Commit Graph

35 Commits

Author SHA1 Message Date
Falk Hüffner ab35de18cb Don't include config.h from assembly, because the mplayer developers
decided it was a great idea to put some (bogus, of course) C function
prototypes there and it doesn't seem worth bothering working around
that since all we wanted is HAVE_GPROF.

Originally committed as revision 3304 to svn://svn.ffmpeg.org/ffmpeg/trunk
2004-07-10 23:22:47 +00:00
Michael Niedermayer bb198e198a interlaced motion estimation
interlaced mpeg2 encoding
  P & B frames
  rate distored interlaced mb decission
  alternate scantable support
4mv encoding fixes (thats also why the regression tests change)
passing height to most dsp functions
interlaced mpeg4 encoding (no direct mode MBs yet)
various related cleanups
disabled old motion estimaton algorithms (log, full, ...) they will either be fixed or removed

Originally committed as revision 2638 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-12-30 16:07:57 +00:00
Michael Niedermayer 2393e65431 100l (forgot alpha & ppc)
Originally committed as revision 2611 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-12-15 09:21:28 +00:00
Falk Hüffner f7728bf61c Use asms instead of builtins when compiling for generic Alpha. Less
ugly.

Originally committed as revision 2270 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-09-13 10:07:46 +00:00
Alex Beregszaszi 669e5093f6 warning fixes
Originally committed as revision 2082 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-07-25 13:15:16 +00:00
Sam Hocevar 2c094d6338 build error on Alpha patch by (Sam Hocevar <sam at zoy dot org>)
Originally committed as revision 1996 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-06-29 00:23:30 +00:00
Michael Niedermayer 4fb518c392 the return of the idct with 16bit output by ("Ivan Kalvachev" <ivan at cacad dot com>)
Originally committed as revision 1983 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-06-21 01:11:31 +00:00
Michael Niedermayer b0368839ac MpegEncContext.(i)dct_* -> DspContext.(i)dct_*
bitexact cleanup

Originally committed as revision 1617 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-03-03 14:54:00 +00:00
Michael Niedermayer df081b2f3b 32 vs. 64bit bugfix by (Dave Huang (dahanc))
Originally committed as revision 1596 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-02-22 13:44:16 +00:00
Falk Hüffner 1a02ea65e0 Implement sad8x8 and sad16x16 with pix_abs.
Originally committed as revision 1428 to svn://svn.ffmpeg.org/ffmpeg/trunk
2003-01-10 13:35:15 +00:00
Falk Hüffner 3354b0c9e5 Kludge around compilation failure on Alpha.
Originally committed as revision 1286 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-11-27 01:20:42 +00:00
Zdenek Kabelac af19f78f2f * using DSPContext - so each codec could use its local (sub)set of CPU extension
* these are untested - users/developers please check

Originally committed as revision 1197 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-11-11 10:04:45 +00:00
Falk Hüffner 8b313a4787 Move Alpha optimized IDCT to own file. Based on a patch by Måns
Rullgård <mru@users.sourceforge.net>.

I've left out the idctCol2 part, because W4 has recently been decreed
to be 16383, and also I doubt it will give a noticeable speedup.

Originally committed as revision 1029 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-10-11 23:01:16 +00:00
Falk Hüffner acd2e05c47 * Fix dct_unquantize_h263_axp for changed block_last_index indexing.
* Set qadd to 0 for intra with h263_aic (whatever that may be).

Originally committed as revision 1009 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-10-07 02:03:21 +00:00
Falk Hüffner 6c38306600 Make compilable again after DCT cleanup. It gives wrong results,
though, so I disabled it for now. Anybody knows offhand what might be
the reason?

Originally committed as revision 1005 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-10-05 00:57:42 +00:00
Falk Hüffner e447d7d65f Some nifty stuff for upcoming patches.
Originally committed as revision 1004 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-10-05 00:12:58 +00:00
Falk Hüffner f9bb4bdffc Add Alpha assembly for pix_abs16x16. Optimized for pca56, no large win
on ev6.

Originally committed as revision 979 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-29 15:14:28 +00:00
Falk Hüffner 8c7b533e8b Synthesize pixels16 functions from pixels functions.
Originally committed as revision 942 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-13 23:27:01 +00:00
Falk Hüffner 60286c8a37 Fix for h % 8 == 4 in put_pixels_axp_asm.
Originally committed as revision 940 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-13 15:44:53 +00:00
Falk Hüffner a343d33ed6 Due to version renumbering, Alpha builtins aren't supported in thegcc
3.2 release.

Originally committed as revision 937 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-13 10:56:51 +00:00
Falk Hüffner 1c73a3fbf2 Adapt to new 8/16 table scheme.
Originally committed as revision 935 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-09-13 09:28:45 +00:00
Falk Hüffner 57d2bce188 MVI optimizations for motion estimation.
Originally committed as revision 831 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-29 23:14:51 +00:00
Falk Hüffner e3e98d3c23 Remove support for variable BSIZE and INCR, as sub_pixels_* is no
longer needed.

Originally committed as revision 754 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-13 16:29:11 +00:00
Falk Hüffner f5abd9fd1a * Improve xy2 routines slightly
* Mark MC pointer arguments as restrict

Originally committed as revision 752 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-13 16:17:38 +00:00
Falk Hüffner 0a12d6fdfd Implement clear_blocks_axp.
Originally committed as revision 722 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-05 19:16:26 +00:00
Falk Hüffner 2dbe7ec87c Make dct_unquantize_h263 work on systems without MVI extension.
Originally committed as revision 715 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-03 23:30:32 +00:00
Falk Hüffner e0580f8c68 Update and activate dct_unquantize_h263_mvi. Thanks to Måns Rullgård
for some improvements.

Originally committed as revision 714 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-03 03:01:06 +00:00
Falk Hüffner dde3f77dbc Use updated motion compensation routines.
Originally committed as revision 713 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-03 01:09:44 +00:00
Falk Hüffner 494e409255 Assembly version of put_pixels. This is currently the function that
takes the most time, and it allows for more efficient unaligned access
and better control over memory latencies.

Originally committed as revision 711 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-01 23:47:01 +00:00
Falk Hüffner d13c54cdc0 Add profiling support.
Originally committed as revision 710 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-01 23:02:36 +00:00
Falk Hüffner bb7d4939ba Implement put_pixels_clamped and add_pixels_clamped in Assembler. This
allows better scheduling of the memory accesses, and is portable among
all compilers.

Originally committed as revision 709 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-07-01 04:26:07 +00:00
Falk Hüffner db42e13bfb Ugly hack to make the assembler accept MVI instructions.
Originally committed as revision 705 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-06-24 22:22:39 +00:00
Falk Hüffner b210c96c0c * Fixed constraints
* Support ccc
* Support gcc 3.2 builtins

Originally committed as revision 704 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-06-24 22:21:30 +00:00
Fabrice Bellard ff4ec49e64 license/copyright change
Originally committed as revision 599 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-05-25 22:45:33 +00:00
Nick Kurshev 1e98dffb7a Alpha optimizations by Falk Hueffner <falk.hueffner@student.uni-tuebingen.de>
Originally committed as revision 274 to svn://svn.ffmpeg.org/ffmpeg/trunk
2002-01-20 14:48:02 +00:00