| Thu, 02 Jun 2005 20:45:35 +0000 |
lorenm |
sort H.264 mmx dsp functions into their own file
libavcodec
|
| Wed, 01 Jun 2005 21:19:00 +0000 |
michael |
adding a few const
libavcodec
|
| Wed, 01 Jun 2005 18:04:01 +0000 |
michael |
100l for myself (breaking amd64)
libavcodec
|
| Wed, 01 Jun 2005 12:03:25 +0000 |
mru |
fix segfault if flush is called before the first frame
libavcodec
|
| Wed, 01 Jun 2005 11:56:58 +0000 |
michael |
merge a few asm blocks so gcc cant unoptimize it (658->631 dezicycles on duron)
libavcodec
|
| Wed, 01 Jun 2005 11:36:32 +0000 |
michael |
get rid of 2 movq (680 -> 658 dezicycles on duron)
libavcodec
|
| Wed, 01 Jun 2005 08:43:40 +0000 |
michael |
avoid one transpose (730->680 dezicycles on duron)
libavcodec
|
| Wed, 01 Jun 2005 04:51:46 +0000 |
lorenm |
10l (symbol mangling)
libavcodec
|
| Wed, 01 Jun 2005 01:18:41 +0000 |
michael |
add rounding bias before the horizontal idct (765->730 dezicyles on duron)
libavcodec
|
| Tue, 31 May 2005 22:48:33 +0000 |
lorenm |
MMX for H.264 iDCT (adapted from x264)
libavcodec
|
| Mon, 30 May 2005 22:02:00 +0000 |
michael |
decrease V_NB_BITS if possible
libavcodec
|
| Mon, 30 May 2005 20:34:05 +0000 |
michael |
more trivial optimizations
libavcodec
|
| Mon, 30 May 2005 20:34:02 +0000 |
mru |
enable x264 multi-threading
libavcodec
|
| Mon, 30 May 2005 19:14:50 +0000 |
michael |
maxdepth is supposed to be a constant, its faster if its too large then if its non constant
libavcodec
|
| Mon, 30 May 2005 15:03:06 +0000 |
michael |
optimizations
libavcodec
|
