This commit was manufactured by cvs2svn to create tag 'xvidcore'.
interlacing bugfix
debug output for cbr
moved HintedMEGet() on intra, to FrameCodeI()
hinted motion estimation support (raw or compressed)
reworked vlc tables
Added -mvec comment
source C files for sad Altivec
splat -> vxor optimizations
Added EPZS and EPZS^2. Unfortunately slower and not really better than PMVfast
optimization
CFLAGS fix
PPC trivial fixes
cygwin Makfile, thanks milan
PPC update
SAD in Altivec
Little fix
fixed crash when using lumi masking in cbr mode (i'm such an idiot)
fixed crash when using lumi masking in cbr mode (last time..)
fixed crash when using lumi masking in cbr mode (removed development code)
fixed crash when using lumi masking in cbr mode
merged image/pMBs create and destroy stuff into single if()
interlacing speedup - transfer strides used instead of manual field-to-frame conversion
interlacing fix and speedup - compensated data wasn't being used in block reconstruction
win32 fix - #define can't immediately follow #define ;-)
Sorry, the previous patch was buggy. This one shoud fix the problem
Big patch, use it with care (tests done with GNU/Linux) : - Trivial warnings in encoder.c line 282 ret1 = ret 2 = 0; - Lot of "New line at end of file missing" warnings fixed (gcc3) - All 2d arrays have been turned into 1d arrays To force alignment of matrices on stack, use DECLARE_ALIGNED_MATRIX (read portab.h) Win32 users should give feedback.
DIRS_PPC update
dct altivec
ASM update
fixed some warnings
LINUX: proper alignement on the stack
aligned memory for better performance
first bits of Altivec support
PPC Altivec version
interlacing support - slow (MBDecideFieldDCT()), buggy (p-frames), incomplete (field motion prediction)
added *.h to msvc project, removed __declspec(align(..)) for _MSC_VER <= 1200
Changed rc_buffersize to 16 as decided in the forum
s/16/CACHE_LINE/
Fixed a bug when malloc aligns the data itself - Added comments
foxer's cbr mode (hijacks the rc_buffersize variable and uses it as reaction delay - easier than changing xvid.h)
read words in memory instead of hw
no message
cosmetica
loop2 now uses CTR
some label cleanups
use simplified mmemonics
CR0 use
use 386 cpu 'bts' in 'cbp |= 1 << (edx-1)
Added first PPC asm port
no message
adopted to new API
cbp_calc for PPC
Added GNU CC alignment
removed intel asm parts
Fixed bug pointed by canard (out of bounds reading and a big mistake in j upper limit)
BSWAP,EMMS macro for LinuxPPC
LinuxPPC Makefile
I prefer this traditional way of alignment
alignment stuff
xvid_malloc/xvid_free
added 3dnow hv interpolation
added interpolate8x8_halfpel_hv_xmm
Disabled DEBUG-messages in Linux, if _DEBUG not set
removed init_timer
removed debug output
disable win32 debug output in release builds (encodes 0.7% faster)
change release compilation inlining mode to "any suitable" - runs faster (just)
prevented RateControlUpdate() being called in non-cbr encodes (bitrate isn't 0 anymore - it's set to 900000 in encoder_create())
INTRA_Q/INTER_Q mcbpc bugfix
readme
no message
no message
qmatrix bugfix
Fixed warning about csp_tmp unitialized
Fixed gcc warnings
Fixed implicit declaration of malloc - #include <stdlib.h>
Fixed brackets for coeff_tab3, coeff_tab7, coeff_tab10 and coorect indentation of all code
Fixed gcc warning about '/*' into comments
Fixed lot of gcc warnings about '/*' into comments
Small cleanup for gcc warnings and code indentation (thx emacs indent-region :-)
In some circumstances acpred_flag was used unitialized - now it's set to zero
New tree structure
bugfix
now compatible with new XviD API
custom qmatrix support
moved sources
