[x264-devel] x264 Development Newsletter: Vol 24

Jason Garrett-Glaser jason at x264.com
Sat Oct 22 02:30:08 CEST 2011


This is the twenty-fourth x264 development newsletter. This is a
regular email containing updates on fixes and improvements in the most
recent x264 push, along with  updates on what's coming next.  Previous
versions can be found in the mailing list archives.

Fixes:

Fix detection of Alpha CPU arch on alphaev67.

Fix use of uninitialized MVs in sub8x8 RDO in some cases.

Fix chroma intra analysis in lossless 4:4:4.

Fix pixel_var2 with 4:2:2 encoding.  Might have caused some artifacts
in rare cases.

Fix integer overflow/signedness errors found by IOC.  One of these
might have been a real bug in extreme situations (really high
resolution).

Fix infinite loop parsing TDecimate Mode 3 timecode v1 files.

Improvements:

Improve profile help: list high422/high444 profiles and don't show
non-high-bit-depth profiles in high-bit-depth builds.

Allow setting a chroma format at compile-time, for slight speed
increase and binary size savings.

Various configure/Makefile fixes, mostly for Solaris.

Update ARM NEON chroma deblock functions to work with NV12 pixel format.

Initial XOP and FMA4 support on Bulldozer CPUs.  A percent or two
faster overall.

Faster i4x4/i8x8 analysis: inline encoding into intra analysis, plus
merge the prediction step with mbcmp_x9.

9-way merged SAD/SATD_x9 for i8x8, similar to the previous i4x4
version.  Significantly faster on SSSE3-and-later CPUs.

Remove obsolete versions of mbcmp_x3, no longer needed on CPUs with
SSSE3 or later due to the existence of mbcmp_x9 on those CPUs.

More x86 versions of missing 4:2:2 asm functions.

Add 4:2:2 version of prefetch_fenc to make sure all of the chroma data
gets prefetched properly.

Add prefetching step to adaptive quant; avoids a bunch of cache misses
and improves speed.

Upcoming:

--device and automatic --level restriction support is in the works, as
part of Google Code-In.  The patch is done, but needs review and adoption.

Automatic range handling (full vs TV) is in the works.

A per-option help system is in the works, as part of Google Code-In.
The patch is done, but needs editing of the help entries.

x262 is under development: a best-in-class MPEG-2 encoder built using
the x264 framework.  Both P and B-frames are done and working.

Google Summer of Code is over, and the two projects, Trellis ME and
H.264/MVC support, are being finished up.

Jason Garrett-Glaser

The x264 Team


More information about the x264-devel mailing list