Mostly seems to help in the MC code, which gets a hundred cycles faster.

Originally committed as revision 24463 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24462 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 31777 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Originally committed as revision 31776 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

sequential geometries instead of running all algorithms sequentially for each
geometry.

Originally committed as revision 31775 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Don't clamp best mv unless it's actually used.

Originally committed as revision 24461 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24460 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24459 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24458 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24457 to svn://svn.ffmpeg.org/ffmpeg/trunk

Apparently the official conformance test vectors don't test this feature,
even though libvpx uses it.

Originally committed as revision 24456 to svn://svn.ffmpeg.org/ffmpeg/trunk

Add MMX idct_dc_add4uv function for this case.
~40% faster chroma idct.

Originally committed as revision 24455 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24454 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24453 to svn://svn.ffmpeg.org/ffmpeg/trunk

Take shortcuts based on statistically common situations.
Add 4-at-a-time idct_dc function (mmx and sse2) since rows of 4 DC-only DCT
blocks are common.
TODO: tie this more directly into the MB mode, since the DC-level transform is
only used for non-splitmv blocks?

Originally committed as revision 24452 to svn://svn.ffmpeg.org/ffmpeg/trunk

Don't prefetch reference frames that were used less than 1/32th of the time so
far in the frame.
This helps speed up to ~2% on videos that, in many frames, make near-zero
(but not entirely zero) use of golden and/or alt-refs.
This is a very common property of videos encoded by libvpx.

Originally committed as revision 24451 to svn://svn.ffmpeg.org/ffmpeg/trunk

erroneous message for last frame.

Originally committed as revision 24450 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24449 to svn://svn.ffmpeg.org/ffmpeg/trunk

~0.3% faster overall.

Originally committed as revision 24448 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24447 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24446 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24445 to svn://svn.ffmpeg.org/ffmpeg/trunk

Prefetch all refs (including altref), but only if they've been used so far this
frame.
~2.5% faster overall.

TODO: Do something even smarter, like using how often each ref has been used
so far, so that a couple blocks of a rarely-used ref don't force us to prefetch
it.

Originally committed as revision 24444 to svn://svn.ffmpeg.org/ffmpeg/trunk

rest of FFmpeg.

Originally committed as revision 31772 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Originally committed as revision 24443 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24442 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24441 to svn://svn.ffmpeg.org/ffmpeg/trunk

use the pixdesc API instead.

Originally committed as revision 24440 to svn://svn.ffmpeg.org/ffmpeg/trunk

irrational number.

Originally committed as revision 24439 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24438 to svn://svn.ffmpeg.org/ffmpeg/trunk

CPUs supporting it.

Originally committed as revision 24437 to svn://svn.ffmpeg.org/ffmpeg/trunk

smaller than 1000

Originally committed as revision 31771 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

the output even if the test crashes.

Originally committed as revision 31770 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Making automatic reconfigure work properly in all cases proved more
difficult than it seemed, and people complained about it.  Instead,
print a warning when the configuration appears to be out of date.

Originally committed as revision 24436 to svn://svn.ffmpeg.org/ffmpeg/trunk

Some converters (ie. unscaled rgb24 -> argb) may write some bytes out of
bounds. Ideally the converters should be fixed, but in the meantime we allocate
more memory to prevent heap corruption.

Originally committed as revision 31768 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Originally committed as revision 24435 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24434 to svn://svn.ffmpeg.org/ffmpeg/trunk

PRINT_LIB_VERSION()      -> PRINT_LIB_INFO()
print_all_lib_versions() -> print_all_libs_info()

Originally committed as revision 24433 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 24432 to svn://svn.ffmpeg.org/ffmpeg/trunk

parameter. Allow a pending factorization.

Originally committed as revision 24431 to svn://svn.ffmpeg.org/ffmpeg/trunk