git-svn-id: file:///var/local/repositories/ffmpeg/trunk@736 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@735 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

* generic optimization:
  - enable ALT and ALIGNED stream
  - move "if(const)" out from loops
  - supress duplicated computing
* first attempt to fix problem with strides


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@734 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@733 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@732 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@731 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@730 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@729 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@728 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@727 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@726 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

new bitstream reader API (old get_bits() based one is emulated and will still be supported in the future cuz its simpler)
minor optimizations to get_vlc


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@725 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@724 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

encoding of slice_height != mb_height
1bit shorter wmv1 headers if bit_rate<50 !? ROTFL M$ is even more stupid than i thought
using dc of non intra blocks for dc prediction of wmv1 if bitrate < 129 && res < 320x240


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@723 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@722 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

* Don't include config.h in dsputil.c


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@721 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@720 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@719 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@718 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@717 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

* Use "restrict" in some dsputil routines (yields large speedup)


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@716 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@715 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

for some improvements.


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@714 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@713 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

be a noticeable win (at least for me :)


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@712 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

takes the most time, and it allows for more efficient unaligned access
and better control over memory latencies.


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@711 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@710 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

allows better scheduling of the memory accesses, and is portable among
all compilers.


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@709 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

colorspace converter tests (memory corruption tests at different width/src/dst)
quite impressive results only 1 of the 13 rgb->rgb converters fails for the c versions
... sadly only 1 of the mmx converters passes though :( ... i feared allready  that last mplayerxp merge reversed the bugfixes :((((


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@708 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@707 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

init dc_scale tables to defaults (fixes mjpeg sig11)


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@706 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@705 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

* Support ccc
* Support gcc 3.2 builtins


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@704 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

code duplication, I'm currently working on the put/add variants, but I
did not get them to be as fast as the old method yet...


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@703 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

4mv bugfix


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@702 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

more mpeg4 fourcc's


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@701 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

fixing wmv1 bugs, it was foolish to belive that m$ would use the same dc prediction as mpeg4 just cuz they changed it a bit from msmpeg4v3


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@700 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@699 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

wmv1


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@698 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

wmv1 support


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@697 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b