Commit Graph

1338 Commits

Author SHA1 Message Date
Scott LaVarnway
ab00d209bc Improved tokenize
For a realtime HD encodings, up to 1.6% gains seen.



Change-Id: If45028e23db95124da63f9d38ffe06e05596cc6e
2011-09-30 12:49:46 -04:00
Johann
3556deaca3 combine loopfilter data access
The data processed by the loopfilter overlaps. At the block level, this
results in some redundant transforms. Grouping the filtering allows for
a single 16x16 transpose (and inversion) instead of three 16x8 transposes
(and three more inversions).

This implementation is x86_64 only. We retain the previous
implementation for x86.

Improvements are obviously material dependant, but it seems to be ~%1 in
tests here.

Change-Id: I467b7ec3655be98fb5f1a94b5d145e5e5a660007
2011-09-30 07:38:35 -07:00
Alpha Lam
7bce513afe Call vp8_find_near_mvs lazily
vp8_find_near_mvs() is being called on all possible reference frames
but the data computed may be used if the loop exits early, which can
be due to x->skip beign set to 1.

Optimize this by call vp8_find_near_mvs() laziy only if it is going
to be used and not computed yet.

Change-Id: Iccdbd4c962a670c9f2c99b8aca8096042ca5dc98
2011-09-30 14:48:18 +01:00
Paul Wilkins
a572ac8327 Merge "CQ and two pass rate control." 2011-09-30 02:57:54 -07:00
Paul Wilkins
b6e27d5f0b CQ and two pass rate control.
Changes to the selection of Q limits for two pass
and two pass CQ mode.

Allowance made for Mode and motion vector costs.
Some refactoring of common code.

For Derf and YT sets CQ mode average improvement
circa 1% (SSIM and Global PSNR).

Some increased tendency to undershoot even when
user CQ not reached.

Patch2: Removed some test code accidentally merged.

Change-Id: Icf74d13af77437c08602571dc7a97e747cce5066
2011-09-30 10:55:52 +01:00
Aaron Watry
69aa303d96 Reduce computational complexity of generic C loop filter.
Change-Id: I1e7f9ed3cd907844a495b9e0073bc140b87e5c06
2011-09-29 17:25:48 -05:00
Attila Nagy
380d64ecb1 Multithreaded encoder, late sync loopfilter
Sync with loopfilter thread just at the beginning of next frame encoding.
This returns control to application faster and allows a better multicore scaling.
When PSNR packets are generated the final filtered frame is needed imediatly
so we cannot delay the sync.

Change-Id: I288d97b5e331d41d6f5bb49d97986fa12ac6f066
2011-09-29 10:06:24 +03:00
John Koleszar
6f9457ec12 Merge "clamp_mvs() using the wrong motion vector information" 2011-09-22 11:54:15 -07:00
John Koleszar
3c85c532bb Merge changes Ie650e9b8,I2427e494
* changes:
  vpxenc: get version string programatically
  Install missing default_coef_probs.h
2011-09-22 11:18:00 -07:00
Johann
9f41a8b0aa Merge "Replace vpx_ports/config.h with vpx_config.h" 2011-09-22 09:30:18 -07:00
John Koleszar
4a6ac727fe Install missing default_coef_probs.h
Make sure that this header is listed as one of the sources, so that it
will be installed if necessary.

Change-Id: I2427e494488126b179151dc21043c1e2c8ba5991
2011-09-22 11:08:24 -04:00
Attila Nagy
1a7d25a484 Replace vpx_ports/config.h with vpx_config.h
Just a clean-up.

Change-Id: Iea5b6dc925dcfa7db548bc1ab1a13d26ed5a2c9a
2011-09-22 13:33:54 +03:00
Fritz Koenig
bd0c3409a8 Move neon only arm functions under arm/neon.
These files don't contain generic arm code, so should
only be compiled by neon.

Change-Id: Ie712823aa04d4235e7cfe7a3b725e73ee4c3e564
2011-09-20 10:51:06 -07:00
Johann
6829e62718 Merge "NEON FDCT updated to match current C code" 2011-09-20 09:51:05 -07:00
Johann
86e07525d5 Merge "NEON walsh transform updated to match C" 2011-09-20 09:50:42 -07:00
Johann
3a16276cf7 Merge "Updated ARMv6 forward transforms to match C" 2011-09-20 09:50:36 -07:00
Johann
fdd51829b1 Merge "Fixed armv5te multiplications" 2011-09-20 09:50:19 -07:00
Tero Rintaluoma
0c2529a812 NEON FDCT updated to match current C code
- Removed fast_fdct4x4_neon and fast_fdct8x4_neon
- Uses now short_fdct4x4 and short_fdct8x4
- Gives ~1-2% speed-up on Cortex-A8/A9

Change-Id: Ib62f2cb2080ae719f8fa1d518a3a5e71278a41ec
2011-09-20 10:20:55 +03:00
Tero Rintaluoma
3c19bc3fb3 Fixed armv5te multiplications
Rd and Rm registers should be different in 'mul'. This register
combination results in unpredictable behaviour. GCC will give
a warning and RVCT an error in this case.

Restriction applies only to armv5 targets and not for armv6 and above.

Change-Id: I378d17c51e1f16a6820814fbed43e115aaabb03e
2011-09-20 09:59:27 +03:00
Stefan Holmer
e529a825f7 Fix necessary for input partitions iface to match the RTP profile
These changes fixes a glitch between the RTP profile and the input
partitions interface. Since there's no way for the user to know the
actual number of partitions, the decoder have to read the
multi_token_paritition bits also when input partitions mode is
enabled.

Included are also a couple of fixes for issues with independent
partitions and uninitialized memory reads.

Change-Id: I6f93b15287d291169ed681898ed3fbcc5dc81837
2011-09-19 15:00:21 +02:00
Tero Rintaluoma
4c3ad66b7f Updated ARMv6 forward transforms to match C
- Updated walsh transform to match C
  (based on Change Id24f3392)
- Changed fast_fdct4x4 and 8x4 to short_fdct4x4 and 8x4
  correspondingly

Change-Id: I704e862f40e315b0a79997633c7bd9c347166a8e
2011-09-19 10:26:59 +03:00
Tero Rintaluoma
2a4b2a000c NEON walsh transform updated to match C
Modified original patch If2f07220885c4c3a0cae0dace34ea0e36124f001
according to comments. Scheduled code a little bit to prevent some
interlocks.

Change-Id: I338f02b881098782f82af63d97f042b85e63e902
2011-09-19 10:15:33 +03:00
John Koleszar
35ce4eb01d Merge "Fixes the boundary checks for extrapolated and interpolated MVs." 2011-09-16 08:09:44 -07:00
Scott LaVarnway
c0ee870b0a clamp_mvs() using the wrong motion vector information
In the "Removed bmi copy to/from BLOCKD" commit, the copy
to the bmi in BLOCKD was eliminated.  The clamp_mvs() used
the bmi in BLOCKD, which now contains incorrect values.  This
patch fixes this problem.

Change-Id: I8eca1eaf4015052b0b63e90876f7ad321aba7cff
2011-09-16 11:03:53 -04:00
Stefan Holmer
b854bbd844 Fixes the boundary checks for extrapolated and interpolated MVs.
Change-Id: I5b47d39d1604f2650d2f2d1ca2a3f40843c8e1ea
2011-09-16 11:58:57 +02:00
Scott LaVarnway
5bc7b3a68e Fixed encoder crash
caused by the "Removed bmi copy to/from BLOCKD" commit.

Change-Id: I9fae71bdc34c8ecc07bb81cd3ccf498b91ce3ec7
2011-09-13 11:46:33 -04:00
Scott LaVarnway
c4b9089bb9 Merge "Skip computation of distortion in vp8_pick_inter_mode if active_map is used" 2011-08-31 07:18:52 -07:00
Scott LaVarnway
222c72e50f Merge "Removed bmi copy to/from BLOCKD" 2011-08-31 06:57:20 -07:00
Alpha Lam
0e05f2c6c9 Skip computation of distortion in vp8_pick_inter_mode if active_map is used
If a block is marked to be inactive then set distortion to 0.

Change-Id: Ib415f19642a2ff7b5cf5cfaedd60ebbd79732272
2011-08-31 14:06:55 +01:00
John Koleszar
800b70a3bf Merge "Recalculate zbin_extra only if regular quantizer is being used" 2011-08-30 12:49:24 -07:00
Alpha Lam
bc9293b815 Recalculate zbin_extra only if regular quantizer is being used
vp8_update_zbin_extra() is called all the time even though the fast
quantizer doesn't use it. Skip this call if fast quantizer is used.

Change-Id: Ia711c38431930cc2486cf59b8466060ef0e9d9db
2011-08-30 19:23:34 +01:00
Yunqing Wang
1f20202e2c Minor modification on key frame decision
This change makes sure that no key frame recoding in real-time mode
even if CONFIG_REALTIME_ONLY is not configured.

Change-Id: Ifc34141f3217a6bb63cc087d78b111fadb35eec2
2011-08-25 16:54:45 -04:00
Fritz Koenig
4797a97215 Quiet warning by removing unused variable.
fwd_boost_score was not being computed or
referenced, so remove declaration.

Change-Id: Iece36cde1ec113e3c6afaff1407d24cdf12bd0a8
2011-08-24 15:47:09 -07:00
Scott LaVarnway
b870947d42 Removed bmi copy to/from BLOCKD
for SPLITMV and B_PRED modes.  Modified code to use the bmi
found in mode_info_context instead of BLOCKD.  On the decode
side, the uvmvs are calculated only when required, instead of
every macroblock.  This is WIP. (bmi should eventually be
removed from BLOCKD)
Small performance gains noticed for RT encodes and decodes.(VGA)

Change-Id: I2ed7f0fd5ca733655df684aa82da575c77a973e7
2011-08-24 14:42:26 -04:00
Fritz Koenig
112bd4e2b4 Fix naming of sse2 idct functions.
Prepend idct function names with vp8_
so that under profiling they show up
associated with libvpx.

Change-Id: I4fe357b50236cb7730a4cc00164c0a3487a1d8b4
2011-08-24 10:25:32 -07:00
Scott LaVarnway
1de5da80c9 Merge "Faster vp8_default_coef_probs" 2011-08-24 07:52:10 -07:00
Johann
85358d04cd Fix data accesses for simple loopfilters
The data that the simple horizontal loopfilter reads is aligned, treat
it accordingly.

For the vertical, we only use the bottom 4 bytes, so don't read in 16
(and incur the penalty for unaligned access).

This shows a small improvement on older processors which have a
significant penalty for unaligned reads.

postproc_mmx.c is unused

Change-Id: I87b29bbc0c3b19ee1ca1de3c4f47332a53087b3d
2011-08-23 20:42:45 -04:00
Fritz Koenig
c5f890af2c Use local labels for jumps/loops in x86 assembly.
Prepend . to local labels in assembly code.  This
allows non unique labels within a file.  Also
makes profiling information more informative
by keeping the function name with the loop name.

Change-Id: I7a983cb3a5ba2413d5dafd0a37936b268fb9e37f
2011-08-23 09:05:29 -07:00
Fritz Koenig
694d4e7777 Reclassify optimized ssim calculations as SSE2.
Calculations were incorrectly classified as either
SSE3 or SSSE3.  Only using SSE2 instructions.
Cleanup function names and make non-RTCD code work
as well.

Change-Id: I48ad0218af0cc51c5078070a08511dee43ecfe09
2011-08-22 12:36:28 -07:00
Fritz Koenig
b7a6f1d20e Merge "Revert "Reclasify optimized ssim calculations as SSE2."" 2011-08-22 12:32:12 -07:00
Fritz Koenig
734b1b2041 Revert "Reclasify optimized ssim calculations as SSE2."
This reverts commit 01376858cd
2011-08-22 11:31:12 -07:00
Fritz Koenig
f8e3d23b99 Merge "Reclasify optimized ssim calculations as SSE2." 2011-08-22 09:20:33 -07:00
Fritz Koenig
01376858cd Reclasify optimized ssim calculations as SSE2.
Calculations were incorrectly classified as either
SSE3 or SSSE3.  Only using SSE2 instructions.
Cleanup function names and make non-RTCD code work
as well.

Change-Id: I29f5c2ead342b2086a468029c15e2c1d948b5d97
2011-08-19 08:51:27 -07:00
John Koleszar
edec5eb5e7 Merge "Copy less when active map is in use" 2011-08-19 07:31:00 -07:00
Alpha Lam
4e8d35a461 Copy less when active map is in use
When active map is specified and the current frame is not a key frame,
golden frame nor a altref frame then copy only those active regions.

This significantly reduces encoding time by as much as 19% on the test
system where realtime encoding is used. This is particularly useful
when the frame size is large (e.g. 2560x1600) and there's only a few
action macroblocks.

Change-Id: If394a813ec2df5a0201745d1348dbde4278f7ad4
2011-08-19 10:29:41 -04:00
Paul Wilkins
744f482350 Small boost to every other frame.
Instead of a single mid GF boost apply a few extra bits to
every other frame. This gives a very small average metrics
improvement on both derf and YT sets.

Also use min GF interval as min KF interval.

Change-Id: Iee238b8cae0ffaed850a5a944ac825cee18da485
2011-08-17 14:14:23 +01:00
Scott LaVarnway
19987dcbfa Faster vp8_default_coef_probs
Copies from a generated table instead of building the
default coeff probabilities during runtime.

Change-Id: I4d9551ea3a2d7d4a4f7ce9eda006495221a8de50
2011-08-16 16:21:21 -04:00
John Koleszar
9cc1611588 Merge v0.9.7-p1 release int 'origin/master'
Change-Id: I93388d2f8846615ad1e26b975308c5e96b9b1918
2011-08-15 17:10:01 -04:00
Stefan Holmer
99d870a472 Don't set the bmi mode when doing error concealment
Since the block will be interpreted as an inter block, the mode will
be interpreted as a motion vector, resulting in bad concealment.

Change-Id: Ifcc685ae1cc883492bce6dbd61e418d91a89b053
2011-08-15 11:46:04 -04:00
Stefan Holmer
ff35649758 Don't set the bmi mode when doing error concealment
Since the block will be interpreted as an inter block, the mode will
be interpreted as a motion vector, resulting in bad concealment.

Change-Id: Ifcc685ae1cc883492bce6dbd61e418d91a89b053
2011-08-15 09:56:07 +02:00
John Koleszar
e96131705a Revert "Improved 1-pass CBR rate control"
This reverts commit b5ea2fbc2c. Further
testing showed noticable keyframe popping in some cases, reverting this
for now to give time for a proper fix.

Conflicts:

	vp8/encoder/onyx_if.c
	vp8/encoder/ratectrl.c

Change-Id: I159f53d1bf0e24c035754ab3ded8ccfd58fd04af
2011-08-12 14:51:36 -04:00
John Koleszar
a4c2211ea3 Propagate macroblock MV to subblocks for error concealment
EC expects the subblock MVs to be populated, but
f1d6cc79e4 removed this code. This
commit restores it, protected by CONFIG_ERROR_CONCEALMENT. May move this
to the EC code more directly in the future.

Change-Id: I44f8f985720cb9a1bf222e59143f9e69abf56ad2
2011-08-12 14:49:35 -04:00
Stefan Holmer
a609be5633 Disable error concealment until first key frame is decoded
When error concealment is enabled the first key frame must
be successfully received before error concealment is activated.
Error concealment will be activated when the delta following
delta frame is received.

Also fixed a couple of bugs related to error tracking in
multi-threading. And avoiding decoding corrupt residual
when we have multiple non-resilient partitions.

Change-Id: I45c4bb296e2f05f57624aef500a874faf431a60d
2011-08-12 14:49:34 -04:00
John Koleszar
cdae03a4eb Fix potential OOB read with Error Concealment
This patch fixes an OOB read when error concealment is enabled and the
partition sizes are corrupt. The partition size read from the bitstream
was not being validated in EC mode.

Change-Id: Ia81dfd4bce1ab29ee78e42320abe52cee8318974
2011-08-12 14:49:34 -04:00
John Koleszar
4645c89889 Merge "Disable error concealment until first key frame is decoded" 2011-08-12 11:45:26 -07:00
John Koleszar
91206793c2 Propagate macroblock MV to subblocks for error concealment
EC expects the subblock MVs to be populated, but
f1d6cc79e4 removed this code. This
commit restores it, protected by CONFIG_ERROR_CONCEALMENT. May move this
to the EC code more directly in the future.

Change-Id: I44f8f985720cb9a1bf222e59143f9e69abf56ad2
2011-08-12 11:34:40 -04:00
Stefan Holmer
3e10be93f2 Disable error concealment until first key frame is decoded
When error concealment is enabled the first key frame must
be successfully received before error concealment is activated.
Error concealment will be activated when the delta following
delta frame is received.

Also fixed a couple of bugs related to error tracking in
multi-threading. And avoiding decoding corrupt residual
when we have multiple non-resilient partitions.

Change-Id: I45c4bb296e2f05f57624aef500a874faf431a60d
2011-08-12 16:10:04 +02:00
John Koleszar
810a06b12c Fix potential OOB read with Error Concealment
This patch fixes an OOB read when error concealment is enabled and the
partition sizes are corrupt. The partition size read from the bitstream
was not being validated in EC mode.

Change-Id: Ia81dfd4bce1ab29ee78e42320abe52cee8318974
2011-08-11 18:07:03 -04:00
Yunqing Wang
b84e8f20c3 Merge "Adjust half-pixel only search" 2011-08-05 12:15:32 -07:00
John Koleszar
238dae8604 Fix source buffer selection
This patch fixes a bug in the interaction between the recode loop and
spatial resampling. If the codec was in a spatial resampling state,
and a subsequent iteration of the recode loop disables resampling,
then the source buffer must be reset to the unscaled source.

Change-Id: I4e4cd47b943f6cd26a47449dc7f4255b38e27c77
2011-08-03 16:13:15 -04:00
Yunqing Wang
b9f19f8917 Adjust half-pixel only search
Changed motion search in vp8_find_best_half_pixel_step() to be the
same as in vp8_find_best_sub_pixel_step(), which checks 5 points
instead of 8 points. This only affects real-time mode with
cpu-used >=9. Tests showed it gives 2% encoding speedup with
a quality loss(psnr) of up to 0.5%.

Change-Id: I16049cad1535002346d46cfdfad345bfc3dc5146
2011-08-03 11:51:07 -04:00
Johann
30e5deae5d update extend frame borders
the neon code made several assumptions which were broken by a recent
change: https://review.webmproject.org/2676

update the code with new assumptions and guard them with a compile time
assert

Change-Id: I32a8378030759966068f34618d7b4b1b02e101a0
2011-08-02 19:26:46 -04:00
James Berry
27ee521753 include asm_com/dec_offsets for make dist
Change-Id: Ia1ad66066a24c01915cd9e3ff75c7e070cc984c8
2011-08-02 13:42:03 -04:00
John Koleszar
f475f0c1bb Merge "include the arm header files in make dist" into cayuga 2011-08-02 05:21:10 -07:00
John Koleszar
06c3d5bb9a Fix building with --disable-postproc
Change-Id: I7e6bc28e7974a376da747300744e0dd5dc1d21e9
2011-08-01 17:50:23 -04:00
Johann
3e8c6d3d35 include the arm header files in make dist
Change-Id: Ibcf5b4b14153f65ce1b53c3bfba87ad2feb17bbd
2011-08-01 17:20:21 -04:00
John Koleszar
6f080f9cec Merge "Convert rc_max_intra_bitrate_pct to control" 2011-07-29 11:57:48 -07:00
John Koleszar
1f71d2e2c8 Correctly track sharpness in vp8cx_pick_filter_level_fast
Make sure to update last_sharpness_level from the current
sharpness_level whenever it changes.

Change-Id: I0258d2f5b11a407abf6176a8d4c4994d925943f0
2011-07-29 12:27:03 -04:00
John Koleszar
1654ae9a2a Convert rc_max_intra_bitrate_pct to control
Since this is the only ABI incompatible change since the last release,
convert it to use the control interface instead. The member of the
configuration struct is replaced with the VP8E_SET_MAX_INTRA_BITRATE_PCT
control.

More significant API changes were expected to be forthcoming when this
control was first introduced, and while they continue to be expected,
it's not worth breaking compatibility for only this change.

Change-Id: I799d8dbe24c8bc9c241e0b7743b2b64f81327d59
2011-07-28 09:17:35 -04:00
Yunqing Wang
2f2302f8d5 Preload reference area in sub-pixel motion search (real-time mode)
This change implemented same idea in change "Preload reference area
to an intermediate buffer in sub-pixel motion search." The changes
were made to vp8_find_best_sub_pixel_step() and vp8_find_best_half
_pixel_step() functions which are called when speed >= 5. Test
result (using tulip clip):

1. On Core2 Quad machine(Linux)
rt mode, speed (-5 ~ -8), encoding speed gain: 2% ~ 3%
rt mode, speed (-9 ~ -11), encoding speed gain: 1% ~ 2%
rt mode, speed (-12 ~ -14), no noticeable encoding speed gain

2. On Xeon machine(Linux)
Test on speed (-5 ~ -14) didn't show noticeable speed change.

Change-Id: I21bec2d6e7fbe541fcc0f4c0366bbdf3e2076aa2
2011-07-27 14:19:10 -04:00
Yunqing Wang
f11613b620 Merge "Fix range checks in motion search" 2011-07-27 09:34:13 -07:00
Yunqing Wang
bde2afbe23 Fix range checks in motion search
There were some situations that the start motion vectors were
out of range. This fix adjusted range checks to make sure they
are checked and clamped.

Change-Id: Ife83b7fed0882bba6d1fa559b6e63c054fd5065d
2011-07-27 10:37:33 -04:00
John Koleszar
db8f0d2ca9 Merge "cosmetics: consistently use [u]int64_t" 2011-07-26 12:57:43 -07:00
James Zern
b45065d38b cosmetics: consistently use [u]int64_t
Removes mixed usage of (unsigned) long long and INT64.
Fixes Issue #208.

Change-Id: I220d3ed5ce4bb1280cd38bb3715f208ce23cf83a
2011-07-26 11:34:36 -07:00
Johann
ca7e346669 Merge ""Eliminated TOKENEXTRABITS" broke the windows build." 2011-07-26 06:34:31 -07:00
Scott LaVarnway
a11624497c "Eliminated TOKENEXTRABITS" broke the windows build.
Fixed.

Change-Id: I3348e8dbcaee6ace263af413701101d77636e5df
2011-07-26 09:33:16 -04:00
Scott LaVarnway
4894b45ced Merge "Eliminated TOKENEXTRABITS" 2011-07-25 14:35:58 -07:00
Scott LaVarnway
76eb402668 Eliminated TOKENEXTRABITS
Noticed small performance gains, depending on material.

Change-Id: I334369f6312bc19aa73481fc3f790ab181e11867
2011-07-25 17:11:24 -04:00
Yunqing Wang
5b0de48ddd Merge "Use CONFIG_FAST_UNALIGNED consistently in codec" 2011-07-25 12:40:50 -07:00
Yunqing Wang
fe270dd527 Specify size for argument pushed to stack
The change fixes building error on Win64.

Change-Id: I63d25b26220c4da8a98ca2e36530cbb802468e6b
2011-07-25 11:30:45 -04:00
Yunqing Wang
65dfcf4696 Use CONFIG_FAST_UNALIGNED consistently in codec
CONFIG_FAST_UNALIGNED is enabled by default. Disable it if it is
not supported by hardware.

Change-Id: I7d6905ed79fed918bca074bd62820b0c929d81ab
2011-07-25 10:11:24 -04:00
Johann
773bcc300d Merge "fix sharpness bug and clean up" 2011-07-22 09:34:55 -07:00
Johann
a04ed0e8f3 fix sharpness bug and clean up
sharpness was not recalculated in vp8cx_pick_filter_level_fast

remove last_filter_type. all values are calculated, don't need to update
the lfi data when it changes.

always use cm->sharpness_level. the extra indirection was annoying.

don't track last frame_type or sharpness_level manually. frame type
only matters for motion search and sharpness_level is taken care of in
frame_init

move function declarations to their proper header

Change-Id: I7ef037bd4bf8cf5e37d2d36bd03b5e22a2ad91db
2011-07-22 12:33:57 -04:00
Yunqing Wang
829179e888 Merge "Preload reference area to an intermediate buffer in sub-pixel motion search" 2011-07-22 06:56:15 -07:00
Yunqing Wang
20bd1446c0 Preload reference area to an intermediate buffer in sub-pixel motion search
In sub-pixel motion search, the search range is small(+/- 3 pixels).
Preload whole search area from reference buffer into a 32-byte
aligned buffer. Then in search, load reference data from this buffer
instead. This keeps data in cache, and reduces the crossing cache-
line penalty. For tulip clip, tests on Intel Core2 Quad machine(linux)
showed encoder speed improvement:
  3.4%   at --rt --cpu-used =-4
  2.8%   at --rt --cpu-used =-3
  2.3%   at --rt --cpu-used =-2
  2.2%   at --rt --cpu-used =-1

Test on Atom notebook showed only 1.1% speed improvement(speed=-4).
Test on Xeon machine also showed less improvement, since unaligned
data access latency is greatly reduced in newer cores.

Next, I will apply similar idea to other 2 sub-pixel search functions
for encoding speed > 4.

Make this change exclusively for x86 platforms.

Change-Id: Ia7bb9f56169eac0f01009fe2b2f2ab5b61d2eb2f
2011-07-22 09:28:06 -04:00
John Koleszar
2bdda84e37 Merge "Increase chrow row alignment to 16 bytes." 2011-07-21 07:32:39 -07:00
Yunqing Wang
c5fe641179 Merge "Add improvements made in good-quality mode to real-time mode" 2011-07-21 07:27:09 -07:00
Timothy B. Terriberry
7d1b37cdac Increase chrow row alignment to 16 bytes.
This is done by expanding luma row to 32-byte alignment, since
 there is currently a bunch of code that assumes that
 uv_stride == y_stride/2 (see, for example, vp8/common/postproc.c,
 common/reconinter.c, common/arm/neon/recon16x16mb_neon.asm,
 encoder/temporal_filter.c, and possibly others; I haven't done a
 full audit).
It also uses replaces the hardcoded border of 16 in a number of
 encoder buffers with VP8BORDERINPIXELS (currently 32), as the
 chroma rows start at an offset of border/2.
Together, these two changes have the nice advantage that simply
 dumping the frame memory as a contiguous blob produces a valid,
 if padded, image.

Change-Id: Iaf5ea722ae5c82d5daa50f6e2dade9de753f1003
2011-07-20 10:20:31 -07:00
Attila Nagy
0afcc76971 encoder: don't set the fragment bit for the last partition
Change-Id: Icb4e4f0d7c3074a8507852178be87541a1cb5bac
2011-07-20 14:09:42 +03:00
Scott LaVarnway
b2d9700f53 Merge "Moved vp8_encode_bool into boolhuff.h" 2011-07-19 08:15:14 -07:00
Johann
6afafc313c remove old armv5 code
armv5 dequantizer is not referenced

Change-Id: Id1cc617dcee35ebd6a406816ec6aaa26e8bbc8ad
2011-07-19 09:20:38 -04:00
Scott LaVarnway
a25f6a9c88 Moved vp8_encode_bool into boolhuff.h
allowing the compiler to inline this function.  For real-time
encodes, this gave a boost of 1% to 2.5%, depending on the
speed setting.

Change-Id: I3929d176cca086b4261267b848419d5bcff21c02
2011-07-19 09:17:25 -04:00
John Koleszar
b5ea2fbc2c Improved 1-pass CBR rate control
This patch attempts to improve the handling of CBR streams with
respect to the short term buffering requirements. The "buffer level"
is changed to be an average over the rc buffer, rather than a long
running average. Overshoot is also tracked over the same interval
and the golden frame targets suppressed accordingly to correct for
overly aggressive boosting.

Testing shows that this is fairly consistently positive in one
metric or another -- some clips that show significant decreases
in quality have better buffering characteristics, others show
improvenents in both.

Change-Id: I924c89aa9bdb210271f2e03311e63de3f1f8f920
2011-07-18 11:48:05 -04:00
Scott LaVarnway
e68894fa03 Merge "Tokenize MB optimized" 2011-07-15 07:54:14 -07:00
Tero Rintaluoma
4e82f01547 Tokenize MB optimized
Optimized C-code of the following functions:
 - vp8_tokenize_mb
 - tokenize1st_order_b
 - tokenize2nd_order_b
Gives ~1-5% speed-up for RT encoding on Cortex-A8/A9
depending on encoding parameters.

Change-Id: I6be86104a589a06dcbc9ed3318e8bf264ef4176c
2011-07-15 11:26:54 +03:00
James Berry
6b6f367c3d bug fix vpx_copy_and_extend_frame size issue
vpx_copy_and_extend_frame could incorrectly
resize uv frames which could result in a crash.

Change-Id: Ie96f7078b1e328b3907a06eebeee44ca39a2e898
2011-07-14 15:58:15 -04:00
John Koleszar
04dce631a2 Remove unused speed features
min_fs_radius, max_fs_radius, full_freq were set but never read.

Change-Id: I82657f4e7f2ba2acc3cbc3faa5ec0de5b9c6ec74
2011-07-14 14:20:25 -04:00
Yunqing Wang
f1f28535c3 Merge "Fix unnecessary casting of B_PREDICTION_MODE (issue 349)" 2011-07-13 13:32:57 -07:00
Yunqing Wang
139577f937 Fix unnecessary casting of B_PREDICTION_MODE (issue 349)
Minor fix.

Change-Id: Iaf93f6e47e882a33c479e57c7a0d0bf321e291c0
2011-07-13 15:52:07 -04:00
Yunqing Wang
0e9a6ed72a Add improvements made in good-quality mode to real-time mode
Several improvements we made in good-quality mode can be added
into real-time mode to speed up encoding in speed 1, 2, and 3
with small quality loss. Tests using tulip clip showed:

--rt --cpu-used=-1
(before change)
PSNR: 38.028
time: 1m33.195s
(after change)
PSNR: 38.014
time: 1m20.851s

--rt --cpu-used=-2
(before change)
PSNR: 37.773
time: 0m57.650s
(after change)
PSNR: 37.759
time: 0m54.594s

--rt --cpu-used=-3
(before change)
PSNR: 37.392
time: 0m42.865s
(after change)
PSNR: 37.375
time: 0m41.949s

Change-Id: I76ab2a38d72bc5efc91f6fe20d332c472f6510c9
2011-07-13 14:51:02 -04:00
Fritz Koenig
84c3cd79d1 Merge "Reduce motion vector search on alt-ref frame." 2011-07-13 10:07:30 -07:00
Johann
211694f67e Merge "update x86 asm for loopfilter" 2011-07-13 04:10:03 -07:00
Johann
8f910594bd Merge "Update armv6 loopfilter to new interface" 2011-07-13 04:09:55 -07:00
Johann
1a219c22b1 Merge "Update armv7 loopfilter to new interface" 2011-07-13 04:09:42 -07:00
Johann
d9b825cff2 Merge "New loop filter interface" 2011-07-13 04:09:26 -07:00
Attila Nagy
c231b0175d Update armv6 loopfilter to new interface
Change-Id: I5fe581d797571a7a9432fbd17fc557591d0c1afa
2011-07-12 12:14:51 +03:00
Attila Nagy
283b0e25ac Update armv7 loopfilter to new interface
Change-Id: I65105a9c63832669237e6a6a7fcb4ea3ea683346
2011-07-12 12:12:25 +03:00
Fritz Koenig
ede0b15c9d Reduce motion vector search on alt-ref frame.
Clamp mv search to accomodate subpixel filtering
of UV mv.

Change-Id: Iab3ed405993ef6bf779ad7cf60863153068fb7d1
2011-07-11 09:05:43 -07:00
Yunqing Wang
587ca06da9 Minor change in pick_inter_mode()
Scott suggested to move vp8_mv_pred() under "case NEWMV" to save
extra checks.

Change-Id: I09e69892f34a08dd425a4d81cfcc83674e344a20
2011-07-08 14:08:45 -04:00
Yunqing Wang
e83d36c053 Merge "Adjust full-pixel clamping and motion vector limit calculation" 2011-07-08 08:39:32 -07:00
Yunqing Wang
40991faeae Adjust full-pixel clamping and motion vector limit calculation
Do mvp clamping in full-pixel precision instead of 1/8-pixel
precision to avoid error caused by right shifting operation.
Also, further fixed the motion vector limit calculation in change:
b748045470

Change-Id: Ied88a4f7ddfb0476eb9f7afc6ceeddbf209fffd7
2011-07-08 11:34:28 -04:00
Johann
01433c5043 update x86 asm for loopfilter
Change-Id: I1ed739522db7c00c189851c7095c1b64ef6412ce
2011-07-08 09:23:38 -04:00
Johann
6ae12c415e Merge "clean up warnings when building arm with rtcd" 2011-07-08 05:16:09 -07:00
Attila Nagy
622958449b New loop filter interface
Separate simple filter with reduced no. of parameters.
MB filter level picking based on precalculated table. Level table updated for
each frame. Inside and edge limits precalculated and updated just when
sharpness changes. HEV threshhold is constant.
ARM targets use scalars and others vectors.

Change works only with --target=generic-gnu
All other targets have to be updated!

Change-Id: I6b73aca6b525075b20129a371699b2561bd4d51c
2011-07-08 09:31:41 +03:00
John Koleszar
973a9c075d Merge "Set VPX_FRAME_IS_DROPPABLE" 2011-07-07 08:11:05 -07:00
John Koleszar
37de0b8bdf Set VPX_FRAME_IS_DROPPABLE
Allow the encoder to inform the application that the encoded frame will not
be used as a reference.

Change-Id: I90e41962325ef73d44da03327deb340d6f7f4860
2011-07-07 10:38:45 -04:00
John Koleszar
b4f70084cc Merge "Properly use GET_GOT/RESTORE_GOT when using GLOBAL()." 2011-07-01 07:14:34 -07:00
Ronald S. Bultje
c8a23ad3f4 Properly use GET_GOT/RESTORE_GOT when using GLOBAL().
This should fix binaries using PIC on x86-32. Also should
fix issue 343.

Change-Id: I591de3ad68c8a8bb16054bd8f987a75b4e2bad02
2011-06-30 14:04:27 -07:00
Yunqing Wang
ae8aa836d5 Merge "Copy macroblock data to a buffer before encoding it" 2011-06-30 11:14:24 -07:00
Yunqing Wang
80c3bbf657 Merge "Bug fix in motion vector limit calculation" 2011-06-30 09:52:03 -07:00
Yunqing Wang
b748045470 Bug fix in motion vector limit calculation
Motion vector limits are calculated using right shifts, which
could give wrong results for negative numbers. James Berry's
test on one clip showed encoder produced some artifacts. This
change fixed that.

Change-Id: I035fc02280b10455b7f6eb388f7c2e33b796b018
2011-06-30 11:20:13 -04:00
Johann
3e4a80cc35 Merge "remove incorrect initialization" 2011-06-30 07:59:08 -07:00
Paul Wilkins
eacaabc592 Merge "Change to arf boost calculation." 2011-06-29 10:03:57 -07:00
Paul Wilkins
11694aab66 Change to arf boost calculation.
In this commit I have added an experimental function
that tests prediction quality either side of a central position
to calculate a suggested boost number for an ARF frame.

The function is passed an offset from the current position and
a number of frames to search forwards and backwards.
It returns a forward, backward and compound boost number.

The new code can be deactivated using #define NEW_BOOST 0

In its current default state the code searches forwards and backwards
from the proposed  position of the next alt ref.

The the old code used a boost number calculated by scanning forward
from the previous GF up to the proposed alt ref frame position.

I have also added some code to try and prevent placement of a gf/arf
where there is a brief flash.

Change-Id: I98af789a5181148659f10dd5dd2ff2d4250cd51c
2011-06-29 18:01:25 +01:00
Johann
fe53107fda remove incorrect initialization
Values were set, then reset. Only set them once.

Change-Id: Iaf43c8467129f2f261f04fa9188b603aa46216b5
2011-06-29 11:54:27 -04:00
Johann
6611f66978 clean up warnings when building arm with rtcd
Change-Id: I3683cb87e9cb7c36fc22c1d70f0799c7c46a21df
2011-06-29 10:51:41 -04:00
John Koleszar
f3a13cb236 Merge "Use MAX_ENTROPY_TOKENS and ENTROPY_NODES more consistently" 2011-06-29 07:29:59 -07:00
Johann
dc004e8c17 Merge "Avoid text relocations in ARM vp8 decoder" 2011-06-28 16:34:10 -07:00
Johann
02c30cdeef Merge "utilize preload in ARMv6 MC/LPF/Copy routines" 2011-06-28 16:33:45 -07:00
John Koleszar
b32da7c3da Use MAX_ENTROPY_TOKENS and ENTROPY_NODES more consistently
There were many instances in the code of vp8_coef_tokens and
vp8_coef_tokens-1, which was a preprocessor macro despite the naming
convention. Replace these with MAX_ENTROPY_TOKENS and ENTROPY_NODES,
respectively.

Change-Id: I72c4f6c7634c94e1fa066cd511471e5592c748da
2011-06-28 17:03:55 -04:00
John Koleszar
9bcf07ae4a Merge "Simplify decode_macroblock." 2011-06-28 12:54:25 -07:00
Gaute Strokkenes
81c0546407 Simplify decode_macroblock.
Change-Id: Ieb2f3827ae7896ae594203b702b3e8fa8fb63d37
2011-06-28 17:01:14 +01:00
Stefan Holmer
7296b3f922 New ways of passing encoded data between encoder and decoder.
With this commit frames can be received partition-by-partition
from the encoder and passed partition-by-partition to the
decoder.

At the encoder-side this makes it easier to split encoded
frames at partition boundaries, useful when packetizing
frames. When VPX_CODEC_USE_OUTPUT_PARTITION is enabled,
several VPX_CODEC_CX_FRAME_PKT packets will be returned
from vpx_codec_get_cx_data(), containing one partition
each. The partition_id (starting at 0) specifies the decoding
order of the partitions. All partitions but the last has
the VPX_FRAME_IS_FRAGMENT flag set.

At the decoder this opens up the possibility of decoding partition
N even though partition N-1 was lost (given that independent
partitioning has been enabled in the encoder) if more info
about the missing parts of the stream is available through
external signaling.

Each partition is passed to the decoder through the
vpx_codec_decode() function, with the data pointer pointing
to the start of the partition, and with data_sz equal to the
size of the partition. Missing partitions can be signaled to
the decoder by setting data != NULL and data_sz = 0. When
all partitions have been given to the decoder "end of data"
should be signaled by calling vpx_codec_decode() with
data = NULL and data_sz = 0.

The first partition is the first partition according to the
VP8 bitstream + the uncompressed data chunk + DCT address
offsets if multiple residual partitions are used.

Change-Id: I5bc0682b9e4112e0db77904755c694c3c7ac6e74
2011-06-28 11:10:17 -04:00
Stefan Holmer
4cb0ebe5b2 Adding support for independent partitions
Adding support in the encoder for generating
independent residual partitions by forcing
equal probabilities over the prev coef entropy
contexts.

Change-Id: I402f5c353255f3ca20eae2620af739f6a498cd21
2011-06-28 11:10:17 -04:00
Mike Hommey
e3f850ee05 Avoid text relocations in ARM vp8 decoder
The current code stores pointers to coefficient tables and loads them to
access the tables contents. As these pointers are stored in the code
sections, it means we end up with text relocations. eu-findtextrel will
thus complain about code not compiled with -fpic/-fPIC.

Since the pointers are stored in the code sections, we can actually cheat
and let the assembler generate relative addressing when accessing the
coefficient tables, and just load their location with adr.

Change-Id: Ib74ae2d3f2bab80b29991355f2dbe6955f38f6ae
2011-06-28 09:11:40 +02:00
Fritz Koenig
be99868bd1 Fix after removal of B_MODE_INFO
Change Ieb746989: Removed B_MODE_INFO missed this.

Change-Id: I32202555581cc2a5d45e729c6650ada4d2df55d3
2011-06-27 09:43:21 -07:00
Johann
8a9a11e8dc Merge "configuration, support disabling any subset of ARM arch" 2011-06-27 08:55:18 -07:00
Stefan Holmer
ba0822ba96 Adding support for error concealment in multi-threaded decoding
Also includes a couple of error concealment bug fixes:
- the segment_id wasn't properly initialized when missing
- when interpolating and no neighbors are found, set to zero
- clear the qcoef buffer when concealing an MB

Change-Id: Id79c876b41d78b559a2241e9cd0fd2cae6198f49
2011-06-27 09:03:06 -04:00
Adrian Grange
deca8cfc44 Fixed initialization of frame buffer ref counters
Only the first frame buffer ref counter was being initialized
because the index was fixed at 0 rather than using i.

Change-Id: Ib842298be4a5e3607f9e21c2cd4bfbee4054ffc4
2011-06-24 08:43:40 -07:00
Yunqing Wang
0d87098e08 Copy macroblock data to a buffer before encoding it
I got this idea from Pascal (Thanks). Before encoding a macroblock,
copy it to a 16x16 buffer, and then read source data from there
instead. This will help keep the source data in cache, and help
with the performance.

Change-Id: Id05f4cb601299150511d59dcba0ae62c49b5b757
2011-06-23 13:54:02 -04:00
John Koleszar
db67dcba6a Revert "Reduce overshoot in 1 pass rate control"
This reverts commit 212f618373.

Further testing shows that the overshoot accumulation/damping is too
aggressive on some clips. Allowing the accumulated overshoot to
decay and limiting to damping to golden frames shows some promise.
But some clips show significant overshoot in the buffer window, so
I think this still needs work.

Change-Id: Ic02a9ca34f55229f9cc04786f4fab54cdc1a3ef5
2011-06-23 11:52:12 -04:00
James Berry
2bd90c13a0 get/set reference buffer dimension check added
vp8_yv12_copy_frame_ptr() expects same size
buffers which was not previously gaurenteed.
Using an improperly allocated buffer would
cause a crash before.

Change-Id: I904982313ce9352474f80de842013dcd89f48685
2011-06-22 13:36:24 -04:00
Yaowu Xu
76495617e0 Merge "adjusting the calculation of errorperbit" 2011-06-21 09:47:42 -07:00
Scott LaVarnway
55c3963c88 Merge "Improved vp8dx_decode_bool" 2011-06-21 07:45:51 -07:00
Yunqing Wang
109c20299c Merge "Remove unnecessary bounds checking in motion search" 2011-06-21 07:23:24 -07:00
Attila Nagy
6f23f24afe configuration, support disabling any subset of ARM arch
Useful for leaving out any version specific asm files.

Change-Id: I233514410eb9d7ca88d2d2c839673122c507fa99
2011-06-21 10:39:01 +03:00
Yaowu Xu
10ed60dc71 adjusting the calculation of errorperbit
RDMULT/RDDIV defines a bit worth of distortion in term of sum squared
difference. This has also been used as errorperbit in subpixel motion
search, where the distortions computed as variance of the difference.
The variance of differences is different from sum squared differences
by amount of DC squared. Typically, for inter predicted MBs, this
difference averages around 10% between the two distortion, so this patch
introduces a 110% constant in deriving errorperbit from RDMULT/RDDIV.

Test on CIF set shows small but positive gain on overall PSNR (.03%)
and SSIM (.07%), overall impact on average PSNR is 0.

Change-Id: I95425f922d037b4d96083064a10c7cdd4948ee62
2011-06-20 16:32:30 -07:00
Scott LaVarnway
67a1f98c2c Improved vp8dx_decode_bool
Relocated the vp8dx_bool_decoder_fill() call, allowing
the compiler to produce better assembly code.  Tests
showed a 1 - 2 % performance boost (x86 using gcc)
for the 720p clip used.

Change-Id: Ic5a4eefed8777e6eefa007d4f12dfc7e64482732
2011-06-20 14:44:16 -04:00
Taekhyun Kim
458fb8f491 utilize preload in ARMv6 MC/LPF/Copy routines
About 9~10% decoding perf improvement on non-Neon ARM cpus

Change-Id: I7dc2a026764e84e9c2faf282b4ae113090326837
2011-06-17 14:04:53 -07:00
Yunqing Wang
2cd1c2855e Remove unnecessary bounds checking in motion search
The starting points are always within the limits, and bounds
checking on these points is not needed. For speed < 5, the
encoded result changes a little because different treatment
is taken while starting point equals the bounds.

Change-Id: I09a402d310f51e305a3519f1601b1d17b05c6152
2011-06-17 14:19:51 -04:00
John Koleszar
a60fc419f5 Merge "Use SSE as BPRED distortion metric consistently" 2011-06-17 09:48:32 -07:00
Ronald S. Bultje
87fd66bb0e Assign boost to GF bit allocation if past frame had no ARF.
Modify the second-pass code to provide a full golden-frame (GF) bit
allocation boost if the past GF group (GFG) had no alt-ref frame (ARF),
even if the current GFG does contain and ARF.

This mostly has no effect on clips, since switching ARFs on/off between
GFGs is not very common. Has a positive effect on e.g. cheer (+0.45 SSIM
at 600kbps) and football (+0.25 SSIM at 600kbps), particularly at high
bitrates. Has a negative effect (-0.04 SSIM at 300kbps) at pamphlet,
which appears only marginally related to this patch, and crew (-0.1 SSIM
at 700kbps).

Change-Id: I2e32899638b59f857e26efeac18a82e0c0b77089
2011-06-16 13:01:27 -04:00
John Koleszar
eb645abeac Merge "Disable specialcase for last frames if the sequence contains ARFs." 2011-06-16 09:56:05 -07:00
John Koleszar
5223016337 Merge "Remove redundant check for KEY_FRAME in multithreaded decoder" 2011-06-15 10:18:06 -07:00
John Koleszar
61599fb59f Use SSE as BPRED distortion metric consistently
The BPRED mode selection uses SSE as a distortion metric, but the early
breakout threshold being used was a variance value.

Change-Id: I42d4602fb9b548bf681a36445701fada5e73aff1
2011-06-15 10:53:37 -04:00
John Koleszar
1ade44b352 Merge "fix --disable-runtime-cpu-detect on x86" 2011-06-15 07:09:09 -07:00
Ronald S. Bultje
299193dd1c Disable specialcase for last frames if the sequence contains ARFs.
firstpass.c contains some rate adjustment code that assures that the
last few frames in a sequence abide by rate limits. If the second-to-
last group of frames contains an alt-ref frame (ARF), the last golden
frame (GF) is zero bytes, and we will thus spend a ridiculously high
number of bits on regular P-frames trying to hit the target rate. This
does slightly enhance the quality of these last few frames, but has
no perceptual value (other than hitting the target rate).

Disabling this code means we consistently (slightly) undershoot the
target rate and consequently do worse on the last few frames of a
clip, which is particularly noticeable for small clips. The quality-
per-bitrate is generally better, ~0.2% better overall on derf-set,
especially on clips such as garden, tennis, foreman at low bitrates.
Has a negative effect on hallmonitor at high bitrates.

Change-Id: I1d63452fef5fee4a0ad2fb2e9af4c9f2e0d86d23
2011-06-15 09:47:00 -04:00
Attila Nagy
c7e6aabbca Remove redundant check for KEY_FRAME in multithreaded decoder
For Intra blocks is enough to check ref_frame == INTRA_FRAME.

Change-Id: I3e2d3064c7642658a9e14011a4627de58878e366
2011-06-15 09:01:27 +03:00
Scott LaVarnway
7be5b6dae4 Merge "Populate bmi for B_PRED only" 2011-06-14 12:04:50 -07:00
Johann
92b0e544f3 fix --disable-runtime-cpu-detect on x86
Change-Id: Ib8e429152c9a8b6032be22b5faac802aa8224caa
2011-06-14 11:31:50 -04:00
Tero Rintaluoma
9909047461 Fix RT only build
Moved encode_intra function from firstpass.c to encodeintra.c to
prevent linking problem in real-time only build. Also changed name
of the function to vp8_encode_intra because it is not a static.

Change-Id: Ibf3c6c1de3152567347e5fbef47d1d39564620a5
2011-06-14 13:39:06 +03:00
James Zern
532c30c83e fix corrupt frame leak
If setup_token_decoder reported an internal error the memory allocated
there would not be freed in the resulting call to _remove_decompressor.

Change-Id: Ib459de222d76b1910d6f449cdcd01663447dbdf6
2011-06-13 17:32:19 -07:00
Scott LaVarnway
223d1b54cf Populate bmi for B_PRED only
Small decode performance gain (~1%) on keyframes.  No
noticeable gains on encode.  Also changed pick_intra4x4mby_modes()
to read the above and left block modes for keyframes only.

Change-Id: I1f4885252f5b3e9caf04d4e01e643960f910aba5
2011-06-13 17:14:11 -04:00
Scott LaVarnway
e71a010646 Calc ref_frame_cost once per frame
instead of every macro block.

Change-Id: I2604e94c6b89e3a8457777e21c8c38406d55b165
2011-06-13 09:58:03 -04:00
John Koleszar
f3ba4c6b82 Merge "bug fix mode_info_context not initialized for error-resilient" 2011-06-09 13:39:47 -07:00
Yaowu Xu
361717d2be remove one set of 16x16 variance funcations
call to this set of functions are replaced by var16x16.

Change-Id: I5ff1effc6c1358ea06cda1517b88ec28ef551b0d
2011-06-09 11:23:05 -07:00
James Berry
45feea4cf0 bug fix mode_info_context not initialized for error-resilient
uninitialized xd->mode_info_context would crash
vpxenc for --error-resilient=1.

Change-Id: I31849e40281e3d65ab63257cfec5e93398997f0b
2011-06-09 12:46:31 -04:00
John Koleszar
af49c11250 Update keyframe activity in non-RD mode
Activity update is no longer dependent on being in RD mode, so update
it unconditionally.

Change-Id: Ib617a6fc210dfc045455e3e4467d7ee5e3d1fa0e
2011-06-09 12:05:31 -04:00
Johann
79327be6c7 use GCC inline magic
Better fix for #326. ICC happens to support the inline magic

Change-Id: Ic367eea608c88d89475cb7b05d73500d2a1bc42b
2011-06-08 16:19:37 -04:00
John Koleszar
8767ac3bc7 Merge "vp8_pick_inter_mode: remove best_bmodes" 2011-06-08 10:59:30 -07:00
John Koleszar
9e4df2bcf5 Merge "vp8_pick_intra_mode: correct returned rate" 2011-06-08 10:58:36 -07:00
John Koleszar
254a7483e5 Merge "Move RD intra block mode selection to rdopt.c" 2011-06-08 10:51:50 -07:00
John Koleszar
001bd51ceb vp8_pick_inter_mode: remove best_bmodes
Since BPRED will be tested at most once, and SPLITMV is not enabled,
there's nothing to clobber the subblock modes, so there's no need to
save and restore them.

Change-Id: I7c3615b69190c10bd068a44df5488d6e8b85a364
2011-06-08 13:50:50 -04:00
Scott LaVarnway
dce64343d6 Merge "Removed unused function parameters" 2011-06-08 10:20:28 -07:00
John Koleszar
91907e0bf4 vp8_pick_intra_mode: correct returned rate
The returned rate was always the 4x4 rate, instead of the rate
matching the selected mode.

Change-Id: I51da31f80884f5e37f3bcc77d1047d31e612ded4
2011-06-08 13:19:12 -04:00
Scott LaVarnway
69d8d386ed Removed unused function parameters
Change-Id: Ib641c624faec28ad9eb99e2b5de51ae74bbcb2a2
2011-06-08 13:01:09 -04:00
Yaowu Xu
1fba1e38ea Adjust errorperbit according to RDMULT in activity masking
In activity masking, RDO constant RDMULT is adjusted on a per MB basis
adaptive to activity with the MB. errorperbit, which is defined as
RDMULT/RDDIV, is a constant used in motion estimation. Previously, in
activity masking, errorperbit is not changed even when RDMULT is changed.
This commit changed to adjust errorperbit according to the change in
RDMULT.

Test in cif set showed a very small but consistent gain by all quality
metrics (average, overall psnr and ssim) when activity masking is on.

Change-Id: I07ded3e852919ab76757691939fe435328273823
2011-06-08 09:45:47 -07:00
Yaowu Xu
5fafa2d524 Merge "Further activity masking changes:" 2011-06-08 09:30:31 -07:00
John Koleszar
96a42aaa2d Move RD intra block mode selection to rdopt.c
This change is analogous to I0b67dae1f8a74902378da7bdf565e39ab832dda7,
which made the move for the non-RD path.

Change-Id: If63fc1b0cd1eb7f932e710f83ff24d91454f8ed1
2011-06-08 12:05:05 -04:00
John Koleszar
e90d17d240 Move intra block mode selection to pickinter.c
This commit moves the intra block mode selection from encodeframe.c
to pickinter.c (in the non-RD case). This allowed pick_intra_mbuv_mode
and pick_intra4x4mby_modes to be made static, and is a step towards
refactoring intra mode selection in the main pickinter loop. Gave a
small perf increase (~0.5%).

Change-Id: I0b67dae1f8a74902378da7bdf565e39ab832dda7
2011-06-08 11:44:57 -04:00
Paul Wilkins
4e81a68af7 Further activity masking changes:
Some further re-structuring of activity masking code.
Still has various experimental switches.
Supports a metric based on intra encode.
Experimental comparison against a fixed activity target  rather
than a frame average, for altering rd and zbin.

Overall the SSIM performance is similar  to TT's original
code but there is a much smaller PSNR hit of circa
0.5% instead of 3.2%

Change-Id: I0fd53b2dfb60620b3f74d7415e0b81c1ac58c39a
2011-06-08 16:03:37 +01:00
Yaowu Xu
7368dd4f8f Merge "remove redundant functions" 2011-06-07 16:36:37 -07:00
Yaowu Xu
59129afc05 Merge "adjust sad per bit constants" 2011-06-07 12:37:04 -07:00
Yaowu Xu
221e00eaa9 adjust sad per bit constants
While investigating the effect of DC values on SAD and SSE in motion
estimation, a side finding indicates the two table of constants need
be adjusted. The adjustment was done by multiplying old constants by
90% with rounding. Also absorb the 1/2 scaling constant into the two
tables. Refer to change Ifa285c3e for background of the 1/2 factor.

Cif set test showed a very small gain on all metric.

Change-Id: I04333527a823371175dd46cb04a817e5b9a8b752
2011-06-07 12:35:03 -07:00
John Koleszar
5c166470a5 Merge "Reduce overshoot in 1 pass rate control" 2011-06-07 12:30:37 -07:00
Scott LaVarnway
346358a5b7 Merge "Wrapped asserts in critical code with CONFIG_DEBUG" 2011-06-07 06:53:51 -07:00
Scott LaVarnway
afb84bb1cc Merge "Removed unused function vp8_treed_read_num" 2011-06-07 06:51:24 -07:00
Scott LaVarnway
0e3bcc6f32 Wrapped asserts in critical code with CONFIG_DEBUG
Change-Id: I5b0aaca06f2e0f40588cb24fb0642b6865da8970
2011-06-07 09:34:47 -04:00
Scott LaVarnway
1374a4db3b Removed unused function vp8_treed_read_num
Change-Id: Id66e70540ee7345876f099139887c1843093907f
2011-06-07 09:32:51 -04:00
Yaowu Xu
d4700731ca remove redundant functions
The encoder defined about 4 set of similar functions to calculate sum,
variance or sse or a combination of them. This commit removed one set
of these functions, get8x8var and get16x16var, where calls to the later
function are replaced with var16x16 by using the fact on a 16x16 MB:
    variance == sse - sum*sum/256

Change-Id: I803eabd1fb3ab177780a40338cbd596dffaed267
2011-06-06 16:44:05 -07:00
Yunqing Wang
03973017a7 Remove hex search's variance calculation while in real-time mode
In real-time mode motion search, there is no need to calculate
variance. This change improved encoding speed by 1% ~ 2%(speed=-5).

Change-Id: I65b874901eb599ac38fe8cf9cad898c14138d431
2011-06-06 19:11:05 -04:00
Johann
04edde2b11 Merge "neon fast quantize block pair" 2011-06-06 13:42:58 -07:00
Johann
da8eb716e8 Merge "adds preload for armv6 encoder asm" 2011-06-06 13:32:13 -07:00
Scott LaVarnway
d1c0ba8f7a Merge "Removed unnecessary bmi motion vector stores." 2011-06-06 07:57:39 -07:00
John Koleszar
824e9410c6 Merge "Don't allow very short GF groups even when the GF is predicted from an ARF." 2011-06-06 07:02:29 -07:00
John Koleszar
212f618373 Reduce overshoot in 1 pass rate control
This patch attempts to reduce the peak bitrate hit by the encoder
when using small buffer windows.

Tested on the CIF set over 200-500kbps using these settings:

  --buf-sz=500 --buf-initial-sz=250 --buf-optimal-sz=250 \
  --undershoot-pct=100

Two pass encodes were tested at best quality. One pass encodes were
tested only at realtime speed 4:

  --rt --cpu-used=-4

The peak datarate (over the specified 500ms window) was measured
for each encode, and averaged together to get metric for
"average peak," computed as SUM(peak)/SUM(target). This patch
reduces the average peak datarate as follows:

  One pass:
    baseline:   1.29715
    this patch: 1.23664

  Two pass:
    baseline:   1.32702
    this patch: 1.37824

This change had a positive effect on our quality metrics as well:

  One pass CBR:
                    Min  / Mean / Max (pct)
    Average PSNR    -0.42 / 2.86 / 27.32
    Overall PSNR    -0.90 / 2.00 / 17.27
    SSIM            -0.05 / 3.95 / 37.46

  Two pass CBR:
                    Min  / Mean / Max (pct)
    Average PSNR    -4.47 / 4.35 / 35.99
    Overall PSNR    -3.40 / 4.18 / 36.46
    SSIM            -4.56 / 6.98 / 53.67

  One pass VBR:
                    Min  / Mean / Max (pct)
    Average PSNR    -5.21 /  0.01 / 3.30
    Overall PSNR    -8.10 / -0.38 / 1.21
    SSIM            -7.38 / -0.11 / 3.17
    (note: most values here were close to the mean, there were a few
     outliers on files that were very sensitive to golden frame size)

  Two pass VBR:
                    Min  / Mean / Max (pct)
    Average PSNR    0.00 / 0.00 / 0.00
    Overall PSNR    0.00 / 0.00 / 0.00
    SSIM            0.00 / 0.00 / 0.00

Neither one pass or two pass CBR mode adheres particularly strictly
to the short term buffer constraints, and two pass is less
consistent, even in the baseline commit. This should be addressed
in a later commit. This likely will hurt the quality numbers, as it
will have to reduce the burstiness of golden frames.

Aside: My work on this commit makes it clear that we need to make
rate control modes "pluggable", where you can easily write a new
one or work on one in isolation.

Change-Id: I1ea9a48f2beedd59891f1288aabf7064956b4716
2011-06-03 16:38:11 -04:00
Scott LaVarnway
f1d6cc79e4 Removed unnecessary bmi motion vector stores.
left_block_mv and above_block_mv will return the MB
motion vector for non SPLITMV macro blocks.

Change-Id: I58dbd7833b4fdcd44b6b72e98ec732c93c2ce4f4
2011-06-03 13:09:46 -04:00
Scott LaVarnway
8c5b73de2a Merge "Removed B_MODE_INFO" 2011-06-03 08:32:30 -07:00
Yunqing Wang
e5c236c210 Adjust bounds checking for hex search in real-time mode
Currently, hex search couldn't guarantee the motion vector(MV)
found is within the limit of maximum MV. Therefore, very large
motion vectors resulted from big motion in the video could cause
encoding artifacts. This change adjusted hex search bounds
checking to make sure the resulted motion vector won't go out
of the range. James Berry, thank you for finding the bug.

Change-Id: If2c55edd9019e72444ad9b4b8688969eef610c55
2011-06-03 08:53:42 -04:00
Scott LaVarnway
773768ae27 Removed B_MODE_INFO
Declared the bmi in BLOCKD as a union instead of B_MODE_INFO.
Then removed B_MODE_INFO completely.

Change-Id: Ieb7469899e265892c66f7aeac87b7f2bf38e7a67
2011-06-02 13:46:41 -04:00
Ronald S. Bultje
9f002bee53 Don't allow very short GF groups even when the GF is predicted from an ARF.
This is basically a slightly modified version of the previous patch,
and it has a moderately positive effect (SSIM/PSNR both +0.08% avg
on derf-set). Most clips show no change, except waterfall/coastguard,
each ~ +0.8% SSIM/PSNR. You can see similar effects in other clips
by shortening their length to terminate at a very short last group
of frames.

Change-Id: I7a70de99ca1f9fe6a8b6ca7a6e30e8a4b64383e4
2011-06-02 09:14:51 -07:00
Yaowu Xu
4ce6928d5b Merge "further clean up of errorperbit and sadperbit" 2011-06-02 08:58:03 -07:00
Yaowu Xu
5b2fb32961 further clean up of errorperbit and sadperbit
this commit makes the usage errorperbit and sadperbit consistent for
encoding modes and passes. Removed all different magic weight factors
associated with errorperbit. Now 1/2 is used for both sadperbit16 and
sadperbit4, the /2 operation is merged into initializations of the 2
variables.

Tests on cif set show .23%, 0.18% and 0.19% gain by avg psnr, overall
psnr and ssim respectively.

Change-Id: Ifa285c3e065ce0a5a77addfc9f95aabf54ee270d
2011-06-01 14:44:06 -07:00
John Koleszar
4101b5c5ed Merge "Bugfix in vp8dx_set_reference" 2011-06-01 13:57:23 -07:00
Henrik Lundin
69ba6bd142 Bugfix in vp8dx_set_reference
The fb_idx_ref_cnt book-keeping was in error. Added an assert to
prevent future errors in the reference count vector. Also fixed a
pointer syntax error.

Change-Id: I563081090c78702d82199e407df4ecc93da6f349
2011-06-01 21:41:12 +02:00
John Koleszar
5610970fe9 Merge "Fix code under #if CONFIG_INTERNAL_STATS." 2011-06-01 11:14:17 -07:00
Ronald S. Bultje
34ba18760f Fix code under #if CONFIG_INTERNAL_STATS.
Change-Id: Iccbd78d91c3071b16fb3b2911523a22092652ecd
2011-06-01 11:10:13 -07:00
Yaowu Xu
50916c6a7d remove some magic weights associated with sad_per_bit
sad_per_bit has been used for a number of motion vector search routines
with different magic weights: 1, 1/2 and 1/4. This commit remove these
magic numbers and use 1/2 for all motion search routines, also reformat
a number of source code lines to within 80 column limit.

Test on cif set shows overall effect is neutral on all metrics. <=0.01%

Change-Id: I8a382821fa4cffc9c0acf8e8431435a03df74885
2011-06-01 10:10:44 -07:00
Tero Rintaluoma
61f0c090df neon fast quantize block pair
vp8_fast_quantize_b_pair_neon function added to quantize
two adjacent blocks at the same time to improve performance.
 - Additional 3-6% speedup compared to neon optimized fast
   quantizer (Tanya VGA@30fps, 1Mbps stream, cpu-used=-5..-16)

Change-Id: I3fcbf141e5d05e9118c38ca37310458afbabaa4e
2011-06-01 10:48:05 +03:00
Scott LaVarnway
9e4f76c154 Merge "vp8_pick_inter_mode code cleanup" 2011-05-31 12:31:46 -07:00
Scott LaVarnway
1a5a1903ea vp8_pick_inter_mode code cleanup
Small code cleanups before attempting to reduce the size
of bmi found in BLOCKD.

Change-Id: Ie9c14adb53afd847716a75bcce067d0e6c04f225
2011-05-31 14:24:42 -04:00
John Koleszar
0a72f568ec Initialize first_time_stamp_ever
Misplaced #endif caused first_time_stamp_ever to only be initialized if
CONFIG_INTERNAL_STATS was set.

Change-Id: I2296a4ab00f7dfb767583edcc5d59b94f48c0621
2011-05-31 12:37:45 -04:00
Tero Rintaluoma
5305e79eae adds preload for armv6 encoder asm
Added preload instructions to armv6 encoder optimizations.
About 5% average speed-up on Tegra2 for VGA@30fps sequence.

Change-Id: I41d74737720fb71ce7a316f07555357822f3347e
2011-05-30 11:10:03 +03:00
John Koleszar
4a4ade6dc8 Merge "bug fix check frame buffer index before copy" 2011-05-27 12:35:06 -07:00
James Berry
8795b52512 bug fix check frame buffer index before copy
in onyx_if.c update_reference_frames() make
sure that frame buffer indexes are not equal
before preforming a buffer copy.  If two frames
share the same buffer the flags will already be
set correctly.

Change-Id: Ida9b5516d08e3435c90f131d2dc19d842cfb536e
2011-05-27 14:59:29 -04:00
Yunqing Wang
4fb5ce6a92 Merge "Use hex search for realtime mode speed>4" 2011-05-27 11:12:50 -07:00
Yunqing Wang
4d052bdd91 Use hex search for realtime mode speed>4
Test showed using hex search in realtime mode largely speed up
encoding process, and still achieves similar quality like the
diamond search we have. Therefore, removed the diamond search
option.

Change-Id: I975767d0ec0539f9f6ed7fdfc09506e39761b66c
2011-05-27 14:05:02 -04:00
Scott LaVarnway
ba420f1097 Merge "Broken EC after MODE_INFO size reduction" 2011-05-27 07:52:04 -07:00
Yunqing Wang
5a8cbb8955 Merge "Remove unused code" 2011-05-27 07:25:25 -07:00
Yunqing Wang
2dc24635ec Remove unused code
Hex search is not called in rdopt.c

Change-Id: I67347f03e13684147a7c77fb9e9147e440bb5e8e
2011-05-27 10:20:49 -04:00
Scott LaVarnway
4f586f7bd0 Broken EC after MODE_INFO size reduction
This patch fixes the compiler errors and the seg fault
when running decode_with_partial_drops.

Change-Id: I7c75369e2fef81d53b790d5dabc327218216838b
2011-05-26 15:13:00 -04:00
John Koleszar
1fe5070b76 Merge "Do not copy data between encoder reference buffers." 2011-05-26 09:58:26 -07:00
Yaowu Xu
9a248f1593 Merge "fix the mix use of errorperbit and sadperbit" 2011-05-26 09:39:41 -07:00
Scott LaVarnway
40b850b458 Merge "Use int_mv instead of MV in vp8_mv_cont" 2011-05-26 07:01:38 -07:00
Yaowu Xu
d8c525b8b1 fix the mix use of errorperbit and sadperbit
error_per_bit and sad_per_bit were designed as estimates of a bit worth
of sum squared error and sum absolute difference respectively. Under
this assumption, error_per_bit should be used in combination with 2nd
order errors (variance or sum squared error) while sad_per_bit should
be used in combination with 1st order SADs in motion estimation. There
were a few places where sad_per_bit has been misused with variances,
this commit changes to use error_per_bit for those places, also changes
parameter names to properly indicate which constant is being used.

On cif set, the change has a universal gain by all metrics: 0.13% by
average/overall psnr and 0.1% by ssim.

Change-Id: I4850fdcc3fd6886b30f784bd843f13dd401215fb
2011-05-25 16:48:10 -07:00
Yunqing Wang
13b56eeb7a Merge " Use var8x8 instead of get8x8var in VP8_UVSSE" 2011-05-25 11:35:42 -07:00
Yunqing Wang
f299d628f3 Merge "Return sse value in vp8_variance SSE2 functions" 2011-05-25 11:31:07 -07:00
Yaowu Xu
22c05c0575 remove code not in use
Change-Id: I6e5e86235d341cce3b02abda26dbeb71940ed955
2011-05-25 09:46:37 -07:00
Yunqing Wang
b6679879b8 Return sse value in vp8_variance SSE2 functions
Minor modification.

Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 11:55:41 -04:00
Attila Nagy
a615c40499 Use var8x8 instead of get8x8var in VP8_UVSSE
'sum' returned by get8x8var is not used and var8x8 has optimizations
  for more platforms.

Change-Id: I4a907fb1a05f285669fb0b95dc71d42182c980f6
2011-05-25 12:54:34 +03:00
Yunqing Wang
d75eb73653 Fix a bug happening while encoding at profile=3
While profile=3, there is no sub-pixel search. Distortion and SSE
have to calculated using get_inter_mbpred_error().

Change-Id: Ifb36e17eef7750af93efa7d0e2870142ef540184
2011-05-24 16:28:23 -04:00
Scott LaVarnway
a39321f37e Use int_mv instead of MV in vp8_mv_cont
Less operations.

Change-Id: Ibb9cd5ae66b8c7c681c9a654d551c8729c31c3ae
2011-05-24 16:01:12 -04:00
Scott LaVarnway
cfab2caee1 Removed unused variable warnings
Change-Id: I6e5e921f03dc15a72da89a457848d519647677a3
2011-05-24 15:17:03 -04:00
Scott LaVarnway
b5278f38b0 Merge "MODE_INFO size reduction" 2011-05-24 12:08:24 -07:00
Scott LaVarnway
e11f21af9a MODE_INFO size reduction
Declared the bmi in MODE_INFO as a union instead of B_MODE_INFO.
This reduced the memory footprint by 518,400 bytes for 1080
resolutions.  The decoder performance improved by ~4% for the
clip used and the encoder showed very small improvements. (0.5%)
This reduction was first mentioned to me by John K. and in a
later discussion by Yaowu.
This is WIP.

Change-Id: I8e175fdbc46d28c35277302a04bee4540efc8d29
2011-05-24 13:24:52 -04:00
John Koleszar
fbea372817 Merge "Fixing bug in VP8_SET_REFERENCE decoder control command" 2011-05-24 05:57:44 -07:00
Yunqing Wang
69aad3a720 Merge "Rewrite hex search function" 2011-05-24 05:26:16 -07:00
Henrik Lundin
a126cd1760 Fixing bug in VP8_SET_REFERENCE decoder control command
In vp8dx_set_reference, the new reference image is written to an
unused reference frame buffer.

Change-Id: I9e4f2cef5a011094bb7ce7b2719cbfe096a773e8
2011-05-24 09:03:43 +02:00
Yaowu Xu
99fb568e67 Merge "use get8x8var directly for non-subpixel motion case in VP8_UVSSE" 2011-05-23 14:49:56 -07:00
Yunqing Wang
7838f4cfff Rewrite hex search function
Reduced some bound checks in hex search function.

Change-Id: Ie5f73a6c227590341c960a74dc508cff80f8aa06
2011-05-23 16:18:52 -04:00
Yaowu Xu
ab2dfd22f3 use get8x8var directly for non-subpixel motion case in VP8_UVSSE
VP8_UVSSE mistakenly used subpixvar8x8 to calculate SSE for non-subpixl
motion cases.

Change-Id: I4a5398bb9ef39c211039f6af4540546d4972e6a9
2011-05-23 09:11:28 -07:00
John Koleszar
ad6fe4a88c Merge "bug fix active_worst_quality set below active_best_quality" 2011-05-20 11:23:10 -07:00
John Koleszar
8196cc85f8 Merge "cleanup: collect twopass variables" 2011-05-20 11:20:44 -07:00
Johann
6d82d2d22e Merge "Fixed iwalsh_neon build problems with RVDS4.1" 2011-05-20 07:51:11 -07:00
Yaowu Xu
1fbc81a970 Merge "revise two function definitions with less parameters" 2011-05-20 07:45:42 -07:00
John Koleszar
a0c11928db Merge "Remove unused members of VP8_COMP" 2011-05-20 07:39:03 -07:00
Yaowu Xu
a4c69e9a0f revise two function definitions with less parameters
Change-Id: Ia96e5bf915e4d3c0ac9c1795114bd9e5dd07327a
2011-05-19 19:06:03 -07:00
Yaowu Xu
1f3f18443d Merge "disable trellis optimization for first pass" 2011-05-19 17:25:31 -07:00
Yaowu Xu
d5b8f7860f disable trellis optimization for first pass
also remove 2 #defines and 1 function declaration that are not in use.

Change-Id: I8f743d0e3dd9ebf1de24a8b0c30ff09f29b00c53
2011-05-19 17:22:14 -07:00
James Berry
caa1b28be3 bug fix active_worst_quality set below active_best_quality
fixed a bug where active_worst_quality could be set
below active_best_quality which could result in an
infinite loop.

Change-Id: I93c229c3bc5bff2a82b4c33f41f8acf4dd194039
2011-05-19 18:10:31 -04:00
John Koleszar
63cb1a7ce0 cleanup: collect twopass variables
This patch collects the twopass specific memebers of VP8_COMP into a
dedicated struct. This is a first step towards isolating the two pass
rate control and aids readability by decorating these variables with
the 'twopass.' namespace. This makes it clear to the reader in what
contexts the variable will be valid, and is a hint that a section of
code might be a good candidate to move to firstpass.c in later
refactoring. There likely will be other rate control modes that need
their own specific data as well.

This notation is probably overly verbose in firstpass.c, so an
alternative would be to access this struct through a pointer like
'rc->' instead of 'cpi->firstpass.' in that file. Feel free to make
a review comment to that effect if you prefer.

Change-Id: I0ab8254647cb4b493a77c16b5d236d0d4a94ca4d
2011-05-19 17:26:09 -04:00
Scott LaVarnway
dba79821f0 Merge "Using partition_info instead of blockd info for splitmv" 2011-05-19 13:22:59 -07:00
John Koleszar
048497720c Remove unused members of VP8_COMP
Various members that were either completely unreferenced or written
and not read.

Change-Id: Ie41ebac0ff0364a76f287586e4fe09a68907806e
2011-05-19 15:49:09 -04:00
Scott LaVarnway
99b9757685 Using partition_info instead of blockd info for splitmv
The partition_info struct contains info just for SPLITMV,
so it should be used instead of BLOCKD.  Eventually, I want
to reduce the size of B_MODE_INFO struct found in BLOCKD, so
this is the first step toward that goal.
Also, since SPLITMV is not supported in vp8_pick_inter_mode(),
the unnecessary mem copies and checks were removed.  For rt
encodes, this gave a slight performance improvement.

Change-Id: I5585c98fa9d5acbde1c7e0f452a01d9ecc080574
2011-05-19 15:03:36 -04:00
Scott LaVarnway
914f7c36d7 Merge "Make hor UV predict ~2x faster (73 vs 132 cycles) using SSSE3." 2011-05-19 11:22:01 -07:00
John Koleszar
c684d5e5f2 Merge "changed configure option name to reduce confusion" 2011-05-19 11:17:08 -07:00
John Koleszar
ff39958cee Merge "Make activity masking functions static" 2011-05-19 11:12:18 -07:00
John Koleszar
21ca4c4d5d Merge "Fix segv without --enable-error-concealment" 2011-05-19 10:58:24 -07:00
John Koleszar
7def902261 Fix segv without --enable-error-concealment
Missed wrapping one function call in #if CONFIG_ERROR_CONCEALMENT.

Change-Id: I5746b1e6e4531670dbed1130467331fe309bdcae
2011-05-19 13:57:45 -04:00
John Koleszar
e3081b2502 Merge "Adding error-concealment to the decoder." 2011-05-19 10:48:58 -07:00
Stefan Holmer
d04f852368 Adding error-concealment to the decoder.
The error-concealer is plugged in after any motion vectors have been
decoded. It tries to estimate any missing motion vectors from the
motion vectors of the previous frame. Intra blocks with missing
residual are replaced with inter blocks with estimated motion vectors.

This feature was developed in a separate sandbox
(sandbox/holmer/error-concealment).

Change-Id: I5c8917b031078d79dbafd90f6006680e84a23412
2011-05-19 13:46:33 -04:00
John Koleszar
a84177b432 Make activity masking functions static
These don't need extern linkage.

Change-Id: I21220ada926380a75ff654f24df84376ccc49323
2011-05-19 11:14:13 -04:00
John Koleszar
87254e0b7b Move quantizer init functions to quantize.c
Group related functions together.

Change-Id: I92fd779225b75a7204650f1decb713142c655d71
2011-05-19 11:07:41 -04:00
Attila Nagy
f96d56c4aa Fixed iwalsh_neon build problems with RVDS4.1
rvct 4.1 was complaining about vstmia.16, store multiple expects 64 data type.
optimized the implementation.

Change-Id: I0701052cabd685c375637bbc3796ff6d88f5972c
2011-05-19 10:27:26 +03:00
Yunqing Wang
00a1e2f8e4 Merge "Modify MVcount in pick_inter_mode to eliminate calling of vp8_find_near_mvs" 2011-05-18 12:53:27 -07:00
Yunqing Wang
9c62f94129 Fix a bug in vp8_clamp_mv function
Scott fixed the bug in MV clamping function in encoder, which
could cause artifacts.

Change-Id: Id05f2794c43c31cdd45e66179c8811f3ee452cb9
2011-05-18 09:52:56 -04:00
Yunqing Wang
f62b33f140 Modify MVcount in pick_inter_mode to eliminate calling of vp8_find_near_mvs
Moved MVcount modification in pick_inter_mode, and eliminated
calling of vp8_find_near_mvs.

Change-Id: Icd47448a1dfc8fdf526f86757d0e5a7f218cb5e8
2011-05-17 10:59:42 -04:00
John Koleszar
eafdc5e10a Merge "Improve framerate adaptation" 2011-05-13 11:18:42 -07:00
Yaowu Xu
5608c14020 Merge "adjusting rd constant slightly by ~10%" 2011-05-13 09:28:26 -07:00
Paul Wilkins
0e86235265 Merge "Restructure of activity masking code." 2011-05-13 09:23:50 -07:00
Paul Wilkins
ff52bf3691 Restructure of activity masking code.
This commit restructures the mb activity masking code
to better facilitate experimentation using different metrics
etc. and also allows for adjustment of the zero bin either
for encode only or both the encode and mode selection
stages

It also uses information from the current frame rather than
the previous frame and the default strength has been
reduced.

Change-Id: Id39b19eace37574dc429f25aae810c203709629b
2011-05-13 10:37:50 +01:00
John Koleszar
5ed116e220 Improve framerate adaptation
This patch improves the accuracy of frame rate estimation by using a
larger, 1 second window. It also more quickly adapts to step changes
in the input frame rate (ie 30fps to 15fps)

Change-Id: I39e48a8f5ac880b4c4b2ebd81049259b81a0218e
2011-05-12 15:07:50 -04:00
Scott LaVarnway
71a7501bcf Removed mv_bits_sadcost
This sad cost is being generated but never used.

Change-Id: I562eebdcb792b743770954feca365b5b37491ecd
2011-05-12 11:20:41 -04:00
Scott LaVarnway
6b25501bf1 Using int_mv instead of MV
The compiler produces better assembly when using int_mv
for assignments.  The compiler shifts and ors the two 16bit
values when assigning MV.

Change-Id: I52ce4bc2bfbfaf3f1151204b2f21e1e0654f960f
2011-05-12 11:08:16 -04:00
Yunqing Wang
6ed81fa5b3 Merge "Modification and issue fix in full-pixel refining search" 2011-05-12 07:20:44 -07:00
Yunqing Wang
b4da1f83e6 Modification and issue fix in full-pixel refining search
Further modification and wrong implementation fix which caused
refining_search and refining_searchx4 result mismatching.

Change-Id: I80cb3a44bf5824413fd50c972e383eebb75f9b6f
2011-05-12 10:18:40 -04:00
Yaowu Xu
bd9d890605 adjusting rd constant slightly by ~10%
This is to reflect the RD improvement in the encoder. The change has a
small positive impact on quality (0.25% by VPXSSIM and 0.05% by PSNR)

Change-Id: Ic66ffc19b10870645088c0624c85556f009fd210
2011-05-11 23:32:06 -07:00
Yaowu Xu
ba6f60dba7 Merge "remove a variable no longer in use" 2011-05-10 20:20:59 -07:00
Yaowu Xu
1bcf4e66bb Merge "fix a bug related to gf_active_flags in multi-threaded encoder" 2011-05-10 19:59:52 -07:00
Yaowu Xu
f7cf439b34 remove a variable no longer in use
The variable is introduced in commit 2e53e9e53 to make more use of
trellis quantization, but this is no longer necessary after RDMULT
was made adaptive in a number of later commits.

Change-Id: I7420522ec7723f38cf77033466c25afb405d52ae
2011-05-10 19:57:51 -07:00
Johann
df2023a6cb set up Global Offset Table in recon
global values were being referenced, but the GOT was not being set up.
as the GOT is only required for PIC, this issue wasn't caught in the
default configuration.

Change-Id: I8006e53776139362a76f2c80cf9d0f8458602b2f
http://code.google.com/p/webm/issues/detail?id=328
2011-05-10 15:58:56 -04:00
Yunqing Wang
c7a56f677d Merge "Use diamond search to replace full search in full-pixel refining search" 2011-05-10 06:59:38 -07:00
Yunqing Wang
cb7b1fb144 Use diamond search to replace full search in full-pixel refining search
In NEWMV mode, currently, full search is used as the refining search
after n-step search. By replacing it with an iterative diamond search
of radius 1 largely reduced the computation complexity, but still
maintained the same encoding quality since the refining search is
done for every macroblock instead of only a small precentage of
macroblocks while using full search.

Tests on the test set showed a 3.4% encoding speed increase with none
psnr & ssim loss.

Change-Id: Ife907d7eb9544d15c34f17dc6e4cfd97cb743d41
2011-05-09 14:07:06 -04:00
Johann
a7d4d3c550 clean up unused variable warnings
Change-Id: I9467d7a50eac32d8e8f3a2f26db818e47c93c94b
2011-05-09 12:56:20 -04:00
Yaowu Xu
89c6017cc0 fix a bug related to gf_active_flags in multi-threaded encoder
Paul pointed out that the pointer to the gf_active_flags is not being
properly incremented in multithreaded encoder. This commit fixes the
issue by making sure the gf_active_ptr points to the starting of next
group of mb rows.

Change-Id: I3246e657d23beabb614dfb880733a68a5fd7e34c
2011-05-06 09:00:44 -07:00
John Koleszar
5c756005aa Merge "Don't override active_worst_quality in 2 pass" 2011-05-06 08:59:05 -07:00
Johann
52490354f3 Merge "neon fast quantizer updated" 2011-05-06 08:54:14 -07:00
John Koleszar
abc9958c52 Don't override active_worst_quality in 2 pass
Commit db5057c introduced a bug in that the active_worst_quality
selected by the 2 pass rate controller was being overridden for key
frames, causing a severe quality loss.

Change-Id: I4865a6fbe3e94e9b4fb9271c7dd68b455d7b371d
2011-05-06 11:48:53 -04:00
Tero Rintaluoma
33fa7c4ebe neon fast quantizer updated
vp8_fast_quantize_b_neon function updated and further optimized.
 - match current C implementation of fast quantizer
 - updated to use asm_enc_offsets for structure members
 - updated ads2gas scripts to handle alignment issues

Change-Id: I5cbad9c460ad8ddb35d2970a8684cc620711c56d
2011-05-06 08:59:52 +03:00
Aron Rosenberg
eeb8117303 Fix semaphore emulation on Windows
The existing emulation of posix semaphores on Windows uses SetEvent()
and WaitForSingleObject(), which implements a binary semaphore, not a
counting semaphore as implemented by posix. This causes deadlock when
used with the expected posix semantics. Instead, this patch uses the
CreateSemaphore() and ReleaseSemaphore() calls (introduced in Windows
2000) which have the expected behavior.

This patch also reverts commit eb16f00, which split a semaphore that
was being used with counting semantics into two binary semaphores.
That commit is unnecessary with corrected emulation.

Change-Id: If400771536a27af4b0c3a31aa4c4e9ced89ce6a0
2011-05-06 00:13:59 -04:00
Yunqing Wang
eb16f00cf2 Fix rare hang in multi-thread encoder on Windows
This patch is to fix a rare hang in multi-thread encoder that was
only seen on Windows. Thanks for John's help in debugging the
problem. More test is needed.

Change-Id: Idb11c6d344c2082362a032b34c5a602a1eea62fc
2011-05-05 10:42:29 -04:00
Johann
ca5c1b17a2 Merge "Loopfilter NEON: Use VMOV for constant vectors instead of VLD." 2011-05-05 06:16:21 -07:00
Yunqing Wang
aeb86d615c Merge "Runtime detection of available processor cores." 2011-05-05 04:59:54 -07:00
Attila Nagy
a6aa389d2f Loopfilter NEON: Use VMOV for constant vectors instead of VLD.
Change-Id: I562b6e01c32bb51d00f3b95faf757fc7dc29a3a3
2011-05-04 11:29:23 +03:00
Yunqing Wang
3fbade23a2 Merge "Modify HEX search" 2011-05-03 11:59:32 -07:00
Yunqing Wang
04ec930abc Modify HEX search
Changed 8-neighbor searching to 4-neighour searching, and continued
searching until the center point is the best match.

Test on test set showed 1.3% encoding speed improvement as well as
0.1% PSNR and SSIM improvement at speed=-5 (rt mode).

Will continue to improve it.

Change-Id: If4993b1907dd742b906fd3f86fee77cc5932ee9a
2011-05-03 14:26:33 -04:00
Yaowu Xu
e9465daee3 Merge "change to use fast ssim code for internal ssim calculations" 2011-05-03 11:20:52 -07:00
Yaowu Xu
6c565fada0 change to use fast ssim code for internal ssim calculations
The commit also removed the slow ssim calculation that uses a 7x7
kernel, and revised the comments to better describe how sample ssim
values are computed and averaged

Change-Id: I1d874073cddca00f3c997f4b9a9a3db0aa212276
2011-05-03 08:36:17 -07:00
John Koleszar
c09d8c1419 Merge "Fix documentation typos" 2011-05-02 06:50:22 -07:00
John Koleszar
a66d8d33dd Fix compile error with --enable-postproc-visualizer
Typo.

Change-Id: I9cc6a4587c3d93c9f0da5e101d376741fc9622a4
2011-05-02 09:28:37 -04:00
Thijs Vermeir
8942f70cdf Fix documentation typos
Change-Id: I97124670926433bf1593c91660d8b8f8482ea9ce
2011-04-30 09:34:59 +02:00
Ronald S. Bultje
5a23352c03 Make hor UV predict ~2x faster (73 vs 132 cycles) using SSSE3.
Change-Id: I658a1df7d825f820573cb2d11ad402f9d2791035
2011-04-29 11:52:09 -07:00
Yaowu Xu
57ad189129 changed configure option name to reduce confusion
Renamed configure option "enable-psnr" to "enable-internal-stats" to
better reflect the purpose of the option and eliminate the confusion
reported in http://code.google.com/p/webm/issues/detail?id=35

Change-Id: If72df6fdb9f1e33dab1329240ba4d8911d2f1f7a
2011-04-29 09:39:05 -07:00
Yunqing Wang
dfa9e2c5ea Merge "Use insertion sort instead of quick sort" 2011-04-29 08:27:58 -07:00
Scott LaVarnway
1b2abc5f49 Merge "Consolidated build inter predictors" 2011-04-29 07:13:49 -07:00
James Berry
f10732554b bug fix removed inline from recon_wrapper_sse2.c
removed inline from recon_wrapper_sse2.c to build
for visual stuido

Change-Id: I74a3482950448e2cdb30e9cd7087145b440d8a22
2011-04-28 15:12:00 -04:00
Scott LaVarnway
219ba87a93 Merge "Use psadbw to get the sum of bytes in a line." 2011-04-28 07:58:20 -07:00
Scott LaVarnway
ccd6f7ed77 Consolidated build inter predictors
Code cleanup.

Change-Id: Ic8b0167851116c64ddf08e8a3d302fb09ab61146
2011-04-28 10:53:59 -04:00
Ronald S. Bultje
1e7ded69cf Use psadbw to get the sum of bytes in a line.
Thanks Jason for pointing that out on #vp8. ;-).

Change-Id: I5330a753e752a8704b78a409597472628e0b26a5
2011-04-27 13:49:21 -07:00
Scott LaVarnway
2e102855f4 Removed unused code in reconinter
The skip flag is never set by the encoder for SPLITMV.

Change-Id: I5ae6457edb3a1193cb5b05a6d61772c13b1dc506
2011-04-27 15:25:32 -04:00
John Koleszar
085fb4b737 Merge "SSE2/SSSE3 optimizations for build_predictors_mbuv{,_s}()." 2011-04-27 12:02:55 -07:00
Ronald S. Bultje
1083fe4999 SSE2/SSSE3 optimizations for build_predictors_mbuv{,_s}().
decoding

before
10.425
10.432
10.423
=10.426

after:
10.405
10.416
10.398
=10.406, 0.2% faster

encoding

before
14.252
14.331
14.250
14.223
14.241
14.220
14.221
=14.248

after
14.095
14.090
14.085
14.095
14.064
14.081
14.089
=14.086, 1.1% faster

Change-Id: I483d3d8f0deda8ad434cea76e16028380722aee2
2011-04-27 11:31:27 -07:00
Yunqing Wang
5abafcc381 Use insertion sort instead of quick sort
Insertion sort performs better for sorting small arrays. In real-
time encoding (speed=-5), test on test set showed 1.7% performance
gain with 0% PSNR change in average.

Change-Id: Ie02eaa6fed662866a937299194c590d41b25bc3d
2011-04-27 13:53:28 -04:00
John Koleszar
64355ecad3 Merge "Speed up VP8DX_BOOL_DECODER_FILL" 2011-04-27 09:03:45 -07:00
John Koleszar
f8ffecb176 Merge "Update VP8DX_BOOL_DECODER_FILL to better detect EOS" 2011-04-27 09:03:24 -07:00
John Koleszar
5e1fd41357 Speed up VP8DX_BOOL_DECODER_FILL
The end-of-buffer check is hoisted out of the inner loop. Gives
about 0.5% improvement on x86_64.

Change-Id: I8e3ed08af7d33468c5c749af36c2dfa19677f971
2011-04-27 10:25:03 -04:00
John Koleszar
9594370e0c Update VP8DX_BOOL_DECODER_FILL to better detect EOS
Allow more reliable detection of truncated bitstreams by being more
precise with the count of "virtual" bits in the value buffer.
Specifically, the VP8_LOTS_OF_BITS value is accumulated into count,
rather than being assigned, which was losing the prior value,
increasing the required tolerance when testing for the error condition.

Change-Id: Ib5172eaa57323b939c439fff8a8ab5fa38da9b69
2011-04-27 10:24:39 -04:00
John Koleszar
db5057c742 Refactor calc_iframe_target_size
Combine calc_iframe_target_size, previously only used for forced
keyframes, with calc_auto_iframe_target_size, which handled most
keyframes.

Change-Id: I227051361cf46727caa5cd2b155752d2c9789364
2011-04-26 16:55:35 -04:00
John Koleszar
81d2206ff8 Move pick_frame_size() to ratectrl.c
This is a first step in cleaning up the redundancies between
vp8_calc_{auto_,}iframe_target_size. The pick_frame_size() function is
moved to ratectrl.c, and made to be the primary interface. This means
that the various calc_*_target_size functions can be made private.

Change-Id: I66a9a62a5f9c23c818015e03f92f3757bf3bb5c8
2011-04-26 16:49:54 -04:00
Scott LaVarnway
0da77a840b Merge "Test vector mismatch fix" 2011-04-26 10:12:37 -07:00
Scott LaVarnway
7a2b9c50a3 Test vector mismatch fix
Fixed test vector mismatch that was introduced
in the "Removed dc_diff from MB_MODE_INFO"
(Ie2b9cdf9e0f4e8b932bbd36e0878c05bffd28931)

Change-Id: I98fa509b418e757b5cdc4baa71202f4168dc14ec
2011-04-26 09:37:19 -04:00
Johann
d5c46bdfc0 Merge "remove simpler_lpf" 2011-04-25 14:51:07 -07:00
Johann
01527e743f remove simpler_lpf
the decision to run the regular or simple loopfilter is made outside the
function and managed with pointers

stop tracking the option in two places. use filter_type exclusively

Change-Id: I39d7b5d1352885efc632c0a94aaf56b72cc2fe15
2011-04-25 17:37:41 -04:00
John Koleszar
fd6da3b2e7 Fix duplicate vp8_compute_frame_size_bounds
Likely introduced by a bad automatic merge from gerrit.

Change-Id: I0c6dd6ec18809cf9492f524d283fa4a3a8f4088b
2011-04-25 14:30:57 -04:00
John Koleszar
1f32b1489c Merge "Remove unused functions" 2011-04-25 11:05:00 -07:00
John Koleszar
47bc1c7013 Remove unused functions
Remove estimate_min_frame_size() and calc_low_ss_err(), as they are
never referenced.

Change-Id: I3293363c14ef70b79c4678ca27aa65b345077726
2011-04-25 13:54:23 -04:00
John Koleszar
cfbfd39de8 Merge "Change rc undershoot/overshoot semantics" 2011-04-25 10:49:32 -07:00
John Koleszar
76557e34d2 Merge "Limit size of initial keyframe in one-pass." 2011-04-25 10:48:13 -07:00
John Koleszar
d9f898ab6d Merge "Add rc_max_intra_bitrate_pct control" 2011-04-25 10:47:57 -07:00
John Koleszar
454cbc96b7 Limit size of initial keyframe in one-pass.
Rather than using a default size of 1/2 or 3/2 seconds for the first
frame, use a fraction of the initial buffer level to give the
application some control.

This will likely undergo further refinement as size limits on key
frames are currently under discussion on codec-devel@, but this gives
much better behavior for small buffer sizes as a starting point.

Change-Id: Ieba55b86517b81e51e6f0a9fe27aabba295acab0
2011-04-25 13:47:20 -04:00
John Koleszar
aa926fbd27 Add rc_max_intra_bitrate_pct control
Adds a control to limit the maximum size of a keyframe, as a function of
the per-frame bitrate. See this thread[1] for more detailed discussion:

[1]: http://groups.google.com/a/webmproject.org/group/codec-devel/browse_thread/thread/271b944a5e47ca38

Change-Id: I7337707642eb8041d1e593efc2edfdf66db02a94
2011-04-25 13:47:14 -04:00
John Koleszar
2089b2cee5 Merge "bug fix possible keyframe context divide by zero" 2011-04-25 09:35:12 -07:00
James Berry
8d5ce819dd bug fix possible keyframe context divide by zero
vp8_adjust_key_frame_context() divides by
estimate_keyframe_frequency() which can
return 0 in the case where --kf-max-dist=0.

Change-Id: Idfc59653478a0073187cd2aa420e98a321103daa
2011-04-25 12:16:36 -04:00
Johann
aeca599087 Merge "keep values in registers during quantization" 2011-04-25 06:52:38 -07:00
Scott LaVarnway
c36b6d4d01 Merge "Removed unnecessary frame type checks" 2011-04-25 06:45:43 -07:00
Scott LaVarnway
5b67329747 Merge "Removed dc_diff from MB_MODE_INFO" 2011-04-25 06:45:32 -07:00
Ronald S. Bultje
496bcbb0de Fix overflow in temporal_filter_apply_sse2().
The accumulator array is an integer array, so use paddd instead of paddw
to add values to it. Fixes overflows when using large --arnr-maxframes
(>8) values.

Change-Id: Iad83794caa02400a65f3ab5760f2517e082d66ae
2011-04-22 10:00:38 -04:00
John Koleszar
73c3d32705 Merge "Remove unused kf rate variables" 2011-04-21 16:54:14 -07:00
Adrian Grange
d2a6eb4b1e Corrected format specifiers in debug print statements
The arguments to these fprintfs are int not long int so
the format specifier should be "%d" and not "%ld". This
was writing garbage in the linux build.

Change-Id: I3d2aa8a448d52e6dc08858d825bf394929b47cf3
2011-04-21 15:45:57 -07:00
Johann
508ae1b3d5 keep values in registers during quantization
add an sse4 quantizer so we can use pinsrw/pextrw and keep values in xmm
registers instead of proxying through the stack. and as long as we're
bumping up, use some ssse3 instructions in the EOB detection (see ssse3
fast quantizer)
pick up about a percent on 32bit and about two on 64bit.

Change-Id: If15abba0e8b037a1d231c0edf33501545c9d9363
2011-04-21 15:47:55 -04:00
Scott LaVarnway
6f6cd3abb9 Removed unnecessary frame type checks
ref_frame is set to INTRA_FRAME for keyframes.  The B_PRED
mode is only used in intra frames.

Change-Id: I9bac8bec7c736300d47994f3cb570329edf11ec0
2011-04-21 14:59:42 -04:00
Scott LaVarnway
3698c1f620 Removed dc_diff from MB_MODE_INFO
The dc_diff flag is used to skip loopfiltering.  Instead
of setting this flag in the decoder/encoder, we now check
for this condition in the loopfilter.

Change-Id: Ie2b9cdf9e0f4e8b932bbd36e0878c05bffd28931
2011-04-21 14:38:36 -04:00
Scott LaVarnway
7a49accd0b Removed force_no_skip
force_no_skip is always set to zero.

Change-Id: I89b61c5e0bee34627a9c07c05f3517e1db76af77
2011-04-20 15:45:12 -04:00
Scott LaVarnway
09c933ea80 Removed redundant checks of the mode_info_context flags
Code cleanup.  The build inter predictor functions are
redundantly checking the mode_info_context for either
INTRA_FRAME or SPLITMV.

Change-Id: I4d58c3a5192a4c2cec5c24ab1caf608bf13aebfb
2011-04-20 14:06:40 -04:00
Attila Nagy
43464e94ed Do not copy data between encoder reference buffers.
Golden and ALT reference buffers were refreshed by copying from
the new buffer. Replaced this by index manipulation.
Also moved all the reference frame updates to one function for
easier tracking.

Change-Id: Icd3e534e7e2c8c5567168d222e6a64a96aae24a1
2011-04-20 15:26:55 +03:00
John Koleszar
ad6a8ca58b Remove unused kf rate variables
Remove tot_key_frame_bits and prior_key_frame_size[] as they were
tracked but never used. Remove intra_frame_target, as it was only
used to initialize prior_key_frame_size.

Refactor vp8_adjust_key_frame_context() some to remove unnecessary
calculations.

Change-Id: Icbc2c83d2b90e184be03e6f9679e678f3a4bce8f
2011-04-19 16:14:57 -04:00
Johann
4a2b684ef4 modify SAVE_XMM for potential 64bit use
the win64 abi requires saving and restoring xmm6:xmm15. currently
SAVE_XMM and RESTORE XMM only allow for saving xmm6:xmm7. allow
specifying the highest register used and if the stack is unaligned.

Change-Id: Ica5699622ffe3346d3a486f48eef0206c51cf867
2011-04-19 10:42:45 -04:00
Johann
a9b465c5c9 Merge "Add save/restore xmm registers in x86 assembly code" 2011-04-19 06:32:10 -07:00
Johann
c7cfde42a9 Add save/restore xmm registers in x86 assembly code
Went through the code and fixed it. Verified on Windows.

Where possible, remove dependencies on xmm[67]

Current code relies on pushing rbp to the stack to get 16 byte
alignment. This broke when rbp wasn't pushed
(vp8/encoder/x86/sad_sse3.asm). Work around this by using unaligned
memory accesses. Revisit this and the offsets in
vp8/encoder/x86/sad_sse3.asm in another change to SAVE_XMM.

Change-Id: I5f940994d3ebfd977c3d68446cef20fd78b07877
2011-04-18 16:30:38 -04:00
Yunqing Wang
48438d6016 Merge "Use sub-pixel search's SSE in mode selection" 2011-04-18 13:20:04 -07:00