Mercurial > libavcodec.hg
graph
-
30% faster ff_vorbis_floor1_render_list, 3% faster overall libavcodecSat, 09 Feb 2008 13:53:15 +0000, by lorenm
-
use fewer registers in apply_welch_window_sse2 libavcodecSat, 09 Feb 2008 05:29:47 +0000, by lorenm
-
sse2 h264 motion compensation. not new code, just separate out the cases that didn't need ssse3. libavcodecWed, 06 Feb 2008 12:32:31 +0000, by lorenm
-
put loop counter in a register if possible. makes some of the qpel functions 3% faster. libavcodecWed, 06 Feb 2008 04:44:21 +0000, by lorenm
-
Check that we have enough input data in IMC decoder. libavcodecTue, 05 Feb 2008 18:39:55 +0000, by reimar
-
add qpel rounder once during hv rather than twice during hv and whatever it's averaged with libavcodecTue, 05 Feb 2008 03:58:13 +0000, by lorenm
-
clean up an ugliness introduced in r11826. this syntax will require fewer changes when adding future sse2 code. libavcodecTue, 05 Feb 2008 01:16:48 +0000, by lorenm
-
Deprecate old and inefficient per instruction asm(). libavcodecMon, 04 Feb 2008 20:03:08 +0000, by michael
-
Make luma_dc_quant_i and luma_dc_quant_p const, fixes a couple of libavcodecMon, 04 Feb 2008 10:26:35 +0000, by diego
-
mpegvideo.h has two function declarations with the 'inline' specifier libavcodecSun, 03 Feb 2008 17:54:30 +0000, by diego
-
use mmx2/3dnow avg functions in avg_qpel*_mc00 libavcodecSun, 03 Feb 2008 16:19:37 +0000, by lorenm
-
remove some movq in ff_h264_idct8_add_mmx. 225->217 cycles on core2. libavcodecSun, 03 Feb 2008 03:21:47 +0000, by lorenm
-
Make avcodec_decode_* functions take const input buffers. libavcodecSat, 02 Feb 2008 22:54:50 +0000, by michael
-
100l, since we already check for buf_size == 769 we should also libavcodecSat, 02 Feb 2008 22:34:44 +0000, by reimar
-
Use bytestream_get_be24 to simplify palette parsing. libavcodecSat, 02 Feb 2008 22:31:11 +0000, by reimar