James Zern [Sat, 11 Oct 2014 09:27:23 +0000 (11:27 +0200)]
args.h: add arg_parse_enum prototype
Change-Id: I64d6e7c2d3f76528ab409ce381267fc5cef13ab1
James Zern [Sat, 11 Oct 2014 09:24:52 +0000 (11:24 +0200)]
vp9_spatial_svc_encoder: fix -bit-depth arg parsing
use arg_parse_enum_or_int like vpxenc. this also fixes a warning as
arg_parse_enum is not currently declared in args.h.
Change-Id: If9ce258d6adb6286eb86f529083929d5fe2b3a56
James Zern [Sat, 11 Oct 2014 09:14:54 +0000 (11:14 +0200)]
yv12config: fix highbitdepth build
Change-Id: Ie8a4caae19fb514d5bd22e0ae35ca0edcd3484ae
Alex Converse [Fri, 10 Oct 2014 18:31:58 +0000 (11:31 -0700)]
Revert "Add adaptation option for VBR."
This reverts commit
869d4ca51957614dcf5093ebb9e322cc8a8405ca.
This breaks the build via conflict with
e18edd5eb651f9b7563cbd829744807402bfe0d8.
Change-Id: If544b99e367a449452834eb8cce600f58c34ec0d
hkuang [Fri, 10 Oct 2014 17:40:21 +0000 (10:40 -0700)]
Merge "Optimize the code to set the refernce frame right after reading the header."
hkuang [Fri, 10 Oct 2014 17:01:05 +0000 (10:01 -0700)]
Merge "Correct the code format."
Paul Wilkins [Fri, 10 Oct 2014 16:22:58 +0000 (09:22 -0700)]
Merge "Add adaptation option for VBR."
Yaowu Xu [Fri, 10 Oct 2014 16:16:19 +0000 (09:16 -0700)]
Merge "vp9/choose_partitioning: add missing clear_system_state"
James Zern [Fri, 10 Oct 2014 15:14:46 +0000 (08:14 -0700)]
vp9/choose_partitioning: add missing clear_system_state
set_vt_partitioning does double math
Change-Id: I8e9d73d5c89b937a5326abf04164d24d9d88c5ef
James Zern [Fri, 10 Oct 2014 13:48:04 +0000 (06:48 -0700)]
Merge changes Ib5030373,Id614cfe6
* changes:
test-data.mk: add some sectional comments
add test-data.mk
Paul Wilkins [Thu, 9 Oct 2014 13:49:41 +0000 (14:49 +0100)]
Add adaptation option for VBR.
Allow min and maxQ to creep when the undershoot
or overshoot exceeds thresholds controlled by the
command line under_shoot_pct and over_shoot_pct
values.
Default is 100%,100% which ~disables adaptation.
Derf results for example undershoot% / overshoot%:-
Head:- Mean abs (%rate error) = 14.4%
This check in:-
25%/25% - Mean abs (%rate error) = 6.7%
PSNR hit -1% SSIM -0.1%
5% / 5% - Mean abs (%rate error) = 2.2%
PSNR hit -3.3% SSIM - 1.1%
Most of the remaining error and most of the quality hit is
at extreme data rates. The adaptation code still has an
exception for material that is in effect static so that we
don't over adjust and over spend on YT slide show type
content.
Change-Id: If25a2449a415449c150acff23df713e9598d64c9
James Zern [Fri, 10 Oct 2014 08:05:47 +0000 (10:05 +0200)]
vp9_avg_intrin_sse2: correct intrinsics include
immintrin.h -> emmintrin.h
fixes build where newer intrinsics are unavailable
Change-Id: I79311b39bfa782fc2abeb45884ecb417050cb9f8
hkuang [Fri, 10 Oct 2014 00:57:45 +0000 (17:57 -0700)]
Correct the code format.
Change-Id: If2de420f8123a4e8bf635dd29205dd74ee174eee
hkuang [Thu, 9 Oct 2014 23:32:36 +0000 (16:32 -0700)]
Optimize the code to set the refernce frame right after reading the header.
Change-Id: I495cf4a366e06e3220ed132500b1ba1c8448f708
hkuang [Thu, 9 Oct 2014 22:44:08 +0000 (15:44 -0700)]
Merge "Remove unnecessary code."
hkuang [Thu, 9 Oct 2014 22:43:31 +0000 (15:43 -0700)]
Merge "Remove unnecessary scale check in set_ref."
Deb Mukherjee [Thu, 9 Oct 2014 22:39:56 +0000 (15:39 -0700)]
Merge "Rename highbitdepth functions to use highbd prefix"
hkuang [Thu, 9 Oct 2014 22:12:12 +0000 (15:12 -0700)]
Remove unnecessary code.
Function will jump to error handler when ref buffer is corrupted.
So "xd->corrupted |= ref_buffer->buf->corrupted;" is useless.
Change-Id: I35353a0637ad0dbb682454e040ef69fa68280bfa
Deb Mukherjee [Wed, 8 Oct 2014 19:43:22 +0000 (12:43 -0700)]
Rename highbitdepth functions to use highbd prefix
Uses highbd_ prefix convention consistently.
Change-Id: I58f7f799a7ff8e32701bcd71c955bcf1cdd4581e
hkuang [Thu, 9 Oct 2014 18:58:39 +0000 (11:58 -0700)]
Remove unnecessary scale check in set_ref.
Scale check has been done in read_inter_block_mode_info.
Change-Id: I6c86f93bd579109ed30ff13a04a30e35f5ae6fc5
Jingning Han [Thu, 9 Oct 2014 18:16:11 +0000 (11:16 -0700)]
Merge "Remove sub8x8 block index from rd_pick_partition argument"
Deb Mukherjee [Thu, 9 Oct 2014 18:14:51 +0000 (11:14 -0700)]
Merge "Subpel search cleanups and enhancements"
Alex Converse [Thu, 9 Oct 2014 18:12:02 +0000 (11:12 -0700)]
Merge "Move the high freq coeff check outside store_coding_context"
Jingning Han [Thu, 9 Oct 2014 02:34:00 +0000 (19:34 -0700)]
Remove sub8x8 block index from rd_pick_partition argument
This parameter is deprecated. Its function is replaced with
other explicit condition check.
Change-Id: I61337e350ba8ca9eb50382db8b4d4acbf45cb7eb
Marco [Thu, 9 Oct 2014 15:41:49 +0000 (08:41 -0700)]
Merge "vp8: Suppress denoising with respect to old reference frames."
Yaowu Xu [Thu, 9 Oct 2014 14:08:48 +0000 (07:08 -0700)]
Merge "Fix src frame buffer copy and extend"
James Zern [Fri, 3 Oct 2014 23:41:56 +0000 (16:41 -0700)]
test-data.mk: add some sectional comments
+ sort
Change-Id: Ib503037376368cec0cf57cbedeb215ab0eac8304
James Zern [Fri, 3 Oct 2014 23:18:58 +0000 (16:18 -0700)]
add test-data.mk
splits the test sources (test.mk) and the test data.
Change-Id: Id614cfe68bfbc09b0e429a37a21ec7a9dcdfadc6
James Zern [Thu, 9 Oct 2014 10:53:01 +0000 (03:53 -0700)]
Merge "set_vt_partitioning: fix type conversion warning"
James Zern [Thu, 9 Oct 2014 09:41:36 +0000 (11:41 +0200)]
set_vt_partitioning: fix type conversion warning
double -> int64
+ make threshold_multiplier an int
Change-Id: I6d3607fdf13d670f57c9d9b04a80acb2be1346a0
James Zern [Thu, 9 Oct 2014 08:44:42 +0000 (10:44 +0200)]
vp9_rtcd_defs: fix vp9_avg_8x8 declaration
vp9_avg_8x8 does not depend on x86inc, fixes 32-bit OS X build
Change-Id: I709b874ea84bf57c8cdb5ac7d43eecc6b8c1a2dd
Deb Mukherjee [Tue, 7 Oct 2014 09:48:08 +0000 (02:48 -0700)]
Subpel search cleanups and enhancements
- Some fixes to surface fit.
- Returns variance function as cost rather than sad in the
pattern search and diamond search functions. Only
vp9_pattern_search_sad function used in bigdia search
uses sad as integer 1-away costs.
- Deploys SUBPEL_TREE_PRUNED_MORE for speed 4+.
Results:
derf [Speed 3]: About +0.036% in coding efficiency without any
discernible speed loss.
derf [Speed 4]: About 2-3% faster at -0.199% loss in coding efficiency.
derf [Speed 5]: About 3-4% faster at -0.149% loss in coding efficiency.
Change-Id: I8462f94f6adb46966ca964f2bd0400977357fd63
Yunqing Wang [Thu, 9 Oct 2014 02:58:18 +0000 (19:58 -0700)]
Merge "Allow mode search breakout at very low prediction errors"
Yunqing Wang [Thu, 2 Oct 2014 23:25:52 +0000 (16:25 -0700)]
Allow mode search breakout at very low prediction errors
In model_rd_for_sb function, the spatial domain SSE and variance
are checked to see if transform coefficients are quantized to 0.
Besides that, this patch adds another set of thresholds that are
much more strict. These thresholds are used to conduct a partition
block level check to measure if all its TX blocks are skippable
for YUV planes. If it is true, x->skip is set for this partition
block, and thus its mode search is terminated.
This speeds up the encoding at very low prediction error case,
such as screen sharing application. This patch covers what
rd_encode_breakout_test() does, so that function is removed.
Borg test at speed 3 shows:
For stdhd set, psnr: +0.008%, ssim: +0.014%;
For derf set, psnr: +0.018%, ssim: +0.025%.
No noticeable speed change.
Change-Id: I4e5f15cf10016a282a68e35175ff854b28195944
Jingning Han [Wed, 8 Oct 2014 22:02:37 +0000 (15:02 -0700)]
Move the high freq coeff check outside store_coding_context
This fixes valgrind message issue 870.
Change-Id: Ibbc2481923a2995029ab05de30c9e8a6e9f0f9a8
Jingning Han [Wed, 8 Oct 2014 00:10:38 +0000 (17:10 -0700)]
Use local variable in vp9_rd_pick_inter_mode_sb
Change-Id: Ie35a965a6b8de536ccaf61ff61498620d22db205
Jingning Han [Wed, 8 Oct 2014 22:25:09 +0000 (15:25 -0700)]
Merge "Clean up header files in vp9_blockd.h and related files"
Jingning Han [Wed, 8 Oct 2014 22:24:56 +0000 (15:24 -0700)]
Merge "Use #define statement for MAX_MB_PLANE"
Marco [Tue, 7 Oct 2014 23:15:32 +0000 (16:15 -0700)]
vp8: Suppress denoising with respect to old reference frames.
If the GOLDEN or ALTREF frame was last updated > x frames in the past,
don't use them for denoising (only consider LAST). Using an old reference
frame for denoising, e.g., if it is a long-term reference or the last key frame,
can cause some visible artifacts, in particular in the aggressive denoising mode.
Change-Id: I239c9fbb092c36cba7e95328f1fa67a58d6a7ed9
Deb Mukherjee [Wed, 8 Oct 2014 19:46:37 +0000 (12:46 -0700)]
Fixes naming bug from previous highbitdepth patch
Change-Id: Ife939930309ceea730dd5df594b8f36e6644bf36
Deb Mukherjee [Wed, 8 Oct 2014 18:44:50 +0000 (11:44 -0700)]
Merge "Refactoring of highbitdepth bitshift operations"
JackyChen [Wed, 8 Oct 2014 18:41:14 +0000 (11:41 -0700)]
Merge "Add unit test for VP8 denoiser."
Yaowu Xu [Wed, 8 Oct 2014 16:06:36 +0000 (09:06 -0700)]
Fix src frame buffer copy and extend
For input source with size that is not multiple of 8, the size is
rounded to 8 and saved in width or height, the original source sizes
are saved in crop_width and crop_height. This commit corrects the
computation of bottom and right extension amounts to use the orignal
sizes, hence crop_width and crop_height.
In addition, this commit also adds the missed initialization for
uv_crop_width and uv_crop_height.
This addresses issue #834
Change-Id: I084543ca7645a4964b88f7cf8ff668f517d3a39b
Deb Mukherjee [Tue, 7 Oct 2014 03:46:11 +0000 (20:46 -0700)]
Refactoring of highbitdepth bitshift operations
Moves the high bitdepth bit-shift operations to tools_common.c
Change-Id: Ifab91574362609cd224d2d6bae5fde00652eb3e4
JackyChen [Tue, 7 Oct 2014 22:30:23 +0000 (15:30 -0700)]
Add unit test for VP8 denoiser.
The unit test is to check the bitexactness of SSE2 code with C code.
Change-Id: I13cc08a557b02357e5d6c2db7a49e93ce0d8e16f
Jim Bankoski [Wed, 8 Oct 2014 16:04:26 +0000 (09:04 -0700)]
Merge "experimental : partition using 1/8 x 1/8 image"
Jim Bankoski [Wed, 8 Oct 2014 16:04:01 +0000 (09:04 -0700)]
Merge "Force better lower quantizer keyframe in case of high quantizer."
Yaowu Xu [Wed, 8 Oct 2014 15:38:15 +0000 (08:38 -0700)]
Revert "Change the y4m buffer allocation size"
This reverts commit
72ef1949388c4c0d82769ef06c6db0195b124b13.
Change-Id: Ic1906bcc6da05d2e6529152b92df809c8e5a2406
Paul Wilkins [Wed, 8 Oct 2014 11:23:30 +0000 (04:23 -0700)]
Merge "Improve two pass VBR accuracy."
Paul Wilkins [Wed, 8 Oct 2014 11:23:20 +0000 (04:23 -0700)]
Merge "Two pass rc changes."
Jim Bankoski [Tue, 7 Oct 2014 23:36:14 +0000 (16:36 -0700)]
experimental : partition using 1/8 x 1/8 image
The concept:
There's too much noise in source pixels for variance and at low bitrate
the reconstructed looks nothing like the source so we have problems
getting good partitionings with either. This skirts the issue by using
a box blur scaled down version for variance calculations. To compare
against source_var_ moved keyframe to be rd based like source_var.
Change-Id: Ie3babdbfadae324b7b5a76bea192893af27f0624
Jim Bankoski [Mon, 22 Sep 2014 20:58:09 +0000 (13:58 -0700)]
Force better lower quantizer keyframe in case of high quantizer.
Change-Id: Ie69a164bc166b6a8819777038d65a7d9f9c3361f
Jingning Han [Tue, 7 Oct 2014 22:34:19 +0000 (15:34 -0700)]
Merge "Remove extra empty line"
Jingning Han [Tue, 7 Oct 2014 22:34:07 +0000 (15:34 -0700)]
Merge "Remove vp9_blockd.h from vp9_common_data.c"
Jingning Han [Tue, 7 Oct 2014 22:33:52 +0000 (15:33 -0700)]
Merge "Replace mi_width_log2() with mi_width_log2_lookup table"
Jingning Han [Tue, 7 Oct 2014 22:33:45 +0000 (15:33 -0700)]
Merge "Take out repeated block width/height lookup functions"
Jingning Han [Tue, 7 Oct 2014 22:12:05 +0000 (15:12 -0700)]
Clean up header files in vp9_blockd.h and related files
This commit breaks the overly broad header files into more
targeted and smaller ones, to help better structure the system
layout.
Change-Id: I7b24559d3ea6e582cf5d9bbe8f71459f9824d71b
Yunqing Wang [Tue, 7 Oct 2014 22:13:44 +0000 (15:13 -0700)]
Merge "Fix skip_txfm issue in rdopt code"
Jingning Han [Tue, 7 Oct 2014 22:00:22 +0000 (15:00 -0700)]
Use #define statement for MAX_MB_PLANE
Change-Id: I3a7f83ab1dbfcedc8a82fe798c2fa30dd9c7d696
Jingning Han [Tue, 7 Oct 2014 21:06:37 +0000 (14:06 -0700)]
Remove extra empty line
Change-Id: I6f2865bb8ba9295f5c45a4cad065aecbe1e63c32
Jingning Han [Tue, 7 Oct 2014 20:42:26 +0000 (13:42 -0700)]
Merge "Move inter filter defs to vp9_filter.h"
Jingning Han [Tue, 7 Oct 2014 20:42:17 +0000 (13:42 -0700)]
Merge "Reduce the scope of the header file used in vp9_context_tree.h"
Jingning Han [Tue, 7 Oct 2014 20:42:13 +0000 (13:42 -0700)]
Merge "Remove redundant header file from vp9_encoder.h"
Jingning Han [Tue, 7 Oct 2014 20:02:54 +0000 (13:02 -0700)]
Remove vp9_blockd.h from vp9_common_data.c
The basic data defs should be above block operation level.
Change-Id: I7dd9836d01120ab75e0c472baac9f15495ed0db5
Yunqing Wang [Tue, 7 Oct 2014 19:30:33 +0000 (12:30 -0700)]
Fix skip_txfm issue in rdopt code
Fixed an encoder crash. Set skip_txfm to 0 for cases that skip_txfm
isn't calculated. Put memcpy of skip_txfm at right place.
Change-Id: Ib3b6afc1b251a85b2a853c8138fb3393f48cfef6
Jingning Han [Tue, 7 Oct 2014 19:45:25 +0000 (12:45 -0700)]
Replace mi_width_log2() with mi_width_log2_lookup table
Change-Id: If0ea98aa139d14d40cd924114e18396aff36b5a5
Yaowu Xu [Tue, 7 Oct 2014 19:36:56 +0000 (12:36 -0700)]
Merge "Add range checking for decoded coefficients."
Jingning Han [Tue, 7 Oct 2014 19:30:33 +0000 (12:30 -0700)]
Take out repeated block width/height lookup functions
The functions b_width_log2 and b_height_log2 only do direct
table fetch. This commit unifies such use cases by using the
table directly and removes these functions.
Change-Id: I3103fc6ba959c1182886a2799d21b8b77c8a7b6b
Jingning Han [Tue, 7 Oct 2014 19:13:33 +0000 (12:13 -0700)]
Move inter filter defs to vp9_filter.h
Add comments on the use case of these definitions. Further reduce
the scope of header file in vp9_context_tree.h.
Change-Id: Ic4a7638e838d0ac441b64abfc56e57354c059d75
Deb Mukherjee [Tue, 7 Oct 2014 19:15:26 +0000 (12:15 -0700)]
Merge "Resolves some static analysis / undefined warnings"
Yaowu Xu [Tue, 7 Oct 2014 18:22:09 +0000 (11:22 -0700)]
Add range checking for decoded coefficients.
The coefficient range checking is enabled when configured with
--enable-debug --enable-coefficient-range-checking
for vpxdec to detect ill-formed input stream. This addresses the
problem raised by issue #792.
Change-Id: I3f9ea541de4dc742dd64389d6c5f543fb1c4f052
Deb Mukherjee [Mon, 6 Oct 2014 21:30:01 +0000 (14:30 -0700)]
Resolves some static analysis / undefined warnings
Also fixes a case of distortion becoming negative and messing
up the RDCOST computation.
Change-Id: Id345af9e8dfff31ade622be5756e51f2cdface53
Jingning Han [Tue, 7 Oct 2014 18:01:05 +0000 (11:01 -0700)]
Reduce the scope of the header file used in vp9_context_tree.h
Change-Id: I264ee35044a5973c7725daba7af870968353a3c1
JackyChen [Tue, 7 Oct 2014 17:51:55 +0000 (10:51 -0700)]
Merge "Add SSE2 code and unit test for VP9 denoiser."
Jingning Han [Tue, 7 Oct 2014 17:49:43 +0000 (10:49 -0700)]
Remove redundant header file from vp9_encoder.h
Change-Id: Ia212390cf8d36db5436bb0f0e1b696f70066341a
Jingning Han [Mon, 6 Oct 2014 22:53:27 +0000 (15:53 -0700)]
Fix eobs buffer pointer mis-use
This commit fixes a buffer pointer mis-use in store_coding_context.
The compression performance for stdhd set of speed 3 is improved by
0.097%. It fixes issue 869.
Change-Id: Idc59e22035eaf39f7133ca04174894374d647ff7
JackyChen [Thu, 18 Sep 2014 23:45:53 +0000 (16:45 -0700)]
Add SSE2 code and unit test for VP9 denoiser.
This SSE2 is based on VP8 denoiser's SSE2 code. In VP8, there are
only 16x16 blocks in denoiser, while in VP9, there are 13 different
block sizes.
By adding this SSE2 code, the improvement of encoder speed is around
20%(using C code vs using SSE2 code), vary for different clips.
The unit test for VP9 denoiser is to confirm that the SSE2 code is
bit-exact with the C code. The unit test covers all block size.
Change-Id: Ic8d8ac26db4ea40a5f146b5678a065af07eaaa3d
Jingning Han [Mon, 6 Oct 2014 17:18:17 +0000 (10:18 -0700)]
Add range check in inverse ADST 16x16
Bit-stream clarification related to Issue 868.
Change-Id: I92a7bc5b7782c9ea5c3f6cceec761742183c9514
Yaowu Xu [Mon, 6 Oct 2014 16:40:15 +0000 (09:40 -0700)]
Merge "Change the y4m buffer allocation size"
Jingning Han [Mon, 6 Oct 2014 16:29:29 +0000 (09:29 -0700)]
Merge "Fix an IOC issue in vp9_rd_pick_inter_mode_sb"
Yaowu Xu [Mon, 6 Oct 2014 14:57:36 +0000 (07:57 -0700)]
Merge "Properly initialize segmentID in nonrd coding path"
Paul Wilkins [Fri, 26 Sep 2014 16:20:01 +0000 (17:20 +0100)]
Improve two pass VBR accuracy.
Adjustments to the GF interval choice and minimum boost.
Adjustment to the calculation of 2 pass worst q.
Compared to 09/29 head there is metrics hit on derf of
(-0.123%,-0.191%)
Compared to the September 29 head and a baseline on
September 18 baseline the accuracy of the VBR rate control
measured on the derf set is as follows:-
Mean error % / Mean abs(error %)
Sept 18 baseline (-7.0% / 14.76%)
Sept 29 head (-15.7%, 19.8%)
This check in (-1.5% / 14.4%)
The mean undershoot is reduced slightly but the
worst case overshoot on e.g. harbour/highway is
increased. This will be addressed in a later patch.
Change-Id: Iffd9b0ab7432a131c98fbaaa82d1e5b40be72b58
Jingning Han [Sun, 5 Oct 2014 19:05:14 +0000 (12:05 -0700)]
Fix an IOC issue in vp9_rd_pick_inter_mode_sb
It is possible that the GOLDEN reference frame is not avaiable, in
which setting the predicted mv will be associated with a residual
value of INT_MAX. This commit checks this condition before
left shift and comparison with that of ALTREF frame, to avoid
overflow issue.
Change-Id: Ib98c3149dbdd016f2fe5beaafb13f67d469dd07c
Deb Mukherjee [Sat, 4 Oct 2014 15:50:30 +0000 (08:50 -0700)]
Some data type changes in vp9_idct.c
Resolves a visual studio warning, and includes some cleanups.
Change-Id: I6a7576ef323c475b7d1c659800cd82c6cb1fd18d
Deb Mukherjee [Sat, 4 Oct 2014 00:10:02 +0000 (17:10 -0700)]
Merge "Remove experimental-bitstream flag for profiles>0"
Yaowu Xu [Fri, 3 Oct 2014 23:14:49 +0000 (16:14 -0700)]
Change the y4m buffer allocation size
This attempts to fix issue #847
Change-Id: Ifa2f96b6b33624004272d83363e089b5b4dc5626
Deb Mukherjee [Fri, 3 Oct 2014 22:41:50 +0000 (15:41 -0700)]
Remove experimental-bitstream flag for profiles>0
Profiles 1-3 are stable enough that we can remove the
experimental bitstream commandline option.
Change-Id: Id2a64a08192d7eac6c45384ee34196330351c6d8
Jingning Han [Fri, 3 Oct 2014 22:24:02 +0000 (15:24 -0700)]
Merge "Fix indent in encode_rd_sb_row"
Jingning Han [Fri, 3 Oct 2014 22:23:54 +0000 (15:23 -0700)]
Merge "Rework partition search skip scheme"
Yaowu Xu [Fri, 3 Oct 2014 21:41:12 +0000 (14:41 -0700)]
Properly initialize segmentID in nonrd coding path
This commit adds proper initialization of segment id for variance AQ
mode in non-rd coding path. It fixes the enc/dec mismatch issue of
rt=7 with --aq-mode=1, as reported in issue #816
Change-Id: I02fa41b96345bf2e66077d5ea553f85ba800f7bb
Deb Mukherjee [Fri, 3 Oct 2014 19:45:39 +0000 (12:45 -0700)]
Merge "Incorporate WRAPLOW macro into non-highbitdepth tx"
Jingning Han [Fri, 3 Oct 2014 17:14:05 +0000 (10:14 -0700)]
Fix indent in encode_rd_sb_row
Change-Id: Icbcfe7b56d88474f4398b4c5b52f6719d551ab4a
Jingning Han [Fri, 3 Oct 2014 00:49:00 +0000 (17:49 -0700)]
Rework partition search skip scheme
This commit enables the encoder to skip split partition search if
the bigger block size has all non-zero quantized coefficients in low
frequency area and the total rate cost is below a certain threshold.
It logarithmatically scales the rate threshold according to the
current block size. For speed 3, the compression performance loss:
derf -0.093%
stdhd -0.066%
Local experiments show 4% - 20% encoding speed-up for speed 3.
blue_sky_1080p, 1500 kbps
51051 b/f, 35.891 dB, 67236 ms ->
50554 b/f, 35.857 dB, 59270 ms (12% speed-up)
old_town_cross_720p, 1500 kbps
14431 b/f, 36.249 dB, 57687 ms ->
14108 b/f, 36.172 dB, 46586 ms (19% speed-up)
pedestrian_area_1080p, 1500 kbps
50812 b/f, 40.124 dB, 100439 ms ->
50755 b/f, 40.118 dB, 96549 ms (4% speed-up)
mobile_calendar_720p, 1000 kbps
10352 b/f, 35.055 dB, 51837 ms ->
10172 b/f, 35.003 dB, 44076 ms (15% speed-up)
Change-Id: I412e34db49060775b3b89ba1738522317c3239c8
Deb Mukherjee [Thu, 2 Oct 2014 22:43:27 +0000 (15:43 -0700)]
Incorporate WRAPLOW macro into non-highbitdepth tx
Incorporates the WRAPLOW macro into the non-highbitdepth transforms
to aid hardware verification between a software C model and an
intended hardware implementation though the use of the configure
options: --enable-experimental --enable-emulate-hardware.
Note that to avoid further discrepancies between the sse/sse2
implementations of the transforms and the C implementation, when the
emulate hardware option is invoked, we also disable sse/sse2/etc.
Also incudes some minor cleanups/renaming etc.
Change-Id: Ib864d8493313927d429cce402982f1c8e45b3287
Deb Mukherjee [Fri, 3 Oct 2014 18:37:47 +0000 (11:37 -0700)]
Merge "Prevent negative cost for highbitdepth"
Deb Mukherjee [Fri, 3 Oct 2014 17:22:21 +0000 (10:22 -0700)]
Prevent negative cost for highbitdepth
Adds proper scaling for highbitdepth in a rdopt cost.
Change-Id: I066694799a7f491b830945ef1c66eb202071c355
Deb Mukherjee [Fri, 3 Oct 2014 07:09:41 +0000 (00:09 -0700)]
rdmult data type change
To fix a VS warning.
Change-Id: I4c530c0afe8d06acdb8cc78b7995aba57a25373d
Deb Mukherjee [Fri, 3 Oct 2014 03:06:04 +0000 (20:06 -0700)]
Merge "Adds highbitdepth support to svc examples"
Deb Mukherjee [Thu, 2 Oct 2014 18:43:05 +0000 (11:43 -0700)]
Adds highbitdepth support to svc examples
Change-Id: I59946642cb5c370726da33f4448a3deaba7d3f11
Yaowu Xu [Thu, 2 Oct 2014 22:15:58 +0000 (15:15 -0700)]
Merge "Make iscan and scan neighbor arrays static const."