
	libavcodec/hevcdsp.c
	libavcodec/hevcdsp.h
	libavcodec/x86/Makefile
	libavcodec/x86/hevc_deblock.asm
	libavcodec/x86/hevcdsp_init.c

See: de7b89fd43 and several others
Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-27 00:20:48 +02:00
James Almer
1ace9573dc x86/hevc_idct: replace old and unused idct functions
Only 8-bit and 10-bit idct_dc() functions are included (adding others should be trivial).

Benchmarks on an Intel Core i5-4200U:

idct8x8_dc
       SSE2   MMXEXT  C
cycles 22     26      57

idct16x16_dc
       AVX2   SSE2    C
cycles 27     32      249

idct32x32_dc
       AVX2   SSE2    C
cycles 62     126     1375

Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Mickaël Raulet <mraulet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-26 18:00:11 +02:00
Pierre Edouard Lepere
1a880b2fb8 hevc: SSE2 and SSSE3 loop filters
Additional contributions by James Almer <jamrial@gmail.com>,
Carl Eugen Hoyos <cehoyos@ag.or.at>, Fiona Glaser <fiona@x264.com> and
Anton Khirnov <anton@khirnov.net>

Signed-off-by: Anton Khirnov <anton@khirnov.net>
2014-07-26 15:01:01 +00:00
Christophe Gisquet
036f11bdb5 x86: hevc_mc: replace simple leas by adds
lea is detrimental for those simple cases. No impact overall to
the change though.

Before:
15017 decicycles in q, 1016152 runs, 32424 skips
15382 decicycles in q_bi, 1013673 runs, 34903 skips
3713 decicycles in e, 2074534 runs, 22618 skips
3901 decicycles in e_bi, 2065509 runs, 31643 skips
7852 decicycles in q_uni, 520165 runs, 4123 skips
2398 decicycles in e_uni, 1043339 runs, 5237 skips

After:
14898 decicycles in q, 1016295 runs, 32281 skips
15119 decicycles in q_bi, 1015392 runs, 33184 skips
3682 decicycles in e, 2073224 runs, 23928 skips
3720 decicycles in e_bi, 2065043 runs, 32109 skips
7643 decicycles in q_uni, 520280 runs, 4008 skips
2363 decicycles in e_uni, 1043780 runs, 4796 skips

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-26 05:41:04 +02:00
Mickaël Raulet
bd0f2d316f x86/hevc: add 12bits support for MC
cherry picked from commit 3fcb7a4595a6f40100a22110a5805e3b7510c0fd

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-26 01:55:20 +02:00
Mickaël Raulet
7df98d8c4d x86/hevc: remove unused constant in deblocking filter
cherry picked from commit a3f7282eaa6f1ab0524fb966c6eade50c3025f99

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-26 01:20:40 +02:00
Mickaël Raulet
7bdcf5c934 x86/hevc: add 12bits support for deblocking filter
cherry picked from commit 97d46afe320c7d61d7b9525e5f5588355cde4bb0

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-26 01:19:42 +02:00
Michael Niedermayer
2904d052b7 Merge commit '7fb993d338d88f2f62e0a358b6c9f3eb9a3a08ac'
* commit '7fb993d338d88f2f62e0a358b6c9f3eb9a3a08ac':
  qpeldsp: Mark source pointer in qpel_mc_func function pointer const

Conflicts:
	libavcodec/h264qpel_template.c
	libavcodec/x86/cavsdsp.c
	libavcodec/x86/rv40dsp_init.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-25 13:05:08 +02:00
Diego Biurrun
7fb993d338 qpeldsp: Mark source pointer in qpel_mc_func function pointer const 2014-07-25 02:52:54 -07:00