Commit Graph

10914 Commits

Author SHA1 Message Date
Michael Niedermayer
e814817b74 Factor mv/ref compare code out.
This is a hair slower (0.15% maybe) but i really dont want to have the
identical code duplicated 3 times because gcc adds odd threaded jumps with
register reshuffling and register safe/restore.

Originally committed as revision 21502 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 10:10:02 +00:00
Michael Niedermayer
3b84924516 Simplify first edge filter condition.
Originally committed as revision 21497 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 02:41:52 +00:00
Michael Niedermayer
b6302d0c55 Cosmetics, mostly indention, 2 or so new fixme comments that i was to lazy
to split out

Originally committed as revision 21496 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 02:20:31 +00:00
Michael Niedermayer
0a32508d90 Make the fast loop filter path work with unavailable left MBs.
This prevents the issue with having to switch between slow and
fast code paths in each row.
0.5% faster loopfilter for cathedral

Originally committed as revision 21495 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 02:15:25 +00:00
Michael Niedermayer
b304767301 get rid of the start variable.
a few cycles faster

Originally committed as revision 21494 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 01:31:06 +00:00
Michael Niedermayer
980bcc554d Unroll main loop so the edge==0 case is seperate.
This allows many things to be simplified away.
h264 decoder is overall 1% faster with a mbaff sample and
0.1% slower with the cathedral sample, probably because the slow loop
filter code must be loaded into the code cache for each first MB of each
row but isnt used for the following MBs.

Originally committed as revision 21493 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-28 01:24:25 +00:00
Rafaël Carré
881b5b80da Fix svq3_* function declarations.
Patch by Rafaël Carré, rafael D carre A gmail

Originally committed as revision 21489 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 22:22:01 +00:00
Måns Rullgård
e3d8807315 SVQ1: 4-byte-align codebook tables
These tables are accessed as uint32 so they must be suitably aligned.

Originally committed as revision 21488 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 21:31:19 +00:00
Reimar Döffinger
7e2ef1b9e1 Fix possible crashes in mlp parser, it tries to go back 7 bytes after
finding the 4-byte signature.
Add a check that ignores the signature if we do not have enough previous
data to go back at least 7 bytes.

Originally committed as revision 21487 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 20:04:19 +00:00
Reimar Döffinger
d780511b94 Reduce log level of "Extracting .. channel downmix" to AV_LOG_DEBUG, the
message is pointless to ordinary users and is printed once per frame.

Originally committed as revision 21486 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 20:02:18 +00:00
Michael Niedermayer
9f72b23c5c Revert Revision: 21362
Log:
	Set CODEC_CAP_SUBFRAMES for adpcm decoders

	This makes ffmpeg stop printing millions of
	  Multiple frames in a packet from stream 0
	when decoding adpcm.
Reason: adpcm has no subframes, the flag does not belong there

Originally committed as revision 21481 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 13:30:05 +00:00
Michael Niedermayer
8670f84cf9 Update comment.
Originally committed as revision 21479 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 13:18:08 +00:00
Michael Niedermayer
e470ef7641 Use table to speedup access to non_zero_count in MBAFF with differing interlacing.
~4 cpu cycles speedup

Originally committed as revision 21474 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 11:14:29 +00:00
Stefano Sabatini
e8a9d01846 Fix misc typos, patch by
Fabian Greffrath base64_decode(PGZhYmlhbkBncmVmZnJhdGguY29tPg==).

Originally committed as revision 21473 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 00:31:52 +00:00
Michael Niedermayer
8652e44acd Simplify left_xy init
Originally committed as revision 21470 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-27 00:15:55 +00:00
Michael Niedermayer
599fe45b8d Split fill_caches() between loopfilter & decode, the 2 no longer where common
enough to justify the messy interleaving.

Originally committed as revision 21469 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 23:54:11 +00:00
Michael Niedermayer
16e5e39ab4 Optimize loop filtering of the left edge in MBAFF.
60 cpu cycles speedup

Originally committed as revision 21467 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 22:59:19 +00:00
Michael Niedermayer
dfe4dc154b use left_xy[1] in mbaff QP loop filter check, this improves the amount that can
be skiped.

Originally committed as revision 21465 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 20:28:58 +00:00
Michael Niedermayer
aebf31236e Optimize mv/ref cache init for left MB.
Originally committed as revision 21464 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 20:25:14 +00:00
Måns Rullgård
f2c694af28 Add UMULH()
This function/macro returns the high half of the unsigned 64-bit
product of two unsigned 32-bit operands.

Originally committed as revision 21463 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 18:05:55 +00:00
Loren Merritt
900479bb74 optimize h264_loop_filter_strength_mmx2
244->160 cycles on core2

Originally committed as revision 21462 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 17:17:48 +00:00
Michael Niedermayer
6548c939ec remove unneeded check
Originally committed as revision 21460 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 15:34:21 +00:00
Michael Niedermayer
18ea2f933c Use left_mb_xy from fill_caches instead of recalculating it.
Originally committed as revision 21459 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 14:57:53 +00:00
Michael Niedermayer
a715af8ff4 Simplify left_xy content for the loop filter, this also makes it closer to
what is needed and its faster too.

Originally committed as revision 21458 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 14:55:19 +00:00
Michael Niedermayer
d5c30c86d0 Simplify loop filter a little by using top/left_type.
Originally committed as revision 21457 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 13:39:26 +00:00
Michael Niedermayer
99344d4372 Set top & left types for deblock in fill_caches().
Originally committed as revision 21456 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 13:38:18 +00:00
Michael Niedermayer
aaa995d7f2 100l typo fix, mixed up +-1 forcing the loop filter skip to never skip.
Originally committed as revision 21455 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 13:30:33 +00:00
Michael Niedermayer
66472bcde0 cosmetic
Originally committed as revision 21454 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 13:28:55 +00:00
Michael Niedermayer
3046c25ec5 Fix qp_thres loop filter check for MBAFF.
Originally committed as revision 21453 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 13:27:22 +00:00
Michael Niedermayer
806ac67b51 Optimize mb neighbor initialization for MBAFF in fill_caches().
~10 cpu cycles speedup.

Originally committed as revision 21452 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 10:35:36 +00:00
Carl Eugen Hoyos
ddee80f384 Metasoft MJPEG codec has Cb and Cr swapped, fixes issue 1611.
Originally committed as revision 21451 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 10:30:24 +00:00
Michael Niedermayer
021dccba1f Fix heap overflow due to lack of nb_components check.
Originally committed as revision 21450 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-25 13:26:10 +00:00
Carl Eugen Hoyos
cc5d4f4c34 Release buffer when decoding is finished.
Originally committed as revision 21449 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-25 10:46:32 +00:00
Michael Niedermayer
2773920698 Optimize decode_cabac_field_decoding_flag().
~4 cpu cycles faster

Originally committed as revision 21447 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-25 02:44:34 +00:00
Alexander Strange
1e4f1c56d7 100l, correct wrong H.264+adaptive MBAFF decoding
Fixes several FATE tests.

Originally committed as revision 21445 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-25 01:56:13 +00:00
Alexander Strange
0b69d6254f H.264: Use 64-/128-bit write-combining macros for copies
2-3% faster decode on x86-32 core2.

Originally committed as revision 21440 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-25 00:30:44 +00:00
Justin Ruggles
008f872f61 Only check frame size if the header is valid.
Originally committed as revision 21439 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 23:55:33 +00:00
Carl Eugen Hoyos
b52fcba9cd Fix decoding ATI VCR2 codec, sample in
samples/V-codecs/ATI-VCR12/ATI/VCR2test.avi
Software decoding was broken since r8865, when two new context variables
were added, but not initialized in vcr2_init_sequence().
Another such context variable was added in r18381.

Originally committed as revision 21438 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 22:47:50 +00:00
Michael Niedermayer
996b099a0f Branchless setting of MB_TYPE_8x8DCT.
Not benchmarked as i failed to find a sample that uses this one. But it should be faster.

Originally committed as revision 21435 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 20:54:09 +00:00
Michael Niedermayer
81afcf1fae Remove cruft.
Originally committed as revision 21434 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 20:52:49 +00:00
Michael Niedermayer
d9ef0d2e14 Mark index as unsigend so gcc knows it doesnt need a sign extension on x86_64.
Any tips on how i can convince gcc that it doesnt need a
mov     %eax, %eax
in every get_bits() ?

Originally committed as revision 21433 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 19:19:09 +00:00
Michael Niedermayer
449d1442a6 a[b-1] -> (a-1)[b].
Helps gcc not to add seperate -1 instructions.

Originally committed as revision 21432 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:42:22 +00:00
Michael Niedermayer
7abc860323 Optimize suffix_length computation, 1 cpu cycle speedup.
Originally committed as revision 21431 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:23:46 +00:00
Michael Niedermayer
eeb1e92feb Simplify suffix_length computation, same speed.
Originally committed as revision 21430 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:18:08 +00:00
Michael Niedermayer
c78295ad1b Optimize level_code computation, 6cpu cycles speedup.
Originally committed as revision 21428 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:17:01 +00:00
Reimar Döffinger
0b882b4009 Fix crash in MLP decoder due to integer overflow.
Probably only DoS, init_get_bits sets buffer to NULL, thus causing a
NULL-dereference directly after.

Originally committed as revision 21426 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:07:29 +00:00
Michael Niedermayer
8ba436171f 1 cpu cycle faster suffix_length calculation.
Originally committed as revision 21425 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 18:05:02 +00:00
Laurent Aimar
f0f86627fe VC-1/WMV3 DXVA2 implementation
It allows VLD VC-1/WMV3 decoding using DXVA2 (GPU assisted decoding API under
VISTA and Windows 7).
 It is implemented by using AVHWAccel API.

Originally committed as revision 21424 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 17:27:00 +00:00
Michael Niedermayer
77d40dce93 Remove is_complex from loop_filter() its useless there in its current form.
Originally committed as revision 21423 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 17:02:44 +00:00
Laurent Aimar
34fecc9afd Added a few missing consts in dxva_h264.
Originally committed as revision 21422 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-24 16:40:58 +00:00