e168a5567a
DCA: 16-byte-align lfe_fir tables
...
Originally committed as revision 22862 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 20:45:29 +00:00
309d16a4a0
DCA: break out lfe_interpolation_fir() inner loops to a function
...
This enables SIMD optimisations of this function.
Originally committed as revision 22861 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 20:45:25 +00:00
fc7e2d34cf
Fix constness for func[12] parameters in ff_parse_expr() and
...
ff_parse_and_eval_expr().
Change func[12] attributes from "** func" to "* const * func".
This is consistent with the semantics of the provided arrays of
functions, which are not supposed to be changed by the ff_parse_*
functions.
Also fix the GCC compilation warnings:
libavcodec/ratecontrol.c: In function ‘ff_rate_control_init’:
libavcodec/ratecontrol.c:109: warning: passing argument 3 of ‘ff_parse_expr’ discards qualifiers from pointer target type
libavcodec/eval.h:69: note: expected ‘double (**)(void *, double)’ but argument is of type ‘double (* const*)(void *, double)’
Originally committed as revision 22860 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 20:22:55 +00:00
82fdcd449c
Change constness for func[12]_name parameters of ff_parse_expr() and
...
ff_parse_and_eval_expr().
Change attribute from "const char **" to "const char * const *".
The name arrays are not supposed to be changed by the function.
Originally committed as revision 22859 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 19:21:19 +00:00
1235429c27
Rename ff_parse() to ff_parse_expr().
...
The new name is more expressive and fits better in the overall naming
scheme for the revisited eval API.
Originally committed as revision 22858 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 19:21:15 +00:00
6c71d2c135
Remove unnecessary header inclusion directives.
...
Originally committed as revision 22857 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 19:21:08 +00:00
2818bdf22b
Extradata length checks for Huffyuv.
...
Patch by Michael Kaufmann hallo $(name) dash $(surname) ch
Originally committed as revision 22856 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 16:16:43 +00:00
843c7aa8fb
DCA: use FASTDIV in decode_blockcode()
...
Originally committed as revision 22855 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 16:05:10 +00:00
f01210a691
ARM: fix NEON synth_filter_float with hardfp calls
...
Originally committed as revision 22852 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 13:28:59 +00:00
a7fc91468c
Remove useless header inclusion.
...
Originally committed as revision 22851 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 13:21:33 +00:00
766fefe8e2
DCA: simplify lfe_interpolation_fir()
...
This reorders the lfe_fir tables, and drops the mirrored half,
such that the loops in lfe_interpolation_fir() can be simplified.
The new loop structure should be easier to implement with SIMD.
Static data size is reduced by 2kB.
3% faster on Cortex-A8.
Originally committed as revision 22849 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 11:14:55 +00:00
b92d483bac
DCA: use a local variable for loop boundary
...
This prevents gcc reloading the value from memory on each iteration
of the loop.
Originally committed as revision 22848 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 11:14:51 +00:00
0dc7df28dd
DCA: use some type-punning in qmf_32_subbands()
...
Originally committed as revision 22847 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-12 11:14:48 +00:00
a367be1e88
Rename ff_eval2() to ff_parse_and_eval_expr().
...
The new name better expresses what the function does.
Originally committed as revision 22845 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 22:32:04 +00:00
4565caf120
Rename ff_parse_eval() to ff_eval_expr().
...
The new name expresses better what the function does.
Originally committed as revision 22844 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 22:32:01 +00:00
f54978f117
Place some empty line in the doxy.
...
Improve readability, also consistent with the predominant doxy style.
Originally committed as revision 22841 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 20:04:15 +00:00
8c21647382
Remove redundant file descriptions from copyright headers.
...
File description is only kept in the @file doxy.
Originally committed as revision 22840 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 20:04:11 +00:00
a3731cadfc
Avoid the use of the symbol ff_expr_s for referencing AVExpr.
...
This way we have to deal only with struct AVExpr and AVExpr, which is
slightly less confusing as the association between the two symbols is
obvious.
Originally committed as revision 22839 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 18:44:51 +00:00
96561c12c5
DECLARE_ALIGNED usage requires #inclusion of 'mem.h'.
...
Originally committed as revision 22838 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 15:54:45 +00:00
80ed701443
Doxument ff_free_expr().
...
Originally committed as revision 22837 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 14:15:32 +00:00
21302fed02
Make ff_fill_linesize() use the information stored in
...
av_pix_fmt_descriptors.
Allow simplification and a more generic implementation.
Originally committed as revision 22836 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 14:09:05 +00:00
f8fea468b2
Rename ff_eval_free() to ff_free_expr().
...
Originally committed as revision 22834 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 11:47:33 +00:00
073f6d5b96
Rename AVEvalExpr to AVExpr, as suggested by Michael.
...
The new name is shorter and less confusing.
Originally committed as revision 22833 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 11:47:29 +00:00
5fccafdbc1
Move AVEvalExpr declaration at the beginning of the file, where it is
...
less distracting.
Originally committed as revision 22832 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-11 11:47:24 +00:00
3ffdd09a7a
alacenc : Report supported input sample formats.
...
Originally committed as revision 22829 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-10 17:32:41 +00:00
e73d1a5efc
ARM: NEON optimised synth_filter_float
...
2.7x faster DCA decoding on Cortex-A8
Originally committed as revision 22828 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-10 16:27:56 +00:00
f462ed1f82
Make synth_filter a function pointer
...
Originally committed as revision 22827 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-10 16:27:53 +00:00
38d52f3ea3
DCA: align some arrays
...
Optimised implementations of the synth filter will require these
arrays 16-byte aligned.
Originally committed as revision 22826 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-10 16:27:47 +00:00
ba659bedb8
Reindent read_sbr_extension.
...
Originally committed as revision 22820 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-08 06:27:39 +00:00
73c2704608
Print an error and skip PS when PS is found but explicitly found but
...
signaled to be absent.
Originally committed as revision 22819 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-08 06:27:35 +00:00
d14662f66a
Add support for PS sync extensions.
...
Originally committed as revision 22818 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-08 06:04:41 +00:00
37216b99e0
Use get_bits_left() in the sync extension check.
...
Originally committed as revision 22817 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-08 06:04:37 +00:00
702b73bc9f
Fix ext_object_type.
...
In the case of explicit non-backwards compible PS, the extension object
type should be set to SBR. See 14496-3:2009 (fourth edition).
Originally committed as revision 22816 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-08 06:04:31 +00:00
77a670e796
Fix segfault when encoder initialization fails.
...
Patch by Craig Thomasson $(name) dot $(surname) ripcode com
Originally committed as revision 22811 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-06 09:52:41 +00:00
2819760b5c
Avoid division by zero
...
Based on clang-scan report http://permalink.gmane.org/gmane.comp.video.ffmpeg.devel/107290
Originally committed as revision 22795 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-04 07:18:10 +00:00
c99baf05c3
Add help for (I)RDFT test in fft-test
...
Originally committed as revision 22793 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-03 15:05:27 +00:00
c2b774a04e
Make code using 1d-DCT consistent with the API change
...
Originally committed as revision 22792 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-03 15:04:15 +00:00
2874c81cc8
Replace all remaining occurrences of AVERROR_NOMEM with
...
AVERROR(ENOMEM).
AVERROR_NOMEM is deprecated and will be dropped at the next libavutil
major bump.
Originally committed as revision 22791 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-03 14:15:00 +00:00
523429220b
aacenc: Error when an unsupported profile is requested
...
Originally committed as revision 22784 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 16:26:10 +00:00
636da41a20
aacenc: Don't lowpass the input unless specifically requested.
...
The heuristic for estimating a good cutoff is busted.
Originally committed as revision 22779 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 14:19:39 +00:00
008593be52
Change default for bidir_refine to 1.
...
Originally committed as revision 22778 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 14:08:51 +00:00
fd8277ffa0
Optimize bidir_refine a bit.
...
compiled code is less tham 1/4 the size.
a tiny bit faster
Originally committed as revision 22777 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 14:04:15 +00:00
a69220cce0
Change bidir refine hash code so we only need to perform a single
...
hash calculation for the whole function.
negligibly faster (about 0.1%)
Originally committed as revision 22775 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 12:34:08 +00:00
2d603902f3
Reduce the size of the bidir refine hashtable from 4096 to 256 and
...
change the hash function to something more sane and simple.
about 1/3 faster, no meassureable change in psnr or size and i gues
its even same md5 with my test file but i forgot testing that.
Originally committed as revision 22773 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 11:06:29 +00:00
e056d2a24b
Move comment that belonged to cmp() back to cmp().
...
Originally committed as revision 22772 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 01:13:59 +00:00
919e749772
slice dice, inline and outline cmp()
...
motion_est.o is now less than half its previous size.
No speedchange meassureable.
Originally committed as revision 22771 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-02 01:07:03 +00:00
00bbc09309
Store range coder state transition table.
...
Use a better table, 2% compression gain for foreman
Originally committed as revision 22763 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-01 18:51:08 +00:00
fa034b4427
Allow hardcoding of ulaw and alaw tables.
...
Originally committed as revision 22762 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-01 17:54:15 +00:00
8eaa6e0e04
Change/simplify the tableprint/tablegen API.
...
Originally committed as revision 22761 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-01 17:11:47 +00:00
27eecec359
Convert two "m" constraints to MANGLE to fix compilation with some compilers.
...
Originally committed as revision 22760 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-01 16:52:14 +00:00
Måns Rullgård
Stefano Sabatini
Michael Kaufmann
Jai Menon
Alex Converse
Craig Thomasson
Reinhard Tartler
Vitor Sessak
Michael Niedermayer
Reimar Döffinger