Also adapt some FATE tests to already cover this.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The AC3 encoder used to be a separate library called "Aften", which
got merged into libavcodec (literally, SVN commits and all).
The merge preserved as much features from the library as possible.
The code had two versions - a fixed point version and a floating
point version. FFmpeg had floating point DSP code used by other
codecs, the AC3 decoder including, so the floating-point DSP was
simply replaced with FFmpeg's own functions.
However, FFmpeg had no fixed-point audio code at that point. So
the encoder brought along its own fixed-point DSP functions,
including a fixed-point MDCT.
The fixed-point MDCT itself is trivially just a float MDCT with a
different type and each multiply being a fixed-point multiply.
So over time, it got refactored, and the FFT used for all other codecs
was templated.
Due to design decisions at the time, the fixed-point version of the
encoder operates at 16-bits of precision. Although convenient, this,
even at the time, was inadequate and inefficient. The encoder is noisy,
does not produce output comparable to the float encoder, and even
rings at higher frequencies due to the badly approximated winow function.
Enter MIPS (owned by Imagination Technologies at the time). They wanted
quick fixed-point decoding on their FPUless cores. So they contributed
patches to template the AC3 decoder so it had both a fixed-point
and a floating-point version. They also did the same for the AAC decoder.
They however, used 32-bit samples. Not 16-bits. And we did not have
32-bit fixed-point DSP functions, including an MDCT. But instead of
templating our MDCT to output 3 versions (float, 32-bit fixed and 16-bit fixed),
they simply copy-pasted their own MDCT into ours, and completely
ifdeffed our own MDCT code out if a 32-bit fixed point MDCT was selected.
This is also the status quo nowadays - 2 separate MDCTs, one which
produces floating point and 16-bit fixed point versions, and one
sort-of integrated which produces 32-bit MDCT.
MIPS weren't all that interested in encoding, so they left the encoder
as-is, and they didn't care much about the ifdeffery, mess or quality - it's
not their problem.
So the MDCT/FFT code has always been a thorn in anyone looking to clean up
code's eye.
Backstory over. Internally AC3 operates on 25-bit fixed-point coefficients.
So for the floating point version, the encoder simply runs the float MDCT,
and converts the resulting coefficients to 25-bit fixed-point, as AC3 is inherently
a fixed-point codec. For the fixed-point version, the input is 16-bit samples,
so to maximize precision the frame samples are analyzed and the highest set
bit is detected via ac3_max_msb_abs_int16(), and the coefficients are then
scaled up via ac3_lshift_int16(), so the input for the FFT is always at least 14 bits,
computed in normalize_samples(). After FFT, the coefficients are scaled up to 25 bits.
This patch simply changes the encoder to accept 32-bit samples, reusing
the already well-optimized 32-bit MDCT code, allowing us to clean up and drop
a large part of a very messy code of ours, as well as prepare for the future lavu/tx
conversion. The coefficients are simply scaled down to 25 bits during windowing,
skipping 2 separate scalings, as the hacks to extend precision are simply no longer
necessary. There's no point in running the MDCT always at 32 bits when you're
going to drop 6 bits off anyway, the headroom is plenty, and the MDCT rounds
properly.
This also makes the encoder even slightly more accurate over the float version,
as there's no coefficient conversion step necessary.
SIZE SAVINGS:
ARM32:
HARDCODED TABLES:
BASE - 10709590
DROP DSP - 10702872 - diff: -6.56KiB
DROP MDCT - 10667932 - diff: -34.12KiB - both: -40.68KiB
DROP FFT - 10336652 - diff: -323.52KiB - all: -364.20KiB
SOFTCODED TABLES:
BASE - 9685096
DROP DSP - 9678378 - diff: -6.56KiB
DROP MDCT - 9643466 - diff: -34.09KiB - both: -40.65KiB
DROP FFT - 9573918 - diff: -67.92KiB - all: -108.57KiB
ARM64:
HARDCODED TABLES:
BASE - 14641112
DROP DSP - 14633806 - diff: -7.13KiB
DROP MDCT - 14604812 - diff: -28.31KiB - both: -35.45KiB
DROP FFT - 14286826 - diff: -310.53KiB - all: -345.98KiB
SOFTCODED TABLES:
BASE - 13636238
DROP DSP - 13628932 - diff: -7.13KiB
DROP MDCT - 13599866 - diff: -28.38KiB - both: -35.52KiB
DROP FFT - 13542080 - diff: -56.43KiB - all: -91.95KiB
x86:
HARDCODED TABLES:
BASE - 12367336
DROP DSP - 12354698 - diff: -12.34KiB
DROP MDCT - 12331024 - diff: -23.12KiB - both: -35.46KiB
DROP FFT - 12029788 - diff: -294.18KiB - all: -329.64KiB
SOFTCODED TABLES:
BASE - 11358094
DROP DSP - 11345456 - diff: -12.34KiB
DROP MDCT - 11321742 - diff: -23.16KiB - both: -35.50KiB
DROP FFT - 11276946 - diff: -43.75KiB - all: -79.25KiB
PERFORMANCE (10min random s32le):
ARM32 - before - 39.9x - 0m15.046s
ARM32 - after - 28.2x - 0m21.525s
Speed: -30%
ARM64 - before - 36.1x - 0m16.637s
ARM64 - after - 36.0x - 0m16.727s
Speed: -0.5%
x86 - before - 184x - 0m3.277s
x86 - after - 190x - 0m3.187s
Speed: +3%
Explicitly insert the scale or aresample filter where it would
have been inserted by the negotiation.
Re-enable conversions if it cannot be done easily.
If a conversion is needed in a test, we want to know about it.
If the negotiation changes and makes new conversion necessary,
we want to know about it even more.
This fixes fate with FF_API_REQUEST_CHANNELS disabled.
Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
* commit '3ad678a85b96fc5fecd60e3d3a31ca5ffc89d67f':
fate: Update ac3 test to the new request_channel_layout option
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
x86: ac3dsp: Remove 3dnow version of ff_ac3_extract_exponents
Conflicts:
tests/fate/ac3.mak
Merged-by: Michael Niedermayer <michaelni@gmx.at>
The function requires increasing the fuzz factor for the ac3/eac3 encode
tests and even so makes fate fail. It only provides a slight encoding
speedup for legacy CPUs that do not support SS2. Thus its benefit is not
worth the trouble it creates and fixing it would be a waste of time.
Use a noise range of -0.707 to 0.707 instead of -0.5 to 0.5
Based on patch by: Mathias Rauen <madshi@gmail.com> and commit by Justin Ruggles (04ea5491)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* commit '14f031d7ecfabba0ef02776d4516aa3dcb7c40d8':
dv: use AVStream.index instead of abusing AVStream.id
lavfi: add ashowinfo filter
avcodec: Add a RFC 3389 comfort noise codec
lpc: Add a function for calculating reflection coefficients from samples
lpc: Add a function for calculating reflection coefficients from autocorrelation coefficients
lavr: document upper bound on number of output samples.
lavr: add general API usage doxy
indeo3: remove duplicate capabilities line.
fate: ac3: Add dependencies
Conflicts:
Changelog
doc/filters.texi
libavcodec/Makefile
libavcodec/allcodecs.c
libavcodec/avcodec.h
libavcodec/codec_desc.c
libavcodec/version.h
libavfilter/Makefile
libavfilter/af_ashowinfo.c
libavfilter/allfilters.c
libavfilter/version.h
libavutil/avutil.h
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit '381dc1a5ec0925b281c573457c413ae643567086':
fate: ac3: Place E-AC-3 tests and AC-3 tests in different groups
fate: Add shorthands for acodec PCM and ADPCM tests
avconv: Drop unused function argument from do_video_stats()
cmdutils: Conditionally compile libswscale-related bits
aacenc: Drop some unused function arguments
rtsp: Avoid a cast when calling strtol
nut: support textual data
nutenc: verbosely report unsupported negative pts
Conflicts:
cmdutils.c
ffmpeg.c
libavformat/nut.c
libavformat/nutenc.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
build: sanitize linking of tools and test programs
fate: Refactor setting of environment variables for groups of tests
Conflicts:
tests/fate/audio.mak
tests/fate/real.mak
tests/fate/voice.mak
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
pcm-mpeg: improve log message wording
fate: add missing $(TARGET_PATH) to ac3-fixed-encode
fate: fix md5sum replacement on some systems
avprobe: correctly set the default formatter
lavr: add x86-optimized function for mixing 2 to 1 s16p with q8 coeffs
lavr: add x86-optimized functions for mixing 2 to 1 s16p with float coeffs
lavr: add C functions for mixing 2 to 1 channels with s16p format
avprobe: move formatter functions in the context
Conflicts:
ffprobe.c
libavcodec/pcm-mpeg.c
tests/fate/ac3.mak
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* commit 'f919cc7df6ab844bc12f89fe7bef4fb915a47725':
fate: fix acodec/vsynth tests for make 3.81
pcm_mpeg: fix number of consumed bytes to include the header.
avfilter: include required header file avfilter.h in video.h
x86: Avoid movs on BUTTERFLYPS when in AVX mode
x86: use new schema for ASM macros
fate: convert codec-regression.sh to makefile rules
fate: allow tests to specify unit size for psnr comparison
fate: teach videogen/rotozoom to output a single raw video stream
http: Add support for reusing the http socket for subsequent requests
http: Add support for using persistent connections
Merged-by: Michael Niedermayer <michaelni@gmx.at>
This makes only tests actually using avconv depend on it.
The remaining tests already depend on what they need.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This will allow decoding to md5 and doing a diff comparison to a reference
checksum instead of a fuzzy stddev or oneoff comparison.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Andreas Rheinhardt
Lynne
Nicolas George
James Almer
Paul B Mahol
Andreas Cadhalpun
Michael Niedermayer
Vittorio Giovara
Timothy Gu
Martin Storsjö
Diego Biurrun
Mathias Rauen
Janne Grunau
Mans Rullgard
Justin Ruggles