Log of /branches/dev-api-4/xvidcore/src/xvid.c
Parent Directory
Revision
1205 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Thu Nov 13 23:11:24 2003 UTC (19 years, 10 months ago) by
edgomez
File length: 19600 byte(s)
Diff to
previous 1197
MMXed the calculation of SSE for 8x8 16bit blocks. This helps quite
a lot VHQ=4 mode.
My tests show with trellis:chroma_me:
- ~20% speed improvement for vhq=4.
- at least 5% when using vhq=1.
Of course this speedup vanishes if more CPU intensive features are
used. CruNcher who used gmc/qpel, noticed "only" a ~5% speed
improvement.
NB: i'm of course talking about overall speed improvement. Such a
small patch for such a big improvement :-)
Revision
1197 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Sun Nov 2 23:02:52 2003 UTC (19 years, 11 months ago) by
edgomez
File length: 19527 byte(s)
Diff to
previous 1195
* Added Dmitry SSE2 iDCT code back.
* Plugged Dmitry iDCT as default for SSE2
* Fixed a bug in xvid_bench that was making it would test some CPU
instruction set w/o host CPU support. xvidcore init was simply
discarding irrelevant cpu flags.
Revision
1190 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Mon Oct 27 01:03:06 2003 UTC (19 years, 11 months ago) by
edgomez
File length: 19458 byte(s)
Diff to
previous 1174
* Ported the ffmpeg fDCT functions (mmx and xmm).
* Modified the skal's versions a bit to allow rolling loops.
* Activated Skal's fDCTs (unrolled versions) for mmx _and_ xmm
(old code was ignoring xmm versions)
* Removed the SSE2 versions (they'll be back later)
* .data -> .rodata
* Applied announced asm CodingStyle to the dct dir
(I'll have to add a section with the said CodingStyle)
Revision
1058 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Mon Jun 9 19:42:25 2003 UTC (20 years, 3 months ago) by
edgomez
File length: 29092 byte(s)
Diff to
previous 1054
Activated simple_idct for MMX - Backward compatibility with bistream version 9 ensures the use of the standard idct_mmx function
Revision
1031 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Sat May 17 13:26:51 2003 UTC (20 years, 4 months ago) by
suxen_drol
File length: 28824 byte(s)
Diff to
previous 982
* dynamic bvop decision stats bug fixed
* added runtime debug level option. this only only activated when _DEBUG is defined.
level is specified globaly, via "xvid_glb_init_t.debug" and XVID_DEBUG_xxx defines.
references to previous DPRINTF_LEVEL_xxx's have been replaced throughout libxvidcore
* futher bitrate zone fixes for plugin_2pass2.c
* removed old ratecontrol.c from msvc project
* xvid_encraw.c specifc:
* added "-max_key_interval integer" option
* renamed -debug to -vop_debug
* added "-debug integer" option, which sets xvid_gbl_init_t.debug field
* s/length/len/ status line
* vfw specific:
* zone/bitrate typos fixed ("you eediot")
* renamed "debug" regkey to "vop_debug"
* added "debug" regeky and textbox which sets xvid_gbl_init_t.debug field
Revision
982 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Thu Apr 10 13:05:54 2003 UTC (20 years, 5 months ago) by
edgomez
File length: 28681 byte(s)
Diff to
previous 948
2003-04-10 GMT Edouard Gomez <ed.gomez@free.fr> patch-14
Summary:
Removed all ABS() macros.
Revision:
xvidcore--devapi4--1.0--patch-14
All ABS macros have been replace with their stdlib.h/math.h
equivalent. This gives a 33% overall speedup for the plain C encoder,
while the ia32 one seems to suffer a small speed loss. However this
speed loss is very small and it seems it depends on the CPU type as
the abs/fabs usage is impacting badly on sad functions but impacting
well on interpolate functions ... weird inst'it ?
modified files:
src/bitstream/mbcoding.c src/global.h src/image/image.c
src/motion/motion_comp.c src/motion/motion_est.c
src/motion/sad.c src/prediction/mbprediction.c
src/utils/mbtransquant.c src/xvid.c
Revision
824 -
(
view)
(
download)
(
annotate)
-
[select for diffs]
Modified
Sun Feb 9 19:32:52 2003 UTC (20 years, 7 months ago) by
edgomez
Original Path:
trunk/xvidcore/src/xvid.c
File length: 13749 byte(s)
Diff to
previous 648
I made all these changes on my own working copy until i had
something really usable wihtout regressions for most users.
I think its place is now in CVS. Here is a ChangeLog summary
generated from my local patches.
base-0
Imported xvidcore 0.9.0 into arch repository
patch-1
Updated to current stable CVS_HEAD
patch-2
Changed build system for Unix OSes
patch-3
Fixes for the new build system in sources.
patch-4
Fixed options and added the --disable-assembly option
patch-5
Fixed a BSD checking in nasm output format.
patch-6
Fixed the "ar" "s" option for some platforms.
patch-7
Changed the way I add strings into variables.
patch-8
Fixed portab.h for _DEBUG target and GCC/ICC compilers.
patch-9
Added Altivec detection (Guillaume Morin)
patch-10
Fixed MacOSX build.
patch-11
Fixed WIN32/_MSC_VER confusion and updated MSVC libxvidcore.dsp project file.
patch-12
DivX4 compatibility layer has been turned into an option (default:disable).
ToDo:
- Fix IA64 assembly file selection.
- Do I disable PPC assembly code ? (according to Guillaume, it needs to be updated)
This form allows you to request diffs between any two revisions of this file.
For each of the two "sides" of the diff,
enter a numeric revision.
