The effect is not really deterministic, as it seems to be a combination
on x86_64 of fewer registers used, different jump offsets and, for all
archs, of likely branches.
Speedup is around 15%.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Those macros take a byte number as shift argument, as this argument
differs between MMX and SSE2 instructions.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This fixes a race condition and use of the wrong field, which become shared
instead of per thread during some AVFrame changes.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Sometimes the input buffers get directly used as raw images and
SIMD optimized video/image filters can sometimes read more than 16 bytes
over the end.
a specific example is the AVX 24bpp to yuv code
This also fixes fate-vsynth3-rgb
Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Since error resilience uses AVFrame pointers instead of references it
has to copy NULL pointers too. After a codec flush the last/next frame
pointers in MpegEncContext are NULL and the old pointers remaining in
ERContext are invalid. Fixes a crash in vlc for android thumbnailer.
Reported and debugged by Adrien Maglo <magsoft@videolan.org>.
grayscale is coded as 4 pixels at a time, the encoder lacks support
for the case where width%4 != 0, and will simply encode less data
leaving random data after decoding at the right side
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
We know that the called function (ff_chroma_inter_body_mmxext)
doesn't touch the redzone, and thus will be kept intact - thus,
this doesn't fix any bug per se.
However, valgrind's memcheck tool intentionally assumes that the
redzone is clobbered on every function call and function return
(see a long comment in valgrind/memcheck/mc_main.c). This avoids
false positives in that tool, at the cost of an extra stack pointer
adjustment.
The other alternative would be a valgrind suppression for this issue,
but that's an extra burden for everybody that wants to run libavcodec
within valgrind.
Signed-off-by: Martin Storsjö <martin@martin.st>
The actual predictor value, set by the trellis code, never
was written back into the variable that was written into
the block header. This was accidentally removed in b304244b.
This significantly improves the audio quality of the trellis
case, which was plain broken since b304244b.
Encoding IMA QT with trellis still actually gives a slightly
worse quality than without trellis, since the trellis encoder
doesn't use the exact same way of rounding as in
adpcm_ima_qt_compress_sample and adpcm_ima_qt_expand_nibble.
CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
This very slightly improves compression
Found-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This is probably not the simplest solution but as this is needed for a bugfix,
simplification is left for later.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Fixes a regression since fb3e380 similar to ticket #2661,
reported by fluffrabbit at aol dot com.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
There's an SSE2 version already, and technically the SSE version
on x86_64 was wrong (using pshufd and pshuflw, SSE2 instructions).
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Michael Niedermayer
Christophe Gisquet
Marton Balint
Carl Eugen Hoyos
Diego Biurrun
Derek Buitenhuis
Ronald S. Bultje
Luca Barbato
Janne Grunau
Vittorio Giovara
Martin Storsjö
James Almer