| Thu, 23 Oct 2003 13:22:20 +0000 |
michael |
minor mmx2 optimization if the dct
libavcodec
|
| Thu, 23 Oct 2003 12:59:21 +0000 |
michael |
inline -> always_inline (842 -> 690 cpu cycles for dct_quantize() difference for the dct itself should be even bigger)
libavcodec
|
| Thu, 23 Oct 2003 12:41:54 +0000 |
michael |
4.9-RC FreeBSD doesnt like lrint() so change to lrintf(), not that bsd supports that but we emulate it ...
libavcodec
|
| Thu, 23 Oct 2003 09:11:56 +0000 |
michael |
optionally merge postscale into quantization table for the float aan dct
libavcodec
|
| Thu, 23 Oct 2003 08:31:26 +0000 |
michael |
10l
libavcodec
|
| Wed, 22 Oct 2003 22:12:25 +0000 |
alex |
10l
libavcodec
|
| Wed, 22 Oct 2003 22:10:22 +0000 |
alex |
100l
libavcodec
|
| Wed, 22 Oct 2003 19:15:55 +0000 |
michael |
oops forgot
libavcodec
|
| Wed, 22 Oct 2003 18:26:38 +0000 |
michael |
floating point AAN DCT
libavcodec
|
| Wed, 22 Oct 2003 15:36:46 +0000 |
michael |
112 bytes smaller MpegEncContext
libavcodec
|
| Wed, 22 Oct 2003 12:01:21 +0000 |
michael |
move blocks away MpegEncContext (7.4 -> 5.8k)
libavcodec
|
| Wed, 22 Oct 2003 11:23:39 +0000 |
michael |
move mmx quantizer matrixes out of MpegEncContext (23k -> 7k) (no meassureable slowdown)
libavcodec
|
| Wed, 22 Oct 2003 10:59:39 +0000 |
michael |
move q_*_matrix out of MpegEncContext (40k ->23k) dct_quantize() is even slightly faster now, dont ask my why ...
libavcodec
|
| Tue, 21 Oct 2003 19:55:31 +0000 |
michael |
dynamic alloc of picture structs instead of putting them in MpegEncContext
libavcodec
|
