This commit enabled assembly code with intel AVX512 VNNI and added unit test for sobel filter
sobel_c: 4537
sobel_avx512icl 2136
Signed-off-by: bwang30 <bin.wang@intel.com>
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
The encoder is fixed point, and uses an MDCT only for analysis. Due
to the slightly different rounding, the encoder makes a different
decision, so the tests have to be adjusted as well.
This patch replaces the transform used in AAC with lavu/tx and removes
the limitation on only being able to decode 960-sample files
with the float decoder.
This commit also removes a whole bunch of unnecessary and slow
lifting steps the decoder did to compensate for the poor accuracy
of the old integer transformation code.
Overall float decoder speedup on Zen 3 for 64kbps: 32%
Fixes ticket #128.
The SVQ1 interframe mean VLC symbols -128 and 128 are incorrectly swapped
in our SVQ1 implementation, resulting in visible artifacts for some videos.
This patch unswaps the order of these two symbols.
The most noticable example of the artiacts caused by this error can be observed in
https://trac.ffmpeg.org/attachment/ticket/128/svq1_set.7z '352_288_k_50.mov'.
The artifacts are not observed when using the reference decoder
(QuickTime 7.7.9 x86 binary).
As a result of this patch, the reference data for the fate-svq1 test
($SAMPLES/svq1/marymary-shackles.mov) must be modified. For this file, our
decoder output is now bitwise identical to the reference decoder. I have
tested patch with various other samples and they are all now bitwise identical.
This enables overriding the rotation as well as horizontal/vertical
flip state of a specific video stream on the input side.
Additionally, switch the singular test that was utilizing the rotation
metadata to instead override the input display rotation, thus leading
to the same result.
There is no MMX code for (add|put|put_signed)_pixels_clamped
since commit bfb28b5ce8, so use
declare_func instead of declare_func_emms() to also test that
we are not in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for diff_bytes since commit
230ea38de1, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for add_int16 since commit
4b6ffc2880, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for llviddsp after commit
fed07efcde, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for pixblockdsp after commit
92b5800277, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for audiodsp after commit
3d716d38ab, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for blockdsp after commit
ee551a21dd, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for vc1_inv_trans_8x8 or
vc1_unescape_buffer, so use declare_func instead of
declare_func_emms() to also test that we are not in MMX
mode after return.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no MMX code for loop filters since commit
6a551f1405, so use declare_func
instead of declare_func_emms() to also test that we are not
in MMX mode after return.
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The data in SGI images is stored planar, so exporting
it via planar pixel formats is natural.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Unfortunately, it is common, and will remain so, that the Bit
manipulations are not enabled at compilation time. This is an official
policy for Debian ports in general (though they do not support RISC-V
officially as of yet) to stick to the minimal target baseline, which
does not include the B extension or even its Zbb subset.
For inline helpers (CPOP, REV8), compiler builtins (CTZ, CLZ) or
even plain C code (MIN, MAX, MINU, MAXU), run-time detection seems
impractical. But at least it can work for the byte-swap DSP functions.
This check is intended to be avoid buffer overflows,
yet there are four problems with it:
1. It has an in-built off-by-one error: len == out_end - out
is perfectly fine and nothing to worry about.
This off-by-one error led to the pixel in the lower-right corner
not being set properly for the back frame of the sample from
the rl2 FATE-test. This pixel is copied to every frame which
is the reason for the update to the reference file of said test.
With this patch, the output of the decoder matches the output
as captured from the reference decoder* (apart from the fact
that said reference somehow lacks the top part of the frame
(copied over from the background frame)).
2. Given that the stride of the buffer may be different
from the width of the video (despite one pixel taking one byte),
there is a second check lateron making the first check redundant
(if one returns immediately; a simple break at the second check
is not sufficient, because it only exits the inner loop).
3. The check is based around the assumption of the stride being
positive (it has this in common with the other check which
will be fixed in a future commit).
4. Even after fixing the off-by-one error, the check in
question is still triggered by all the non-background frames
in the FATE sample as well as by A1100100.RL2. In all these
cases, they use len == 255 and val == 128. For videos with
background frame this just means "copy from the background
frame", which would be done anyway lateron.* Yet for videos
without it copying it is necessary to avoid leaving
uninitialized parts in the video.
*: Available in https://samples.mplayerhq.hu/game-formats/voyeur-rl2/
**: Due to this, the code that copies the rest from the
back frame is no longer executed for any of the samples
available on the sample server. Given that these are only
the files from the demo version of this game, I don't know
whether this code is executed for any file in existence or not.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
RVV defines a total of 12 different extensions, including:
- 5 different instruction subsets:
- Zve32x: 8-, 16- and 32-bit integers,
- Zve32f: Zve32x plus single precision floats,
- Zve64x: Zve32x plus 64-bit integers,
- Zve64f: Zve32f plus Zve64x,
- Zve64d: Zve64f plus double precision floats.
- 6 different vector lengths:
- Zvl32b (embedded only),
- Zvl64b (embedded only),
- Zvl128b,
- Zvl256b,
- Zvl512b,
- Zvl1024b,
- and the V extension proper: equivalent to Zve64f and Zvl128b.
In total, there are 6 different possible sets of supported instructions
(including the empty set), but for convenience we allocate one bit for
each type sets: up-to-32-bit ints (RVV_I32), floats (RVV_F32),
64-bit ints (RVV_I64) and doubles (RVV_F64).
Whence the vector size is needed, it can be retrieved by reading the
unprivileged read-only vlenb CSR. This should probably be a separate
helper macro if needed at a later point.
This introduces compile-time and run-time CPU detection on RISC-V. In
practice, I doubt that FFmpeg will ever see a RISC-V CPU without all of
I, F and D extensions, and if it does, it probably won't have run-time
detection. So the flags are essentially always set.
But as things stand, checkasm wants them that way. Compare the ARMV8
flag on AArch64. We are nowhere near running short on CPU flag bits.
This also tests writing slice data in the unaligned mode
(some of these files use CAVLC) as well as updating
side data as well as parsing ISOBMFF avcc extradata.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
~4x faster than the C version.
The shuffles in the 15pt dim1 are seriously expensive. Not happy with it,
but I'm contempt.
Can be easily converted to pure AVX by removing all vpermpd/vpermps
instructions.
Old one was written with the assumption only even inputs would be given.
This very messy replacement supports even and odd inputs, and supports
AVX2 for extra speed. The buffers given are usually quite big (4k samples),
so the speedup is worth it.
The new SSE version is still faster than the old inline asm version by 33%.
Also checkasm is provided to make sure this monstrosity works.
This fixes some FATE tests.
