log cabac.h @ 4043:87694a28120c libavcodec

age author description
Thu, 19 Oct 2006 18:22:03 +0000 michael unused variable libavcodec
Thu, 19 Oct 2006 01:19:03 +0000 michael optimize sign decoding code in decode_residual() libavcodec
Wed, 18 Oct 2006 09:37:25 +0000 gpoirier Kill a warning with MSVC libavcodec
Tue, 17 Oct 2006 22:18:29 +0000 michael x86 asm version of the decode significance loop (not 8x8) of decode_residual() 5% faster decode_residual() on P3 libavcodec
Tue, 17 Oct 2006 09:38:37 +0000 michael cosmetic (%%eax->%0) libavcodec
Mon, 16 Oct 2006 21:47:19 +0000 diego Fix crash with illegal instruction, cmov is available on 686 and later only. libavcodec
Mon, 16 Oct 2006 21:22:47 +0000 diego Expand some #endif comments. libavcodec
Mon, 16 Oct 2006 11:11:20 +0000 michael fix !CMOV_IS_FAST case (iam not really happy with the fix but i didnt come up with a better one quickly) libavcodec
Sun, 15 Oct 2006 21:04:10 +0000 michael 10l libavcodec
Sun, 15 Oct 2006 20:50:05 +0000 michael this code will not work with PIC as it needs 7 registers and gcc doesnt support that in PIC libavcodec
Sun, 15 Oct 2006 20:40:50 +0000 michael shift CABACContext.range right, this reduces the number of shifts needed in get_cabac() and is slightly faster on P3 (and should be much faster on P4 as the P4 except the more recent variants lacks an integer shifter and so shifts have ~10 times longer latency then simple operations like adds) libavcodec
Fri, 13 Oct 2006 14:21:25 +0000 michael dehack *ps_state indexing in the branchless decoder libavcodec
Thu, 12 Oct 2006 21:32:56 +0000 michael add "memory" to the clobber list we change memory so we need it, this also fixes some problems with gcc svn libavcodec
Thu, 12 Oct 2006 14:49:19 +0000 michael prevent "mb level" get_cabac() calls from being inlined (3% faster decode_mb_cabac() on P3) libavcodec
Thu, 12 Oct 2006 07:51:18 +0000 gpoirier adds some useful comments after some of the #else, #elseif, libavcodec
Wed, 11 Oct 2006 23:17:58 +0000 diego Rename ABS macro to FFABS. libavcodec
Wed, 11 Oct 2006 17:59:40 +0000 michael slightly faster on P3 slightly slower on athlon and probably faster on P4 libavcodec
Wed, 11 Oct 2006 16:39:50 +0000 michael moving lps state transition code a little up in the branched asm code (1% faster on P3) libavcodec
Wed, 11 Oct 2006 16:11:41 +0000 michael write cabac low and range variables as early as possible to prevent stalls from reading them before they where written, the P4 is said to disslike that alot, on P3 its 2% faster (START/STOP_TIMER over decode_residual) libavcodec
Wed, 11 Oct 2006 15:20:08 +0000 michael use ecx instead of cl (no speed change on P3 but might avoid partial register stalls on some cpus) libavcodec
Wed, 11 Oct 2006 14:44:17 +0000 michael make state transition tables global as they are constant and the code is slightly faster that way libavcodec
Wed, 11 Oct 2006 13:25:29 +0000 michael 10l libavcodec
Wed, 11 Oct 2006 13:21:42 +0000 michael make lps_range a global table its constant anyway (saves 1 addition for accessing it) libavcodec
Wed, 11 Oct 2006 12:23:40 +0000 michael enable CMOV_IS_FAST as its faster or equal speed on every cpu (duron, athlon, PM, P3) from which ive seen benchmarks, it might be slower on P4 but noone has posted benchmarks ... libavcodec
Tue, 10 Oct 2006 08:16:41 +0000 diego BRANCHLESS_CABAD --> BRANCHLESS_CABAC_DECODER libavcodec
Tue, 10 Oct 2006 06:56:51 +0000 michael moving another bit&1 out, this is as fast as with it in there, but it makes more sense with it outside of the loop libavcodec
Tue, 10 Oct 2006 01:17:39 +0000 michael move the &1 out of the asm so gcc can optimize it away in inlined cases (yes this is slightly faster) libavcodec
Tue, 10 Oct 2006 01:08:39 +0000 michael replace a few and/sub/... by cmov libavcodec
Mon, 09 Oct 2006 21:57:10 +0000 michael reading 8bit mem into a 8bit register needs 2 uops on P4, 8bit->32bit with zero extension needs just 1 libavcodec
Mon, 09 Oct 2006 21:39:07 +0000 michael on the P4 inc needs twice as much time a add libavcodec