RepoMirrors/ffmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2025-02-24 15:57:00 +00:00
Code Issues Projects Releases Wiki Activity
FFmpeg git repo
101,949 Commits 37 Branches 400 Tags 273 MiB
C 90.3%
Assembly 7.8%
Makefile 1.3%
C++ 0.2%
Objective-C 0.1%
Other 0.1%
870bfe16a1
Go to file
Martin Storsjö 870bfe16a1 aarch64: h264pred: Optimize the inner loop of existing 8 bit functions 
Move the loop counter decrement further from the branch instruction,
this hides the latency of the decrement.

In loops that first load, then store (the horizontal prediction cases),
do the decrement after the load (where the next instruction would
stall a bit anyway, waiting for the result of the load).

In loops that store twice using the same destination register,
also do the decrement between the two stores (as the second store
would need to wait for the updated destination register from the
first instruction).

In loops that store twice to two different destination registers,
do the decrement before both stores, to do it as soon before the
branch as possible.

This gives minor (1-2 cycle) speedups in most cases (modulo measurement
noise), but the horizontal prediction functions get a rather notable
speedup on the Cortex A53.

Before:                     Cortex A53     A72     A73
pred8x8_dc_8_neon:                60.7    46.2    39.2
pred8x8_dc_128_8_neon:            30.7    18.0    14.0
pred8x8_horizontal_8_neon:        42.2    29.2    18.5
pred8x8_left_dc_8_neon:           52.7    36.2    32.2
pred8x8_mad_cow_dc_0l0_8_neon:    48.2    27.7    25.7
pred8x8_mad_cow_dc_0lt_8_neon:    52.5    33.2    34.7
pred8x8_mad_cow_dc_l0t_8_neon:    52.5    31.7    33.2
pred8x8_mad_cow_dc_l00_8_neon:    43.2    27.0    25.5
pred8x8_plane_8_neon:            112.2    86.2    88.2
pred8x8_top_dc_8_neon:            40.7    23.0    21.2
pred8x8_vertical_8_neon:          27.2    15.5    14.0
pred16x16_dc_8_neon:              91.0    73.2    70.5
pred16x16_dc_128_8_neon:          43.0    34.7    30.7
pred16x16_horizontal_8_neon:      86.0    49.7    44.7
pred16x16_left_dc_8_neon:         87.0    67.2    67.5
pred16x16_plane_8_neon:          236.0   175.7   173.0
pred16x16_top_dc_8_neon:          53.2    39.0    41.7
pred16x16_vertical_8_neon:        41.7    29.7    31.0

After:
pred8x8_dc_8_neon:                59.0    46.7    42.5
pred8x8_dc_128_8_neon:            28.2    18.0    14.0
pred8x8_horizontal_8_neon:        34.2    29.2    18.5
pred8x8_left_dc_8_neon:           51.0    38.2    32.7
pred8x8_mad_cow_dc_0l0_8_neon:    46.7    28.2    26.2
pred8x8_mad_cow_dc_0lt_8_neon:    55.2    33.7    37.5
pred8x8_mad_cow_dc_l0t_8_neon:    51.2    31.7    37.2
pred8x8_mad_cow_dc_l00_8_neon:    41.7    27.5    26.0
pred8x8_plane_8_neon:            111.5    86.5    89.5
pred8x8_top_dc_8_neon:            39.0    23.2    21.0
pred8x8_vertical_8_neon:          27.2    16.0    14.0
pred16x16_dc_8_neon:              85.0    70.2    70.5
pred16x16_dc_128_8_neon:          42.0    30.0    30.7
pred16x16_horizontal_8_neon:      66.5    49.5    42.5
pred16x16_left_dc_8_neon:         81.0    66.5    67.5
pred16x16_plane_8_neon:          235.0   175.7   173.0
pred16x16_top_dc_8_neon:          52.0    39.0    41.7
pred16x16_vertical_8_neon:        40.2    33.2    31.0

Despite this, a number of these functions still are slower than
what e.g. GCC 7 generates - this shows the relative speedup of the
neon codepaths over the compiler generated ones:

                           Cortex A53    A72    A73
pred8x8_dc_8_neon:               0.86   0.65   1.04
pred8x8_dc_128_8_neon:           0.59   0.44   0.62
pred8x8_horizontal_8_neon:       1.51   0.58   1.30
pred8x8_left_dc_8_neon:          0.72   0.56   0.89
pred8x8_mad_cow_dc_0l0_8_neon:   0.93   0.93   1.37
pred8x8_mad_cow_dc_0lt_8_neon:   1.37   1.41   1.68
pred8x8_mad_cow_dc_l0t_8_neon:   1.21   1.17   1.32
pred8x8_mad_cow_dc_l00_8_neon:   1.24   1.19   1.60
pred8x8_plane_8_neon:            3.36   3.58   3.76
pred8x8_top_dc_8_neon:           0.97   0.99   1.43
pred8x8_vertical_8_neon:         0.86   0.78   1.18
pred16x16_dc_8_neon:             1.20   1.06   1.49
pred16x16_dc_128_8_neon:         0.83   0.95   0.99
pred16x16_horizontal_8_neon:     1.78   0.96   1.59
pred16x16_left_dc_8_neon:        1.06   0.96   1.32
pred16x16_plane_8_neon:          5.78   6.49   7.19
pred16x16_top_dc_8_neon:         1.48   1.53   1.94
pred16x16_vertical_8_neon:       1.39   1.34   1.98

In particular, on Cortex A72, many of these functions are slower
than the compiler generated code, while they're more beneficial on
e.g. the Cortex A73.

Signed-off-by: Martin Storsjö <martin@martin.st>
2021-04-14 15:23:44 +03:00
compat atomics: Fix the win32 atomic_exchange function 2021-04-04 11:06:08 +03:00
doc doc/muxers: add entries for raw muxers 2021-04-11 12:09:16 +05:30
ffbuild ffbuild/common: Make deletion of templates possible 2021-02-07 09:45:04 +01:00
fftools fftools/ffmpeg_filter: Fix check for mjpeg encoder 2021-04-13 21:58:25 +02:00
libavcodec aarch64: h264pred: Optimize the inner loop of existing 8 bit functions 2021-04-14 15:23:44 +03:00
libavdevice libavdevice/gdigrab: fix capture of windows with non-ASCII titles 2021-04-13 19:34:33 +03:00
libavfilter lavfi/qsvvpp: support async depth 2021-04-11 23:18:20 +08:00
libavformat avformat/dashdec: Also fetch final partial segment 2021-04-14 15:28:04 +08:00
libavresample
libavutil avutil/frame: Return 0 on success in av_frame_ref() 2021-04-05 18:36:51 +02:00
libpostproc Bump minor versions after release branch 2021-03-20 01:02:11 +01:00
libswresample libswresample/audioconvert: Fix undefined NULL + 0 2021-04-01 14:15:20 +02:00
libswscale libswscale/x86/yuv2yuvX: Removes unrolling for mmx and mmxext 2021-04-01 20:47:52 +02:00
presets
tests tests/fate: add tests for PNG side/meta data 2021-04-08 11:03:15 +02:00
tools tools/target_dem_fuzzer: Fix packet leak 2021-04-01 10:10:21 +02:00
.gitattributes
.gitignore
.mailmap mailmap: add entry for myself 2021-03-09 02:09:55 +00:00
.travis.yml
Changelog Changelog: Add new <next> line after 4.4 2021-04-09 06:30:31 +02:00
configure configure: select child muxers for rtp_mpegts 2021-03-21 15:50:23 +05:30
CONTRIBUTING.md
COPYING.GPLv2
COPYING.GPLv3
COPYING.LGPLv2.1
COPYING.LGPLv3
CREDITS
INSTALL.md
LICENSE.md
MAINTAINERS MAINTAINERS: add myself as adpcm maintainer 2021-03-25 12:51:10 +10:00
Makefile avcodec: move core AVCodecContext functions from util.c to a new file 2021-03-19 15:35:35 -03:00
README.md
RELEASE Bump Versions before release/4.4 branch 2021-03-20 01:01:12 +01:00

README.md

FFmpeg README

FFmpeg is a collection of libraries and tools to process multimedia content such as audio, video, subtitles and related metadata.

Libraries

  • libavcodec provides implementation of a wider range of codecs.
  • libavformat implements streaming protocols, container formats and basic I/O access.
  • libavutil includes hashers, decompressors and miscellaneous utility functions.
  • libavfilter provides a mean to alter decoded Audio and Video through chain of filters.
  • libavdevice provides an abstraction to access capture and playback devices.
  • libswresample implements audio mixing and resampling routines.
  • libswscale implements color conversion and scaling routines.

Tools

  • ffmpeg is a command line toolbox to manipulate, convert and stream multimedia content.
  • ffplay is a minimalistic multimedia player.
  • ffprobe is a simple analysis tool to inspect multimedia content.
  • Additional small tools such as aviocat, ismindex and qt-faststart.

Documentation

The offline documentation is available in the doc/ directory.

The online documentation is available in the main website and in the wiki.

Examples

Coding examples are available in the doc/examples directory.

License

FFmpeg codebase is mainly LGPL-licensed with optional components licensed under GPL. Please refer to the LICENSE file for detailed information.

Contributing

Patches should be submitted to the ffmpeg-devel mailing list using git format-patch or git send-email. Github pull requests should be avoided because they are not part of our review process and will be ignored.