| Commit message (Expand) | Author | Age | Files | Lines |
* | PIC fix | Michael Niedermayer | 2006-11-27 | 1 | -1/+1 |
* | Reenable AMD64 optimizations for cabac accidentially disabled in r6852 | Reimar Döffinger | 2006-11-01 | 1 | -8/+5 |
* | Add ARCH_X86_32 as a new define for 32 bit x86 architectures and change | Diego Biurrun | 2006-11-01 | 1 | -5/+5 |
* | Fix compilation with PIC enabled, BRANCHLESS_GET_CABAC is defined under | Diego Biurrun | 2006-10-29 | 1 | -2/+2 |
* | CABAC assembler optimizations ported to AMD64 | Reimar Döffinger | 2006-10-23 | 1 | -52/+64 |
* | decode_significance_8x8_x86() | Michael Niedermayer | 2006-10-20 | 1 | -0/+56 |
* | Protect code that uses CMOV instructions with HAVE_CMOV, | Guillaume Poirier | 2006-10-20 | 1 | -2/+1 |
* | another instruction less in decode_significance_x86() -> 1% faster ion P3 | Michael Niedermayer | 2006-10-20 | 1 | -3/+2 |
* | 1 instruction less | Michael Niedermayer | 2006-10-20 | 1 | -1/+2 |
* | reordering instructions a little in decode_significance_x86() -> 2 instructio... | Michael Niedermayer | 2006-10-20 | 1 | -7/+5 |
* | factorize get_cabac asm (0.5% slower but its much cleaner) | Michael Niedermayer | 2006-10-20 | 1 | -168/+59 |
* | Fix PIC compilation, some defines were under #ifdef !PIC but used | Bernhard Rosenkränzer | 2006-10-19 | 1 | -3/+3 |
* | unused variable | Michael Niedermayer | 2006-10-19 | 1 | -1/+0 |
* | optimize sign decoding code in decode_residual() | Michael Niedermayer | 2006-10-19 | 1 | -0/+81 |
* | Kill a warning with MSVC | Jindřich Makovička | 2006-10-18 | 1 | -2/+3 |
* | x86 asm version of the decode significance loop (not 8x8) of decode_residual(... | Michael Niedermayer | 2006-10-17 | 1 | -0/+165 |
* | cosmetic (%%eax->%0) | Michael Niedermayer | 2006-10-17 | 1 | -10/+10 |
* | Fix crash with illegal instruction, cmov is available on 686 and later only. | Diego Biurrun | 2006-10-16 | 1 | -1/+1 |
* | Expand some #endif comments. | Diego Biurrun | 2006-10-16 | 1 | -2/+2 |
* | fix !CMOV_IS_FAST case (iam not really happy with the fix but i didnt come up... | Michael Niedermayer | 2006-10-16 | 1 | -1/+2 |
* | 10l | Michael Niedermayer | 2006-10-15 | 1 | -1/+1 |
* | this code will not work with PIC as it needs 7 registers and gcc doesnt suppo... | Michael Niedermayer | 2006-10-15 | 1 | -1/+1 |
* | shift CABACContext.range right, this reduces the number of shifts needed in g... | Michael Niedermayer | 2006-10-15 | 1 | -46/+48 |
* | dehack *ps_state indexing in the branchless decoder | Michael Niedermayer | 2006-10-13 | 1 | -2/+3 |
* | add "memory" to the clobber list we change memory so we need it, this also fi... | Michael Niedermayer | 2006-10-12 | 1 | -2/+2 |
* | prevent "mb level" get_cabac() calls from being inlined (3% faster decode_mb_... | Michael Niedermayer | 2006-10-12 | 1 | -1/+9 |
* | adds some useful comments after some of the #else, #elseif, | Guillaume Poirier | 2006-10-12 | 1 | -8/+8 |
* | Rename ABS macro to FFABS. | Diego Biurrun | 2006-10-11 | 1 | -1/+1 |
* | slightly faster on P3 slightly slower on athlon and probably faster on P4 | Michael Niedermayer | 2006-10-11 | 1 | -0/+10 |
* | moving lps state transition code a little up in the branched asm code (1% fas... | Michael Niedermayer | 2006-10-11 | 1 | -2/+2 |
* | write cabac low and range variables as early as possible to prevent stalls fr... | Michael Niedermayer | 2006-10-11 | 1 | -3/+4 |
* | use ecx instead of cl (no speed change on P3 but might avoid partial register... | Michael Niedermayer | 2006-10-11 | 1 | -4/+4 |
* | make state transition tables global as they are constant and the code is slig... | Michael Niedermayer | 2006-10-11 | 1 | -19/+14 |
* | 10l | Michael Niedermayer | 2006-10-11 | 1 | -2/+2 |
* | make lps_range a global table its constant anyway (saves 1 addition for acces... | Michael Niedermayer | 2006-10-11 | 1 | -15/+11 |
* | enable CMOV_IS_FAST as its faster or equal speed on every cpu (duron, athlon,... | Michael Niedermayer | 2006-10-11 | 1 | -0/+1 |
* | BRANCHLESS_CABAD --> BRANCHLESS_CABAC_DECODER | Diego Biurrun | 2006-10-10 | 1 | -3/+3 |
* | moving another bit&1 out, this is as fast as with it in there, but it makes m... | Michael Niedermayer | 2006-10-10 | 1 | -2/+1 |
* | move the &1 out of the asm so gcc can optimize it away in inlined cases (yes ... | Michael Niedermayer | 2006-10-10 | 1 | -1/+1 |
* | replace a few and/sub/... by cmov | Michael Niedermayer | 2006-10-10 | 1 | -1/+10 |
* | reading 8bit mem into a 8bit register needs 2 uops on P4, 8bit->32bit with ze... | Michael Niedermayer | 2006-10-09 | 1 | -7/+7 |
* | on the P4 inc needs twice as much time a add | Michael Niedermayer | 2006-10-09 | 1 | -1/+1 |
* | 10l | Michael Niedermayer | 2006-10-09 | 1 | -4/+0 |
* | reverse remainder of the failed attempt to optimize *state=c->mps_state[s] | Michael Niedermayer | 2006-10-09 | 1 | -8/+8 |
* | x86 branchless cabac decoder | Michael Niedermayer | 2006-10-09 | 1 | -1/+68 |
* | optimize branchless C CABAC decoder | Michael Niedermayer | 2006-10-09 | 1 | -3/+4 |
* | move outcommented START/STOP_TIMER to a hopefully better place for benchmarki... | Michael Niedermayer | 2006-10-09 | 1 | -2/+0 |
* | drop failed attempt to optimize *state= c->mps_state[s]; | Michael Niedermayer | 2006-10-09 | 1 | -17/+1 |
* | 10l bugfix for some disabled code | Michael Niedermayer | 2006-10-09 | 1 | -2/+2 |
* | first try of a handwritten get_cabac() for x86, this is 10-20% faster on P3 d... | Michael Niedermayer | 2006-10-09 | 1 | -0/+85 |