]> git.sesse.net Git - x264/log
x264
8 years agoaarch64: Remove commas LLVM's assembler complains about
Janne Grunau [Tue, 18 Aug 2015 08:25:08 +0000 (10:25 +0200)]
aarch64: Remove commas LLVM's assembler complains about

8 years agoarm: Implement x264_sub8x16_dct_dc_neon
Martin Storsjö [Thu, 13 Aug 2015 20:59:31 +0000 (23:59 +0300)]
arm: Implement x264_sub8x16_dct_dc_neon

checkasm timing      Cortex-A7      A8     A9
sub8x16_dct_dc_c            6386    3901   4080
sub8x16_dct_dc_neon         1491    698    917

8 years agoarm: Optimize x264_deblock_h_chroma_neon
Martin Storsjö [Thu, 13 Aug 2015 20:59:28 +0000 (23:59 +0300)]
arm: Optimize x264_deblock_h_chroma_neon

Shuffle both chroma components together as a 16 bit unit, and
don't write the unchanged columns (like in x264_deblock_h_luma_neon
and in the aarch64 version of the function).

This causes a minor slowdown for x264_deblock_v_chroma_neon, but
it is negligible compared to the speedup.

checkasm timing      Cortex-A7    A8    A9
deblock_chroma[1]_c         4817  4057  3601
deblock_chroma[1]_neon      1249  716   817   (before)
deblock_chroma[1]_neon      1249  766   845   (after)

deblock_h_chroma_420_c      3699  3275  2830
deblock_h_chroma_420_neon   2068  1414  1400  (before)
deblock_h_chroma_420_neon   1838  1355  1291  (after)

8 years agoaarch64: Remove leftover commented out code
Martin Storsjö [Thu, 13 Aug 2015 20:59:27 +0000 (23:59 +0300)]
aarch64: Remove leftover commented out code

8 years agoaarch64: Simplify the decimate_score functions
Martin Storsjö [Thu, 13 Aug 2015 20:59:26 +0000 (23:59 +0300)]
aarch64: Simplify the decimate_score functions

After doing a left shift by the number of bits returned by clz,
only bits set to zero can be shifted out, so if the register
was nonzero to start with (which is checked), it can't become
zero here.

8 years agoarm: Use aligned loads in x264_coeff_last15_neon
Martin Storsjö [Thu, 13 Aug 2015 20:59:25 +0000 (23:59 +0300)]
arm: Use aligned loads in x264_coeff_last15_neon

After subtracting 2, the pointer will be aligned.

checkasm timing      Cortex-A7    A8    A9
coeff_last15_c              423   375   230
coeff_last15_neon           350   420   404  (before)
coeff_last15_neon           350   400   394  (after)

8 years agoarm: Simplify x264_predict_8x8c_p_neon
Martin Storsjö [Thu, 13 Aug 2015 20:59:24 +0000 (23:59 +0300)]
arm: Simplify x264_predict_8x8c_p_neon

This gets rid of a few unnecessary (and confusing) steps in
calculating the increment to i00.

checkasm timing      Cortex-A7    A8    A9
intra_predict_8x8c_p_c      5525  4732  4755
intra_predict_8x8c_p_neon   1719  1140  1262  (before)
intra_predict_8x8c_p_neon   1663  1142  1255  (after)

8 years agolavf: Use the prefixed name for pixel format enum
Vittorio Giovara [Tue, 15 Sep 2015 13:40:14 +0000 (15:40 +0200)]
lavf: Use the prefixed name for pixel format enum

8 years agoaarch64: fix x264_mbtree_propagate_cost_neon
Janne Grunau [Wed, 2 Sep 2015 22:21:58 +0000 (00:21 +0200)]
aarch64: fix x264_mbtree_propagate_cost_neon

The branch conditon caused the loop to execute one time more than
intended. Detected by a memory corruption on arm with the 1 to 1 port of
the function.

8 years agoaarch64: Fix integral_init4/8h_neon
Martin Storsjö [Thu, 13 Aug 2015 20:59:22 +0000 (23:59 +0300)]
aarch64: Fix integral_init4/8h_neon

The stride is the number of uint16_t elements and thus needs
to be shifted.

This issue had slipped unnoticed since checkasm didn't actually
verify the output of these functions.

8 years agox86: Fix integral_init4/8h_avx2
Henrik Gramner [Thu, 27 Aug 2015 17:53:00 +0000 (19:53 +0200)]
x86: Fix integral_init4/8h_avx2

The AVX2 implementation was using the wrong offsets. It went undetected due to
the checkasm test being incorrect.

8 years agoSimplify inclusion of x264.h in C++ projects
Mark Webster [Wed, 5 Aug 2015 03:28:17 +0000 (04:28 +0100)]
Simplify inclusion of x264.h in C++ projects

Name all structs to support forward declarations.
Add a conditional extern "C" wrapper in x264.h itself instead of having to
specify it in every location where it's included.

8 years agocheckasm: Properly save rdx/edx in checkasm_call() on x86
Henrik Gramner [Sun, 16 Aug 2015 19:59:26 +0000 (21:59 +0200)]
checkasm: Properly save rdx/edx in checkasm_call() on x86

If the return value doesn't fit in a single register rdx/edx can in some
cases be used in addition to rax/eax.

Doesn't affect any of the existing checkasm tests but it's more correct
behavior and it might be useful in the future.

8 years agox86: Enable SSE2 by default on x86-32
Henrik Gramner [Tue, 11 Aug 2015 15:19:35 +0000 (17:19 +0200)]
x86: Enable SSE2 by default on x86-32

It makes more sense to tune the defaults to benefit the vast majority of users.

Anyone still using a Pentium III for video encoding is of course free to
explicitly set different flags when compiling.

8 years agomsvs/icl: Improve default CFLAGS
Henrik Gramner [Mon, 10 Aug 2015 20:30:21 +0000 (22:30 +0200)]
msvs/icl: Improve default CFLAGS

Use -fp:fast as a substitute for -ffast-math.
Increase warning level from -W0 to -W1 (the default setting).
Disable -GS (stack cookies) on MSVS. It's disabled by default on ICL.

8 years agoUse a relative $SRCPATH for out-of-tree builds
Henrik Gramner [Wed, 12 Aug 2015 20:23:31 +0000 (22:23 +0200)]
Use a relative $SRCPATH for out-of-tree builds

Fixes out-of-tree MSVS builds on Cygwin.

8 years agocygwin: Enable MSVS support
Henrik Gramner [Sat, 8 Aug 2015 20:26:38 +0000 (22:26 +0200)]
cygwin: Enable MSVS support

`cl -showIncludes` creates absolute Windows paths for some files, attempt
to convert those to Unix paths.

Use relative paths for dependencies located in or below the working directory
in order to mimic the behavior of gcc and to make the paths more readable.

Make the dependency generation script a bit more robust in general.

8 years agocltostr.sh: Minor fixes
Henrik Gramner [Sat, 8 Aug 2015 16:34:21 +0000 (18:34 +0200)]
cltostr.sh: Minor fixes

8 years agoSimplify version.sh
Henrik Gramner [Sat, 8 Aug 2015 10:21:54 +0000 (12:21 +0200)]
Simplify version.sh

Also remove some non-POSIX syntax and improve robustness.

As a bonus the script now runs about 2-3 times faster.

`git rev-list --count` could be used to simplify things even further,
but that functionality was added in git 1.7.2 so keep `wc -l` for now
to maintain compatibility with older git versions.

8 years agomsvs: Fix cl detection in non-English environments
장영훈 [Fri, 7 Aug 2015 05:43:24 +0000 (14:43 +0900)]
msvs: Fix cl detection in non-English environments

8 years agox86inc: Sync minor changes from ffmpeg/libav
Henrik Gramner [Mon, 3 Aug 2015 19:05:11 +0000 (21:05 +0200)]
x86inc: Sync minor changes from ffmpeg/libav

8 years agomatroska: Add comments for the remaining element names
Henrik Gramner [Wed, 29 Jul 2015 17:30:52 +0000 (19:30 +0200)]
matroska: Add comments for the remaining element names

8 years agoSilence various static analyzer warnings
Henrik Gramner [Wed, 29 Jul 2015 17:30:41 +0000 (19:30 +0200)]
Silence various static analyzer warnings

Those are false positives, but it doesn't hurt to get rid of them.

8 years agomingw: Enable the tsaware linker flag
Henrik Gramner [Sun, 26 Jul 2015 21:13:29 +0000 (23:13 +0200)]
mingw: Enable the tsaware linker flag

Avoids an irrelevant compatibility layer in Terminal Services environments.

https://msdn.microsoft.com/en-us/library/cc834995.aspx

8 years agomsvs: Don't redefine snprintf for VS2015
Henrik Gramner [Sun, 26 Jul 2015 21:13:26 +0000 (23:13 +0200)]
msvs: Don't redefine snprintf for VS2015

Visual Studio 2015 has a proper snprintf implementation.

8 years agomsvs: Prefer link.exe from the same directory as cl.exe
Henrik Gramner [Sun, 26 Jul 2015 21:13:19 +0000 (23:13 +0200)]
msvs: Prefer link.exe from the same directory as cl.exe

/usr/bin/link from coreutils may be located before the MSVS linker in $PATH
which causes linking to fail due to using the wrong binary.

8 years agoframe_dump: check fseek() return value
Henrik Gramner [Sun, 26 Jul 2015 22:10:00 +0000 (00:10 +0200)]
frame_dump: check fseek() return value

8 years agox264_vfprintf: use va_copy
Henrik Gramner [Sun, 26 Jul 2015 22:08:38 +0000 (00:08 +0200)]
x264_vfprintf: use va_copy

It's undefined behavior to use the same va_list twice.

This most likely didn't cause any issues in practice since the string would
have to be larger than 4 KiB to trigger the fallback path.

Use workaround for ICL as it doesn't define va_copy even for C99.

8 years agoparam_parse: Fix framerate rounding issues
Henrik Gramner [Sun, 26 Jul 2015 22:08:31 +0000 (00:08 +0200)]
param_parse: Fix framerate rounding issues

8 years agoaarch64: Remove broken CFLAGS in configure
Marcin Juszkiewicz [Mon, 1 Jun 2015 09:24:45 +0000 (11:24 +0200)]
aarch64: Remove broken CFLAGS in configure

GCC doesn't have an "-arch" switch, but works when that entire line is removed.

8 years agoppc: Add little-endian PowerPC support
Rong Yan [Mon, 20 Jul 2015 08:34:20 +0000 (03:34 -0500)]
ppc: Add little-endian PowerPC support

8 years agomips: MSA quant optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:46 +0000 (17:48 +0530)]
mips: MSA quant optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: MSA predict optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:45 +0000 (17:48 +0530)]
mips: MSA predict optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: MSA pixel optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:44 +0000 (17:48 +0530)]
mips: MSA pixel optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: MSA deblock optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:43 +0000 (17:48 +0530)]
mips: MSA deblock optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: MSA dct optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:42 +0000 (17:48 +0530)]
mips: MSA dct optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: MSA mc optimizations
Rishikesh More [Thu, 18 Jun 2015 12:18:40 +0000 (17:48 +0530)]
mips: MSA mc optimizations

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: Common MSA macros
Rishikesh More [Thu, 18 Jun 2015 12:18:38 +0000 (17:48 +0530)]
mips: Common MSA macros

Add macros for load/store, slide, shift, transpose and basic arithmetic
operations required by subsequent patches.

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: Add MSA support to checkasm
Rishikesh More [Tue, 12 May 2015 14:08:09 +0000 (19:38 +0530)]
mips: Add MSA support to checkasm

Signed-off-by: Rishikesh More <rishikesh.more@imgtec.com>
8 years agomips: Initial MSA support
Kaustubh Raste [Fri, 17 Apr 2015 12:08:58 +0000 (17:38 +0530)]
mips: Initial MSA support

MSA is the MIPS SIMD Architecture.

Add X264_CPU_MSA define.
Update configure to detect MIPS platform and set flags.
CPU-specific gcc options are expected through --extra-cflags.

Sample command line for mips32r5:
    ./configure --host=mipsel-linux-gnu --cross-prefix=<TOOLCHAIN>/mips-mti-linux-gnu-
    --extra-cflags="-EL -mips32r5 -msched-weight -mload-store-pairs"

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
8 years agoLimit autodetection of threads number according to the source height
Anton Mitrofanov [Thu, 16 Jul 2015 21:22:29 +0000 (00:22 +0300)]
Limit autodetection of threads number according to the source height

8 years agoFine-tune of frame's size predictors at ratecontrol start
Anton Mitrofanov [Thu, 16 Jul 2015 16:04:59 +0000 (19:04 +0300)]
Fine-tune of frame's size predictors at ratecontrol start

This is attempt to improve VBV at start of video with a lot of threads which
delay feedback for predictors.

8 years agoUse forced frame types in slicetype analysis
Anton Mitrofanov [Thu, 16 Jul 2015 13:15:56 +0000 (16:15 +0300)]
Use forced frame types in slicetype analysis

This should improve MBTree and VBV when a lot of forced frame types are used.

8 years agox86: SSSE3 and AVX2 implementations of plane_copy_swap
Henrik Gramner [Mon, 1 Dec 2014 21:05:42 +0000 (22:05 +0100)]
x86: SSSE3 and AVX2 implementations of plane_copy_swap

For NV21 input.

8 years agoNV21 input support
Yu Xiaolei [Fri, 6 Jun 2014 08:05:27 +0000 (16:05 +0800)]
NV21 input support

Eliminates an extra copy when encoding Android camera preview images.

Checkasm test by Janne Grunau.
ARM assembly with improvements from Janne Grunau.

8 years agodeblock: Write combining
Henrik Gramner [Tue, 23 Jun 2015 15:00:47 +0000 (17:00 +0200)]
deblock: Write combining

8 years agoGet rid of some tabs and trailing whitespaces
Henrik Gramner [Tue, 23 Jun 2015 12:59:59 +0000 (14:59 +0200)]
Get rid of some tabs and trailing whitespaces

8 years agox86: Experimental nasm support
Henrik Gramner [Sat, 23 May 2015 17:44:16 +0000 (19:44 +0200)]
x86: Experimental nasm support

Enables the use of nasm as an alternative to yasm.

Note that nasm cannot assemble x264 with PIC enabled since it currently doesn't
support [symbol-$$] addressing which is used extensively by x264's PIC code.
This includes all 64-bit Windows and 64-bit OS X builds, even non-shared.

For the above reason nasm is currently intentionally not auto-detected, instead
the assembler must be explicitly specified using "AS=nasm ./configure".

Also drop -O2 from ASFLAGS since it's simply ignored anyway.

8 years agox86inc: Prevent warnings when using `struc` and `endstruc`
Timothy Gu [Tue, 26 May 2015 17:12:42 +0000 (19:12 +0200)]
x86inc: Prevent warnings when using `struc` and `endstruc`

struc and endstruc attempts to revert to the previous section state set by
the SECTION macro.

Use the primitive [SECTION] directive instead of the SECTION macro for the
.note.GNU-stack section to prevent it from being emitted again during endstruc.

8 years agox86inc: Drop SECTION_TEXT macro
Henrik Gramner [Wed, 27 May 2015 19:38:14 +0000 (21:38 +0200)]
x86inc: Drop SECTION_TEXT macro

The .text section is already 16-byte aligned by default on all supported
platforms so `SECTION_TEXT` isn't any different from `SECTION .text`.

8 years agox86inc: Disable vpbroadcastq workaround in newer yasm versions
Henrik Gramner [Sat, 23 May 2015 11:38:05 +0000 (13:38 +0200)]
x86inc: Disable vpbroadcastq workaround in newer yasm versions

The bug was fixed in 1.3.0, so only perform the workaround in earlier versions.

8 years agoPrefer Unicode versions of Windows API calls
Henrik Gramner [Sun, 24 May 2015 20:57:00 +0000 (22:57 +0200)]
Prefer Unicode versions of Windows API calls

Just for consistency, doesn't affect behavior.

8 years agoGet rid of fPIC warnings when compiling a shared library on Windows
Henrik Gramner [Sun, 24 May 2015 21:21:20 +0000 (23:21 +0200)]
Get rid of fPIC warnings when compiling a shared library on Windows

PIC is always enabled when compiling for Windows so gcc complains when using
-fPIC since it doesn't do anything.

8 years agomatroska: Write the correct DocTypeVersion when using frame-packing
Henrik Gramner [Sat, 25 Jul 2015 20:42:59 +0000 (22:42 +0200)]
matroska: Write the correct DocTypeVersion when using frame-packing

The StereoMode element is only valid with DocTypeVersion 3 or higher.

8 years agodump_yuv: Fix file handle leak
Anton Mitrofanov [Fri, 24 Jul 2015 21:21:52 +0000 (00:21 +0300)]
dump_yuv: Fix file handle leak

8 years agomp4: Fix file handle leak
Anton Mitrofanov [Fri, 24 Jul 2015 21:20:47 +0000 (00:20 +0300)]
mp4: Fix file handle leak

8 years agoflv: Check fseek() and fwrite() return values
Henrik Gramner [Tue, 23 Jun 2015 22:40:45 +0000 (00:40 +0200)]
flv: Check fseek() and fwrite() return values

8 years agoflv: Fix memory and file handle leaks
Henrik Gramner [Tue, 23 Jun 2015 22:22:56 +0000 (00:22 +0200)]
flv: Fix memory and file handle leaks

8 years agoavs: Fix file handle leak
Henrik Gramner [Tue, 23 Jun 2015 23:23:35 +0000 (01:23 +0200)]
avs: Fix file handle leak

8 years agomatroska: Fix memory leak
Henrik Gramner [Tue, 23 Jun 2015 11:38:02 +0000 (13:38 +0200)]
matroska: Fix memory leak

8 years agordo: Fix potential CAVLC overflow issues
Henrik Gramner [Tue, 23 Jun 2015 11:24:29 +0000 (13:24 +0200)]
rdo: Fix potential CAVLC overflow issues

8 years agoslurp_file: Various minor bug fixes
Henrik Gramner [Tue, 23 Jun 2015 20:08:35 +0000 (22:08 +0200)]
slurp_file: Various minor bug fixes

 * Fix unsigned <= 0 check.
 * Add additional size sanity check on 32-bit systems.
 * Don't read uninitialized data if fread() fails.

8 years agoparam_parse: Check strdup() return value
Henrik Gramner [Tue, 23 Jun 2015 20:47:53 +0000 (22:47 +0200)]
param_parse: Check strdup() return value

8 years agoparam_parse: Fix memory leak
Henrik Gramner [Tue, 23 Jun 2015 13:38:16 +0000 (15:38 +0200)]
param_parse: Fix memory leak

8 years agoAdd FreeBSD's stdint.h header guard to allowed list
Anton Mitrofanov [Fri, 19 Jun 2015 13:01:12 +0000 (16:01 +0300)]
Add FreeBSD's stdint.h header guard to allowed list

Patch written by Koop Mast <kwm@FreeBSD.org>

8 years agox86: Prevent overread of src in plane_copy_interleave
Henrik Gramner [Fri, 22 May 2015 17:23:33 +0000 (19:23 +0200)]
x86: Prevent overread of src in plane_copy_interleave

Could only occur in 4:2:2 with height == 1.

Also enable asm for inputs with different U/V strides as long as the strides
have identical signs.

8 years agocheckasm: Fix incorrect memcmp size for ARM architecture
Anton Mitrofanov [Wed, 20 May 2015 20:10:20 +0000 (23:10 +0300)]
checkasm: Fix incorrect memcmp size for ARM architecture

8 years agoFix possible use of uninitialized MVs in lookahead analysis for B-frames
Anton Mitrofanov [Sun, 26 Apr 2015 17:51:05 +0000 (20:51 +0300)]
Fix possible use of uninitialized MVs in lookahead analysis for B-frames

9 years agoCatch incorrect usage of libx264 API for delayed frames flushing
Anton Mitrofanov [Tue, 21 Apr 2015 20:08:19 +0000 (23:08 +0300)]
Catch incorrect usage of libx264 API for delayed frames flushing

9 years agoFix detection of system libx264 configuration
Anton Mitrofanov [Sat, 7 Mar 2015 20:00:09 +0000 (23:00 +0300)]
Fix detection of system libx264 configuration

9 years agoCosmetic changes
Anton Mitrofanov [Mon, 23 Feb 2015 11:23:18 +0000 (14:23 +0300)]
Cosmetic changes

9 years agoUpdate configure for auto detection of system libx264 configuration
Anton Mitrofanov [Tue, 30 Dec 2014 23:15:05 +0000 (02:15 +0300)]
Update configure for auto detection of system libx264 configuration

9 years agoAdd tile format frame packing value
Anton Mitrofanov [Tue, 3 Feb 2015 11:51:28 +0000 (14:51 +0300)]
Add tile format frame packing value

Defined in 2014-02 edition.

9 years agoStricter validation of crop-rect values
Anton Mitrofanov [Tue, 3 Feb 2015 10:39:14 +0000 (13:39 +0300)]
Stricter validation of crop-rect values

9 years agoAdd mono frame packing value
Vittorio Giovara [Tue, 20 Jan 2015 16:15:56 +0000 (16:15 +0000)]
Add mono frame packing value

Defined in 2013-04 edition.

9 years agoValidate frame packing value instead of clipping
Vittorio Giovara [Tue, 20 Jan 2015 15:57:41 +0000 (15:57 +0000)]
Validate frame packing value instead of clipping

9 years agox86inc: Correctly warn on use of SSE2 instructions in SSE functions
Christophe Gisquet [Tue, 3 Feb 2015 19:40:41 +0000 (20:40 +0100)]
x86inc: Correctly warn on use of SSE2 instructions in SSE functions

SSE2 instructions that are XMM-implementations of pre-existing MMX/MMX2
instructions did not issue warnings when used in SSE functions. Handle
it by also checking the register type when such instructions are used.

9 years agox86inc: Fix instantiation of YMM registers
Christophe Gisquet [Tue, 3 Feb 2015 17:02:30 +0000 (18:02 +0100)]
x86inc: Fix instantiation of YMM registers

9 years agomatroska: Correctly write display width and height in stereo mode
Vittorio Giovara [Tue, 20 Jan 2015 16:28:54 +0000 (16:28 +0000)]
matroska: Correctly write display width and height in stereo mode

According to the specifications, when stereo mode is set, these values
represent the single view size.

9 years agoUse POC type 0 for AVC-Intra
Kieran Kunhya [Tue, 20 Jan 2015 15:38:00 +0000 (09:38 -0600)]
Use POC type 0 for AVC-Intra

Based on a patch from Capella Systems

9 years agoFix ARCH variable name conflict with BSD ports (bsd.port.mk) read-only variable
Anton Mitrofanov [Sat, 3 Jan 2015 12:46:19 +0000 (15:46 +0300)]
Fix ARCH variable name conflict with BSD ports (bsd.port.mk) read-only variable

9 years agoFix negative percentages in final stats output
Anton Mitrofanov [Sat, 27 Dec 2014 17:35:39 +0000 (20:35 +0300)]
Fix negative percentages in final stats output

They were caused by integer overflow when encoding long UHD video.

9 years agoBump dates to 2015
Anton Mitrofanov [Sat, 3 Jan 2015 20:35:23 +0000 (23:35 +0300)]
Bump dates to 2015

9 years agox86: Update intel compiler cpu dispatcher override for new versions of ICC/ICL
Anton Mitrofanov [Mon, 15 Dec 2014 15:49:23 +0000 (18:49 +0300)]
x86: Update intel compiler cpu dispatcher override for new versions of ICC/ICL

9 years agoNew AQ mode: auto-variance AQ with bias to dark scenes
Anton Mitrofanov [Tue, 6 Sep 2011 17:53:29 +0000 (21:53 +0400)]
New AQ mode: auto-variance AQ with bias to dark scenes

Also known as --aq-mode 3 or auto-variance AQ modification.

9 years agoImprove HRD conformance
Anton Mitrofanov [Tue, 28 Aug 2012 23:02:27 +0000 (03:02 +0400)]
Improve HRD conformance

9 years agox86: SSE and AVX implementations of plane_copy
Henrik Gramner [Fri, 28 Nov 2014 22:24:56 +0000 (23:24 +0100)]
x86: SSE and AVX implementations of plane_copy

Also remove the MMX2 implementation and fix src overread for height == 1.

9 years agoUpdate to the latest version of gas-preprocessor.pl from http://git.libav.org/?p...
Anton Mitrofanov [Mon, 29 Sep 2014 19:26:19 +0000 (23:26 +0400)]
Update to the latest version of gas-preprocessor.pl from http://git.libav.org/?p=gas-preprocessor.git

Contributions by Janne Grunau, Martin Storsjo, Mans Rullgard, David Conrad, Martin Aumuller and others

9 years agoaarch64: cabac_encode_{decision,bypass,terminal}_asm
Janne Grunau [Tue, 18 Nov 2014 23:33:55 +0000 (00:33 +0100)]
aarch64: cabac_encode_{decision,bypass,terminal}_asm

benchmarks on a Nexus 9 (nvidia denver):
101.3 cycles in x264_cabac_encode_decision_c,   67105369 runs, 3495 skips
 97.3 cycles in x264_cabac_encode_decision_asm, 67105493 runs, 3371 skips
132.8 cycles in x264_cabac_encode_terminal_c,    1046950 runs, 1626 skips
116.1 cycles in x264_cabac_encode_terminal_asm,  1048424 runs, 152 skips
 92.4 cycles in x264_cabac_encode_bypass_c,     16776192 runs, 1024 skips
 89.6 cycles in x264_cabac_encode_bypass_asm,   16776453 runs, 763 skips

Cycle counts are not as stable as one would like. The dynamic code
optimisation seems to produce different results for small chnages in a
binary. Repeated runs with the same binary produce stable results
though (ignoring the first run).

9 years agocheckasm: add cycle counter read for aarch64
Janne Grunau [Thu, 6 Nov 2014 08:20:17 +0000 (09:20 +0100)]
checkasm: add cycle counter read for aarch64

Needs kernel support since user space access to the cycle counter is not
allowed on all available AArch64 systems (Android 5 and iOS).

9 years agoaarch64: nal_escape_neon
Janne Grunau [Wed, 5 Nov 2014 10:35:13 +0000 (11:35 +0100)]
aarch64: nal_escape_neon

3-4 times faster.

9 years agoaarch64: {plane_copy,memcpy_aligned,memzero_aligned}_neon
Janne Grunau [Fri, 31 Oct 2014 13:49:04 +0000 (14:49 +0100)]
aarch64: {plane_copy,memcpy_aligned,memzero_aligned}_neon

2-3 times faster than C.

9 years agoaarch64: x264_mbtree_propagate_{cost,list}_neon
Janne Grunau [Wed, 29 Oct 2014 17:17:48 +0000 (18:17 +0100)]
aarch64: x264_mbtree_propagate_{cost,list}_neon

x264_mbtree_propagate_cost_neon is ~7 times faster.
x264_mbtree_propagate_list_neon is 33% faster.

9 years agoaarch64: x264_denoise_dct_neon
Janne Grunau [Tue, 21 Oct 2014 13:18:49 +0000 (15:18 +0200)]
aarch64: x264_denoise_dct_neon

3.5 times faster.

9 years agoaarch64: x264_coeff_level_run{4,8,15,16}
Janne Grunau [Mon, 20 Oct 2014 11:12:14 +0000 (13:12 +0200)]
aarch64: x264_coeff_level_run{4,8,15,16}

All functions ~33% faster.

9 years agoaarch64: NEON asm for intra luma deblocking
Janne Grunau [Tue, 14 Oct 2014 17:20:52 +0000 (19:20 +0200)]
aarch64: NEON asm for intra luma deblocking

deblock_luma_intra[0]_neon is 2 times fastes,
deblock_luma_intra[1]_neon is ~4 times faster.

9 years agoaarch64: x264_deblock_h_chroma_422_neon
Janne Grunau [Mon, 13 Oct 2014 15:29:22 +0000 (17:29 +0200)]
aarch64: x264_deblock_h_chroma_422_neon

deblock_h_chroma_422 2.5 times faster

9 years agoaarch64: x264_deblock_h_chroma_mbaff_neon
Janne Grunau [Mon, 13 Oct 2014 10:43:50 +0000 (12:43 +0200)]
aarch64: x264_deblock_h_chroma_mbaff_neon

deblock_chroma_420_mbaff_neon  2 times faster

9 years agoaarch64: NEON asm for intra chroma deblocking
Janne Grunau [Fri, 10 Oct 2014 08:29:15 +0000 (10:29 +0200)]
aarch64: NEON asm for intra chroma deblocking

deblock_h_chroma_420_intra, deblock_h_chroma_422_intra and
x264_deblock_h_chroma_intra_mbaff_neon are ~3 times faster.
deblock_chroma_intra[1] is ~4 times faster than C.

9 years agoaarch64: add myself as author to aarch64/mc.h
Janne Grunau [Tue, 2 Sep 2014 08:27:22 +0000 (10:27 +0200)]
aarch64: add myself as author to aarch64/mc.h