summaryrefslogtreecommitdiff
path: root/src/post
AgeCommit message (Collapse)Author
2014-12-16Use correct absolute value function with floats and doublesBrad Smith
2014-12-07Fix compiling greedy2frame_template_sse2.cPetri Hintukainen
Some compilers generate invalid assembler code (64-bit register prefix) when %q is used in 32-bit code. Maybe compiler bug ? ("Produces the quad word register name for an operand if the target supports quad word. Otherwise, it produces a word register name.")
2014-11-19Revert part of 12593:f7c8f59bf518Petri Hintukainen
Fixes build issues.
2014-11-18goom/convolve_fx: add 64bit MMX version.Torsten Jager
2014-11-06Copyright year update by hg log.Torsten Jager
2014-11-06post: silence warnings.Torsten Jager
For historical reasons, arg #4 of xine->config->register_enum () is of type (char **). That string array will not be modified though.
2014-08-04Add missing LIBINTLCarlo Bramini
2014-06-06Copyright year update by hg log.Torsten Jager
2014-05-19Fix Solaris problem in goom assembler.Jonathan Perkin
Solaris cannot handle this MMX section, failing with: error: can't find a register in class GENERAL_REGS while reloading 'asm'.
2014-05-19Updated comment why code is disabled in X86_64Petri Hintukainen
2014-05-19moved register initializations out of loop. do not require registers for ↵Petri Hintukainen
arguments.
2014-05-19Require less GPRs for asm parameters.Alan Barrett
In affine_1d_MMX(), move the initialisation of %mm3 and %mm4 registers into a separate asm statement, to give the compiler more freedom for register allocation. Fixes a problem with gcc-4.5.4 on NetBSD/i386.
2014-05-18Only compile MMX/SSE code on x86/x86_64.Thomas Klausner
The greedy2frame code uses MMX or SSE instructions. Only compile this code on architectures where they exist.
2014-04-22fix debug build with clangPetri Hintukainen
2014-04-20Fix build with clangPetri Hintukainen
2014-03-10Silence compiler warnings on register_enum () arg #4.Torsten Jager
Making them all "const char * const *" did work too (even with Kaffeine build/run), but that would be an API change.
2013-10-14Copyright year update by hg log info #2.Torsten Jager
2013-10-01Fixed converting pointer to intPetri Hintukainen
2013-10-01Hide warining: 'fields[?]' may be used uninitialized in this functionPetri Hintukainen
2013-09-30gsl_read_file(): check for read errorPetri Hintukainen
2013-09-30goom: comment out unused functionsPetri Hintukainen
2013-09-30goom: Mark data static const. Use union to fix initialization.Petri Hintukainen
2013-09-30goom: mark static functions staticPetri Hintukainen
2013-09-30goom: fix declarations of functions without parametersPetri Hintukainen
2013-09-30goom: fix warning: * may be used uninitialized in this functionPetri Hintukainen
2013-09-30goom: add const to eliminate warningsPetri Hintukainen
2013-09-30Hide warningPetri Hintukainen
2013-09-30Fixed warningsPetri Hintukainen
2013-09-30Fix READ_PREFETCH_2048 so that it won't be optimized out, re-ordered or ↵Petri Hintukainen
interleaved by compiler. Simplify generated code (no add).
2013-09-13Copyright year update by hg log infoTorsten Jager
script execution time: 55"
2013-07-11Fix spurious executable permissions.Xavier Bachelot
2013-07-03Fix FSF address in a bunch of filesXavier Bachelot
2012-12-23goom: single pass rgb to yuy2 conversionTorsten Jager
2012-06-15Fixed greedy2frame_template_sse2.c when only 4 registers are availableRoland Scheidegger
2012-06-10Fix building on non-x86 systemsPetri Hintukainen
2012-06-09Merge from 1.1.Darren Salt
--HG-- rename : src/libfaad/xine_faad_decoder.c => src/audio_dec/xine_faad_decoder.c
2012-06-09Fix a possible NULL dereference when cleaning up in the FFT vis plugin code.Darren Salt
2012-05-15speedy.c: added vfilter_chroma_332_packed422_scanline_sse2()Petri Hintukainen
2012-05-15speedy.c: added diff_factor_packed422_scanline_sse2()Petri Hintukainen
Tested with Atom N550 + SD video: - ~25% speedup with unaligned data - ~50% speedup aligned data (scanlines were properly aligned with all my test clips)
2012-05-25Fixed compiling DeinterlaceGreedy2Frame_SSE2() when only 5 GP registers are ↵Petri Hintukainen
available
2012-05-22Added SSE2 version of DeinterlaceGreedy2FrameRoland Scheidegger
2012-05-22Removed incorrect .alignPetri Hintukainen
2012-05-22Simplify mmx constant loadingRoland Scheidegger
2012-05-22Renamed DeinterlaceGreedy2Frame_SSE() to DeinterlaceGreedy2Frame_MMXEXT()Roland Scheidegger
2012-01-12Fixed leakPetri Hintukainen
2012-01-11Fixed pointer castsPetri Hintukainen
* * * Fixed pointer cast
2012-04-15simplify greedy2frame deinterlacer a bitRoland Scheidegger
Cuts roughly 10% of the instructions (with sse), results should be identical. Not sure why it was that complicated in the first place, the simplification is possible because the code gave a score of 1 to top and bottom comparisons, and 2 for the middle one, and weaved when all scores added together were more than 2. This is equivalent to weave when (cmp(m) AND (cmp(b) OR cmp(t))) which is a much better match for the available hw instructions. This also reduces the number of constant loads a lot, and the patch moves up some memory loads a bit which can never hurt.
2012-01-12Merge from 1.1Petri Hintukainen
--HG-- rename : src/libspudvb/xine_spudvb_decoder.c => src/spu_dec/spudvb_decoder.c
2011-10-17Merge from 1.1Petri Hintukainen
2011-10-15Disable strict-aliasing where it caused compiler warningsChristian Ruppert
generated by cgit v1.2.3 (git 2.25.1) at 2025-06-05 11:41:11 +0000