Fritz Koenig [Mon, 27 Jun 2011 16:43:21 +0000 (09:43 -0700)]
Fix after removal of B_MODE_INFO
Change Ieb746989: Removed B_MODE_INFO missed this.
Change-Id: I32202555581cc2a5d45e729c6650ada4d2df55d3
Johann [Mon, 27 Jun 2011 15:55:18 +0000 (08:55 -0700)]
Merge "configuration, support disabling any subset of ARM arch"
Stefan Holmer [Thu, 16 Jun 2011 09:44:50 +0000 (11:44 +0200)]
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
Adrian Grange [Fri, 24 Jun 2011 15:43:40 +0000 (08:43 -0700)]
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
John Koleszar [Thu, 23 Jun 2011 15:47:09 +0000 (11:47 -0400)]
Revert "Reduce overshoot in 1 pass rate control"
This reverts commit
212f6183739d448ad5fa2ccf1b4edd30829b2806.
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
John Koleszar [Thu, 23 Jun 2011 15:18:10 +0000 (08:18 -0700)]
Merge "Fix parallel install"
John Koleszar [Thu, 23 Jun 2011 15:09:10 +0000 (08:09 -0700)]
Merge changes I2807b5a1,I59b020c2
* changes:
vpxenc: add rate histogram display
vpxenc: add quantizer histogram display
John Koleszar [Tue, 21 Jun 2011 20:42:33 +0000 (16:42 -0400)]
vpxenc: add rate histogram display
Add the --rate-hist=n option, which displays a histogram with n
buckets for the rate over the --buf-sz window.
Change-Id: I2807b5a1525c7972e9ba40839b37e92b23ceffaf
John Koleszar [Mon, 20 Jun 2011 21:09:29 +0000 (17:09 -0400)]
vpxenc: add quantizer histogram display
Add the --q-hist=n option, which displays a histogram with n buckets for
the quantizer selected on each frame.
Change-Id: I59b020c26b0acae0b938685081d9932bd98df5c9
James Berry [Wed, 22 Jun 2011 16:41:17 +0000 (12:41 -0400)]
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
Alexis Ballier [Wed, 22 Jun 2011 16:48:27 +0000 (12:48 -0400)]
Fix parallel install
Require the destination to be present before trying to create the symlink.
See: http://bugs.gentoo.org/show_bug.cgi?id=323805
Change-Id: I14ed4a9792dedc289885a9a43bc5a86cb792206d
Yaowu Xu [Tue, 21 Jun 2011 16:47:42 +0000 (09:47 -0700)]
Merge "adjusting the calculation of errorperbit"
Scott LaVarnway [Tue, 21 Jun 2011 14:45:51 +0000 (07:45 -0700)]
Merge "Improved vp8dx_decode_bool"
Yunqing Wang [Tue, 21 Jun 2011 14:23:24 +0000 (07:23 -0700)]
Merge "Remove unnecessary bounds checking in motion search"
Attila Nagy [Wed, 15 Jun 2011 10:17:40 +0000 (13:17 +0300)]
configuration, support disabling any subset of ARM arch
Useful for leaving out any version specific asm files.
Change-Id: I233514410eb9d7ca88d2d2c839673122c507fa99
Yaowu Xu [Mon, 20 Jun 2011 23:30:26 +0000 (16:30 -0700)]
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
Scott LaVarnway [Mon, 20 Jun 2011 18:44:16 +0000 (14:44 -0400)]
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
Yunqing Wang [Fri, 17 Jun 2011 18:19:51 +0000 (14:19 -0400)]
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
John Koleszar [Fri, 17 Jun 2011 16:48:32 +0000 (09:48 -0700)]
Merge "Use SSE as BPRED distortion metric consistently"
Ronald S. Bultje [Thu, 16 Jun 2011 17:01:27 +0000 (13:01 -0400)]
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
John Koleszar [Thu, 16 Jun 2011 16:56:05 +0000 (09:56 -0700)]
Merge "Disable specialcase for last frames if the sequence contains ARFs."
John Koleszar [Thu, 16 Jun 2011 14:19:46 +0000 (07:19 -0700)]
Merge "gen_msvs_proj: write boolean for Debug attribute"
James Zern [Wed, 15 Jun 2011 21:42:06 +0000 (14:42 -0700)]
gen_msvs_proj: write boolean for Debug attribute
Replace =1 with =true for yasm tool element. This aids in upgrading
e.g., vs9 project files to vs10.
build/x86-msvs/yasm.xml generated during conversion will require the
Separator attribute to be removed for the build to complete
successfully.
Change-Id: If75c4f9a925529740048882003e9d766c5ac4f0c
John Koleszar [Wed, 15 Jun 2011 17:18:06 +0000 (10:18 -0700)]
Merge "Remove redundant check for KEY_FRAME in multithreaded decoder"
John Koleszar [Mon, 13 Jun 2011 15:21:01 +0000 (11:21 -0400)]
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
John Koleszar [Wed, 15 Jun 2011 14:09:09 +0000 (07:09 -0700)]
Merge "fix --disable-runtime-cpu-detect on x86"
Ronald S. Bultje [Wed, 15 Jun 2011 13:47:00 +0000 (09:47 -0400)]
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
Attila Nagy [Wed, 15 Jun 2011 10:27:39 +0000 (13:27 +0300)]
Guard vpx_config.h against multiple inclusions
Change-Id: Iabe2be73af2b92c53687755b31b77448fba385d2
Attila Nagy [Mon, 13 Jun 2011 12:13:10 +0000 (15:13 +0300)]
Remove redundant check for KEY_FRAME in multithreaded decoder
For Intra blocks is enough to check ref_frame == INTRA_FRAME.
Change-Id: I3e2d3064c7642658a9e14011a4627de58878e366
Scott LaVarnway [Tue, 14 Jun 2011 19:04:50 +0000 (12:04 -0700)]
Merge "Populate bmi for B_PRED only"
Johann [Tue, 14 Jun 2011 15:31:50 +0000 (11:31 -0400)]
fix --disable-runtime-cpu-detect on x86
Change-Id: Ib8e429152c9a8b6032be22b5faac802aa8224caa
Paul Wilkins [Tue, 14 Jun 2011 13:01:44 +0000 (06:01 -0700)]
Merge "Fix RT only build"
Tero Rintaluoma [Tue, 14 Jun 2011 10:39:06 +0000 (13:39 +0300)]
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
Tero Rintaluoma [Tue, 14 Jun 2011 08:29:35 +0000 (11:29 +0300)]
Update -linux-rvct targets
- Updated -linux-rvct targets to support RVDS 4.0 and later.
- Changed optimization flag to -Otime because -O3 ruined performance
for RVCT linux targets.
- Added support for --enable-small for RVCT
- RVCT created library should be able to link with GCC
- Supports building shared linux libraries
Change-Id: Ic62589950d86c3420fd4d908b8efb870806d1233
James Zern [Tue, 14 Jun 2011 00:29:49 +0000 (17:29 -0700)]
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
Scott LaVarnway [Mon, 13 Jun 2011 21:14:11 +0000 (17:14 -0400)]
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
Scott LaVarnway [Mon, 13 Jun 2011 13:50:54 +0000 (09:50 -0400)]
Calc ref_frame_cost once per frame
instead of every macro block.
Change-Id: I2604e94c6b89e3a8457777e21c8c38406d55b165
Tero Rintaluoma [Fri, 10 Jun 2011 11:05:53 +0000 (14:05 +0300)]
Fix make clean for asm offset files
Automatically created assembly offset files added to CLEAN-OBJS list
for proper cleanup. This will fix following build error:
1) Build for the workstation
./conigure
make
make clean
2) Build for ARM platform
./configure --target=armv7-linux-gcc
make ==> this will fail because it uses old asm_*_offset.asm files
Change-Id: Id5275c470390ca81b8db086a15ad75af39b80703
John Koleszar [Thu, 9 Jun 2011 20:39:47 +0000 (13:39 -0700)]
Merge "bug fix mode_info_context not initialized for error-resilient"
Yaowu Xu [Wed, 8 Jun 2011 20:45:29 +0000 (13:45 -0700)]
remove one set of 16x16 variance funcations
call to this set of functions are replaced by var16x16.
Change-Id: I5ff1effc6c1358ea06cda1517b88ec28ef551b0d
James Berry [Thu, 9 Jun 2011 16:46:31 +0000 (12:46 -0400)]
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
John Koleszar [Wed, 8 Jun 2011 16:09:02 +0000 (12:09 -0400)]
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
Johann [Wed, 8 Jun 2011 18:43:34 +0000 (14:43 -0400)]
use GCC inline magic
Better fix for #326. ICC happens to support the inline magic
Change-Id: Ic367eea608c88d89475cb7b05d73500d2a1bc42b
Johann [Wed, 8 Jun 2011 18:47:18 +0000 (11:47 -0700)]
Merge "Revert "Use shared object files for ELF""
Johann [Wed, 8 Jun 2011 18:36:04 +0000 (11:36 -0700)]
Revert "Use shared object files for ELF"
Broke RVCT. New magic coming for ICC. Stay tuned!
This reverts commit
c73eb2ffff674e4e9fa7d5209d531aeec6ac5edd
John Koleszar [Wed, 8 Jun 2011 17:59:30 +0000 (10:59 -0700)]
Merge "vp8_pick_inter_mode: remove best_bmodes"
John Koleszar [Wed, 8 Jun 2011 17:58:36 +0000 (10:58 -0700)]
Merge "vp8_pick_intra_mode: correct returned rate"
John Koleszar [Wed, 8 Jun 2011 17:51:50 +0000 (10:51 -0700)]
Merge "Move RD intra block mode selection to rdopt.c"
John Koleszar [Wed, 8 Jun 2011 17:50:50 +0000 (13:50 -0400)]
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
Scott LaVarnway [Wed, 8 Jun 2011 17:20:28 +0000 (10:20 -0700)]
Merge "Removed unused function parameters"
John Koleszar [Wed, 8 Jun 2011 17:08:12 +0000 (13:08 -0400)]
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
Scott LaVarnway [Wed, 8 Jun 2011 17:01:09 +0000 (13:01 -0400)]
Removed unused function parameters
Change-Id: Ib641c624faec28ad9eb99e2b5de51ae74bbcb2a2
Yaowu Xu [Tue, 7 Jun 2011 20:59:46 +0000 (13:59 -0700)]
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
Yaowu Xu [Wed, 8 Jun 2011 16:30:31 +0000 (09:30 -0700)]
Merge "Further activity masking changes:"
John Koleszar [Wed, 8 Jun 2011 16:05:05 +0000 (12:05 -0400)]
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
John Koleszar [Wed, 8 Jun 2011 15:24:52 +0000 (11:24 -0400)]
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
Paul Wilkins [Wed, 8 Jun 2011 15:00:59 +0000 (16:00 +0100)]
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
Yaowu Xu [Tue, 7 Jun 2011 23:36:37 +0000 (16:36 -0700)]
Merge "remove redundant functions"
Yaowu Xu [Tue, 7 Jun 2011 19:37:04 +0000 (12:37 -0700)]
Merge "adjust sad per bit constants"
Yaowu Xu [Fri, 20 May 2011 06:12:40 +0000 (23:12 -0700)]
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
John Koleszar [Tue, 7 Jun 2011 19:30:37 +0000 (12:30 -0700)]
Merge "Reduce overshoot in 1 pass rate control"
Scott LaVarnway [Tue, 7 Jun 2011 13:53:51 +0000 (06:53 -0700)]
Merge "Wrapped asserts in critical code with CONFIG_DEBUG"
Scott LaVarnway [Tue, 7 Jun 2011 13:51:24 +0000 (06:51 -0700)]
Merge "Removed unused function vp8_treed_read_num"
John Koleszar [Tue, 7 Jun 2011 13:44:32 +0000 (06:44 -0700)]
Merge "Use shared object files for ELF"
Scott LaVarnway [Tue, 7 Jun 2011 13:34:47 +0000 (09:34 -0400)]
Wrapped asserts in critical code with CONFIG_DEBUG
Change-Id: I5b0aaca06f2e0f40588cb24fb0642b6865da8970
Scott LaVarnway [Tue, 7 Jun 2011 13:32:51 +0000 (09:32 -0400)]
Removed unused function vp8_treed_read_num
Change-Id: Id66e70540ee7345876f099139887c1843093907f
Yaowu Xu [Mon, 6 Jun 2011 23:42:58 +0000 (16:42 -0700)]
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
Yunqing Wang [Thu, 2 Jun 2011 18:47:06 +0000 (14:47 -0400)]
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
Johann [Mon, 6 Jun 2011 20:42:58 +0000 (13:42 -0700)]
Merge "neon fast quantize block pair"
Johann [Mon, 6 Jun 2011 20:32:13 +0000 (13:32 -0700)]
Merge "adds preload for armv6 encoder asm"
Johann [Tue, 31 May 2011 15:59:04 +0000 (11:59 -0400)]
Use shared object files for ELF
Fixes #326
Change-Id: I5f2a4257430ef62f674190acefd43a0474821288
Scott LaVarnway [Mon, 6 Jun 2011 14:57:39 +0000 (07:57 -0700)]
Merge "Removed unnecessary bmi motion vector stores."
John Koleszar [Mon, 6 Jun 2011 14:02:29 +0000 (07:02 -0700)]
Merge "Don't allow very short GF groups even when the GF is predicted from an ARF."
John Koleszar [Tue, 3 May 2011 15:17:05 +0000 (11:17 -0400)]
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
Scott LaVarnway [Fri, 3 Jun 2011 17:05:16 +0000 (13:05 -0400)]
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
Scott LaVarnway [Fri, 3 Jun 2011 15:32:30 +0000 (08:32 -0700)]
Merge "Removed B_MODE_INFO"
Yunqing Wang [Thu, 2 Jun 2011 21:33:17 +0000 (17:33 -0400)]
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
Scott LaVarnway [Thu, 2 Jun 2011 17:46:41 +0000 (13:46 -0400)]
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
Ronald S. Bultje [Thu, 2 Jun 2011 16:14:51 +0000 (09:14 -0700)]
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
Yaowu Xu [Thu, 2 Jun 2011 15:58:03 +0000 (08:58 -0700)]
Merge "further clean up of errorperbit and sadperbit"
Yaowu Xu [Wed, 1 Jun 2011 18:41:05 +0000 (11:41 -0700)]
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
John Koleszar [Wed, 1 Jun 2011 20:57:23 +0000 (13:57 -0700)]
Merge "Bugfix in vp8dx_set_reference"
Henrik Lundin [Wed, 1 Jun 2011 19:41:12 +0000 (21:41 +0200)]
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
John Koleszar [Wed, 1 Jun 2011 18:14:17 +0000 (11:14 -0700)]
Merge "Fix code under #if CONFIG_INTERNAL_STATS."
Ronald S. Bultje [Wed, 1 Jun 2011 18:10:13 +0000 (11:10 -0700)]
Fix code under #if CONFIG_INTERNAL_STATS.
Change-Id: Iccbd78d91c3071b16fb3b2911523a22092652ecd
Yaowu Xu [Thu, 1 Jul 2010 01:58:54 +0000 (18:58 -0700)]
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
Tero Rintaluoma [Mon, 9 May 2011 07:09:41 +0000 (10:09 +0300)]
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
Scott LaVarnway [Tue, 31 May 2011 19:31:46 +0000 (12:31 -0700)]
Merge "vp8_pick_inter_mode code cleanup"
Scott LaVarnway [Tue, 31 May 2011 18:24:42 +0000 (14:24 -0400)]
vp8_pick_inter_mode code cleanup
Small code cleanups before attempting to reduce the size
of bmi found in BLOCKD.
Change-Id: Ie9c14adb53afd847716a75bcce067d0e6c04f225
John Koleszar [Tue, 31 May 2011 16:37:45 +0000 (12:37 -0400)]
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
Tero Rintaluoma [Mon, 30 May 2011 08:10:03 +0000 (11:10 +0300)]
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
John Koleszar [Fri, 27 May 2011 19:35:06 +0000 (12:35 -0700)]
Merge "bug fix check frame buffer index before copy"
James Berry [Fri, 27 May 2011 18:57:25 +0000 (14:57 -0400)]
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
Yunqing Wang [Fri, 27 May 2011 18:12:50 +0000 (11:12 -0700)]
Merge "Use hex search for realtime mode speed>4"
Yunqing Wang [Fri, 27 May 2011 17:31:15 +0000 (13:31 -0400)]
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
Scott LaVarnway [Fri, 27 May 2011 14:52:04 +0000 (07:52 -0700)]
Merge "Broken EC after MODE_INFO size reduction"
Yunqing Wang [Fri, 27 May 2011 14:25:25 +0000 (07:25 -0700)]
Merge "Remove unused code"
Yunqing Wang [Fri, 27 May 2011 14:20:49 +0000 (10:20 -0400)]
Remove unused code
Hex search is not called in rdopt.c
Change-Id: I67347f03e13684147a7c77fb9e9147e440bb5e8e
Scott LaVarnway [Thu, 26 May 2011 19:13:00 +0000 (15:13 -0400)]
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
John Koleszar [Thu, 26 May 2011 16:58:26 +0000 (09:58 -0700)]
Merge "Do not copy data between encoder reference buffers."