]>
granicus.if.org Git - libvpx/log
Hui Su [Thu, 14 Jul 2016 00:13:46 +0000 (00:13 +0000)]
Merge "Refactor codes about motion search" into nextgenv2
Jingning Han [Wed, 13 Jul 2016 21:35:29 +0000 (21:35 +0000)]
Merge "Fix highbd inter prediction filter sse4 overwriting issue" into nextgenv2
Jingning Han [Wed, 13 Jul 2016 19:46:59 +0000 (12:46 -0700)]
Fix highbd inter prediction filter sse4 overwriting issue
Properly handle the case where the height is an integer multiple
of 4.
Change-Id: I11ac188c13f78db20902e2e333c60ce76ce837c5
Yue Chen [Wed, 13 Jul 2016 18:40:48 +0000 (18:40 +0000)]
Merge "Optimize and cleanup obmc predictor and rd search." into nextgenv2
hui su [Wed, 6 Jul 2016 17:37:18 +0000 (10:37 -0700)]
Refactor codes about motion search
1. Add "best_mv" in MACROBLOCK to store the best motion vector
during motion search, so that we don't need to pass its pointer
to various motion search functions.
2. Declare some functions as static when possible.
3. Fix some indents.
Change-Id: I0778146c0866cbc55e245988c59222577ea8260e
Geza Lore [Fri, 8 Jul 2016 12:53:39 +0000 (13:53 +0100)]
Optimize and cleanup obmc predictor and rd search.
Use vpx_blend_a64_hmask and vpx_blend_a64_vmask to speed up
computing the obmc predictor. Clean up calc_target_weighted_pred.
Encoder speedup: 1.3%
Decoder speedup: 6.5%
Change-Id: I0c774fe53d22399e92a10d1daf3af0010d88d2c5
Geza Lore [Tue, 12 Jul 2016 10:41:54 +0000 (11:41 +0100)]
Add SSE4.1 vpx_obmc_variance* implementations and cosmetics
Speedup for these functions: 4x
Also include some cosmetic changes to SAD functions
Change-Id: I344c32c795492507ae08742f52d035a13f583799
Pascal Massimino [Wed, 13 Jul 2016 03:09:52 +0000 (03:09 +0000)]
Merge "Clean up FunctionEquivalenceTest." into nextgenv2
Geza Lore [Tue, 12 Jul 2016 14:26:36 +0000 (15:26 +0100)]
Clean up FunctionEquivalenceTest.
remove use of tuple in favor of struct.
Change-Id: If3b1aa5c2fc3cfe1446fff7a8fd270f2ca85fedf
Aamir Anis [Tue, 12 Jul 2016 23:56:37 +0000 (23:56 +0000)]
Merge "Fix for loop filter selection procedure" into nextgenv2
Aamir Anis [Fri, 8 Jul 2016 21:36:36 +0000 (14:36 -0700)]
Fix for loop filter selection procedure
Fixed best error reported by loop filter selection, this value is used
during loop restoration to pick best mode. Baseline remains unchanged,
change in BDRate for loop restoration experiment:
-0.628 -> -0.625 for lowres,
-1.262 -> -1.283 for highres.
Change-Id: I69ef1608bc232b250ac46f59e31fdbed1a999dcd
Yi Luo [Tue, 12 Jul 2016 19:28:48 +0000 (19:28 +0000)]
Merge "HBD convolution filtering (10/12 taps) SSE4.1 optimization" into nextgenv2
Yi Luo [Fri, 8 Jul 2016 22:41:59 +0000 (15:41 -0700)]
HBD convolution filtering (10/12 taps) SSE4.1 optimization
- For experiment EXT_INTERP under high bit depth.
- Add unit test to verify bit-exact.
- Speed performance improvement:
On Xeon E5-2680, park_joy_1080p_12.y4m, 50 frames, encoding time
drops from
6682503 ms to
5390270 ms.
Change-Id: Iea4debf5414f3accf1eb5672abeab56a0539ac77
Geza Lore [Tue, 12 Jul 2016 12:20:04 +0000 (13:20 +0100)]
Cleanup obmc_sad function prototypes.
Name 'wsrc', 'mask' and 'pre' explicitly, rather than
using 'b', 'm' and 'a'.
Change-Id: Iaee6d1ac1211b0b05b47cf98b50570089b12d600
James Zern [Tue, 12 Jul 2016 02:39:24 +0000 (02:39 +0000)]
Merge changes from topic 'missing-proto' into nextgenv2
* changes:
vp10/encoder/rdopt.c: make a function static
vp10/encoder/rd.c: make a function static
vp10_convolve_ssse3.c: make some functions static
vp10/encoder/bitstream.[hc]: correct a prototype
vp10/common/idct.h: add some missing prototypes
highbd_quantize_intrin_sse2.c: add missing rtcd include
vp10: add some missing includes
Yue Chen [Tue, 12 Jul 2016 01:21:32 +0000 (01:21 +0000)]
Merge "Cosmetics for vp10/common/vp10_rtcd_defs.pl" into nextgenv2
James Zern [Sat, 2 Jul 2016 00:50:06 +0000 (17:50 -0700)]
vp10/encoder/rdopt.c: make a function static
+ remove vp10_ prefix
quiets a -Wmissing-prototypes warning
BUG=b/
29584271
Change-Id: I8821c38009b90296280f9b14233e73c92076e81f
James Zern [Sat, 2 Jul 2016 00:48:43 +0000 (17:48 -0700)]
vp10/encoder/rd.c: make a function static
+ remove vp10_ prefix
quiets a -Wmissing-prototypes warning
BUG=b/
29584271
Change-Id: I6b5d71f8120a6d1fee4c782beb4c6d6eef980f65
James Zern [Sat, 2 Jul 2016 00:43:20 +0000 (17:43 -0700)]
vp10_convolve_ssse3.c: make some functions static
quiets -Wmissing-prototypes warnings
BUG=b/
29584271
Change-Id: I4d2eb7f4b45d7b829421976641b3212bcf29e7dd
James Zern [Sat, 2 Jul 2016 00:47:11 +0000 (17:47 -0700)]
vp10/encoder/bitstream.[hc]: correct a prototype
quiets a -Wmissing-prototypes warning
BUG=b/
29584271
Change-Id: I91aba2a75dccd6752bdf91837564c2aa45817c09
James Zern [Sat, 2 Jul 2016 00:38:47 +0000 (17:38 -0700)]
vp10/common/idct.h: add some missing prototypes
quiets the warning of the same name
BUG=b/
29584271
Change-Id: I220cd58e1060f77e3910472fed1b167add3a08f8
James Zern [Sat, 2 Jul 2016 00:45:46 +0000 (17:45 -0700)]
highbd_quantize_intrin_sse2.c: add missing rtcd include
quiets -Wmissing-prototypes warnings
BUG=b/
29584271
Change-Id: Iff5214df0d1781810afbfc20bfaf664f109e2f29
James Zern [Sat, 2 Jul 2016 00:41:26 +0000 (17:41 -0700)]
vp10: add some missing includes
quiets some -Wmissing-prototypes warnings
BUG=b/
29584271
Change-Id: I9174728459fcabb6d9ac0028ae58029e52c0da92
Yue Chen [Mon, 11 Jul 2016 21:23:08 +0000 (14:23 -0700)]
Cosmetics for vp10/common/vp10_rtcd_defs.pl
Change-Id: Iaf8c6f0b1e340f0406df2871a3dc2ded19b7009a
Debargha Mukherjee [Mon, 11 Jul 2016 22:30:33 +0000 (22:30 +0000)]
Merge "Add a few branch hints to vp10_optimize_b." into nextgenv2
Debargha Mukherjee [Mon, 11 Jul 2016 22:30:16 +0000 (22:30 +0000)]
Merge "Optimize and cleanup supertx predictor." into nextgenv2
Debargha Mukherjee [Mon, 11 Jul 2016 19:30:03 +0000 (19:30 +0000)]
Merge "Improve vpx_blend_* functions." into nextgenv2
Geza Lore [Fri, 8 Jul 2016 13:10:53 +0000 (14:10 +0100)]
Optimize and cleanup supertx predictor.
Use vpx_blend_a64_hmask and vpx_blend_a64_vmask to speed up
computing the supertx predictor.
Decoder speedup of up to 4% has been observed.
Change-Id: I255a5ba4cc24f78dc905d25b6e2f7fbafac13253
Geza Lore [Mon, 11 Jul 2016 11:43:47 +0000 (12:43 +0100)]
Improve vpx_blend_* functions.
- Made source buffers pointers to const.
- Renamed vpx_blend_mask6b to vpx_blend_a64_mask. This is more
indicative that the function does alpha blending. The 6, or 6b
suffix was misleading, as the max mask value (64) does not fit into
6 bits.
- Added VPX_BLEND_* macros to use when needing to blend scalars.
- Use VPX_BLEND_A256 in combine_interintra to be more explicit about
the operation being done.
- Added versions of vpx_blend_a64_* which take 1D horizontal/vertical
masks directly and apply them to all rows/columns
(vpx_blend_a64_hmask and vpx_blend_a64_vmask). The SSE4.1 optimzied
horizontal version now falls back on the 2D version. This can be
improved upon if it show up high enough in a profile.
- All vpx_blend_a64_* functions now support block sizes down to 1x1
(ie: a single pixel). This is for usage convenience. The SSE4.1
optimized versions fall back on the C implementation if
w <= 2 or h <= 2. This can again be improved if it becomes hot code.
Change-Id: I13ab3835146ffafe3e1d74d8e9cf64a5abe4144d
Pascal Massimino [Mon, 11 Jul 2016 13:17:17 +0000 (06:17 -0700)]
remove ROUNDZ_* macros in favor of just ROUND_* ones
Change-Id: I263088be8d71018deb9cc6a9d2c66307770b824d
Geza Lore [Mon, 11 Jul 2016 11:29:16 +0000 (11:29 +0000)]
Merge "Fix unused warning without ext-interp" into nextgenv2
Debargha Mukherjee [Sat, 9 Jul 2016 06:50:32 +0000 (06:50 +0000)]
Merge "Refactor and clean up on blend_mask6" into nextgenv2
Yue Chen [Sat, 9 Jul 2016 01:07:26 +0000 (01:07 +0000)]
Merge "Fix assertion failures in mips+msa setting" into nextgenv2
Yue Chen [Fri, 8 Jul 2016 23:47:19 +0000 (16:47 -0700)]
Fix assertion failures in mips+msa setting
Directly call c functions, otherwise when EXT_TX is enabled, hybrid
transform other than combination of DCT/ADST has not been implemented, thus
will cause assertion failures in the switch loops in vp10_fhtnxn_msa() and
vp10_ihtnxn_nxn_add_msa().
BUG=webm:1239
Change-Id: I2379a07e5406f9489edcd2f3205682f679c9b091
Jingning Han [Fri, 8 Jul 2016 22:21:27 +0000 (22:21 +0000)]
Merge "Properly reset rate and distortion value for zero pred residual case" into nextgenv2
Debargha Mukherjee [Fri, 8 Jul 2016 20:27:31 +0000 (13:27 -0700)]
Refactor and clean up on blend_mask6
Change-Id: Ie9188471e7dc07ab9c95b22f258b1662e895c533
Jingning Han [Fri, 8 Jul 2016 18:25:13 +0000 (18:25 +0000)]
Merge "Integrate ext-interp into dual filter framework" into nextgenv2
Geza Lore [Fri, 8 Jul 2016 18:16:05 +0000 (19:16 +0100)]
Add a few branch hints to vp10_optimize_b.
vp10_optimize_b now takes between 40% to 60% of the TOTAL runtime
of the encoder, depending on bit-rate. It also contains 2/3 to 3/4
of the mispredicted branch instructions in the whole program.
Adding a few branch hints makes vp10_optimize_b around 2-5% faster
(dependig on bit-rate) when compiled with gcc/clang.
Change-Id: I1572733e18b4166bc10591b958c5018a9561fa2b
Sarah Parker [Fri, 8 Jul 2016 17:40:54 +0000 (17:40 +0000)]
Merge "Make new_quant bin widths to be uniform" into nextgenv2
Jingning Han [Tue, 10 May 2016 16:56:20 +0000 (09:56 -0700)]
Integrate ext-interp into dual filter framework
The combination of the two experiments improves the compression
performance gains:
lowres 2.5%
midres 2.1%
Change-Id: Id26c0a9474ce08893aa1d946365c7ff850fab57a
Jingning Han [Fri, 8 Jul 2016 05:41:56 +0000 (22:41 -0700)]
Properly reset rate and distortion value for zero pred residual case
When the prediction residuals are all zero, reset the coeff rate
cost and the distortion value to be zero. This change doesn't affect
lowres set significantly, but improves several clips in the midres
set, like sintel_480p and mobisode2_480p, by a few percents. The
average performance for midres set is improved by 0.2%.
Change-Id: Idd5ebf2652e556a1b1c569fe3c48dacef3f11c32
Geza Lore [Fri, 8 Jul 2016 14:48:02 +0000 (15:48 +0100)]
Fix unused warning without ext-interp
Change-Id: Ibb63c492eb8278d115262b8fc3cbc761c406b107
Jingning Han [Fri, 8 Jul 2016 01:11:17 +0000 (01:11 +0000)]
Merge "Fix ioc in trellis optimization with hbd" into nextgenv2
Sarah Parker [Thu, 7 Jul 2016 23:22:32 +0000 (16:22 -0700)]
Make new_quant bin widths to be uniform
Change-Id: Iceeca8ecbc43919b43189352a307479d666d1dad
Debargha Mukherjee [Thu, 7 Jul 2016 22:47:57 +0000 (22:47 +0000)]
Merge "RD costing fix in loop-restoration expt" into nextgenv2
Debargha Mukherjee [Thu, 7 Jul 2016 21:55:38 +0000 (21:55 +0000)]
Merge "Remove redundant code in new_quant" into nextgenv2
Debargha Mukherjee [Thu, 7 Jul 2016 20:05:12 +0000 (20:05 +0000)]
Merge "Clean up build_wedge_inter_predictor_from_buf" into nextgenv2
Debargha Mukherjee [Thu, 7 Jul 2016 19:54:54 +0000 (12:54 -0700)]
RD costing fix in loop-restoration expt
Change-Id: I8dbc1002f5d6bf8f2409db8c6be4346f1df0590c
Jingning Han [Thu, 7 Jul 2016 18:59:22 +0000 (11:59 -0700)]
Fix ioc in trellis optimization with hbd
Use int64_t type for distortion. This avoids integer overflow
issues in the trellis optimization function in high bit-depth
settings.
Change-Id: I550c3ca9f11a3191ef8638a152887018cd476141
Debargha Mukherjee [Thu, 7 Jul 2016 18:55:20 +0000 (11:55 -0700)]
Remove redundant code in new_quant
Change-Id: Ie2534c7c0cc3fc59e7389b55cb066f2b347d846e
Geza Lore [Wed, 6 Jul 2016 14:54:29 +0000 (15:54 +0100)]
Remove various testing utilities.
test/assertion_helpers.h
test/randomise.{cc,h}
test/snapshot.h
Modfiy blend_mask6_test.cc not to rely on these.
Change-Id: I88b8933fe0a729a606797e5cd421795a544c612d
Geza Lore [Thu, 7 Jul 2016 12:12:24 +0000 (13:12 +0100)]
Clean up build_wedge_inter_predictor_from_buf
Change-Id: I715f8ffa3e81056a74ca8ac94793009afb781221
Debargha Mukherjee [Thu, 7 Jul 2016 05:55:06 +0000 (05:55 +0000)]
Merge "Reinstate tests for wedge partition selection optimizations." into nextgenv2
Debargha Mukherjee [Thu, 7 Jul 2016 05:54:45 +0000 (05:54 +0000)]
Merge "Add tests for vpx_sum_squares_i16." into nextgenv2
Yue Chen [Thu, 7 Jul 2016 01:12:19 +0000 (01:12 +0000)]
Merge "Add SSE4.1 vpx_obmc_sad* implementations." into nextgenv2
James Zern [Wed, 6 Jul 2016 23:25:34 +0000 (23:25 +0000)]
Merge changes I9433d858,Iafd05637,If08ce6ca into nextgenv2
* changes:
tests: remove redundant round() definition
remove visual studio < 2010 workarounds
configure: remove old visual studio support (<2010)
Yue Chen [Wed, 6 Jul 2016 22:44:27 +0000 (22:44 +0000)]
Merge "Refactoring in preparation for OBMC optimizations." into nextgenv2
Geza Lore [Tue, 5 Jul 2016 12:36:25 +0000 (13:36 +0100)]
Add SSE4.1 vpx_obmc_sad* implementations.
Speedup for these functions: 4x
Change-Id: I21baa04f53c6ab308ea3edf3ebacc62970e97454
hui su [Wed, 29 Jun 2016 22:16:31 +0000 (15:16 -0700)]
mcomp.c: rename variables and remove unnecessary codes
Change-Id: I4ad4061875fa1c8f3801efbcdcb0da47e7c032a5
Geza Lore [Tue, 21 Jun 2016 19:23:13 +0000 (20:23 +0100)]
Add tests for vpx_sum_squares_i16.
Change-Id: I529c34d5bfa85719cb6499a9a3c9d907eccccd56
Geza Lore [Tue, 21 Jun 2016 19:29:21 +0000 (20:29 +0100)]
Reinstate tests for wedge partition selection optimizations.
This reinstates the tests from commit
efda2831e5f758b4f350679b5c55c0b9282449b0 with the appropriate
fixes for 32 bit x86 builds.
Change-Id: Ib331906c5b448ca964895ee9cbfd4266f67d1089
Geza Lore [Mon, 4 Jul 2016 10:47:19 +0000 (11:47 +0100)]
Refactoring in preparation for OBMC optimizations.
- Use int32_t instead of int in vpx_obmc{variance,sad} functions
- Remove weigthed_src and obmc mask strides and assume contiguous
buffers. These inputs can always be packed as contiguous arrays.
Change-Id: I74c09b3fb3337f13d39e13a9cb61e140536f345d
Wei-ting Lin [Thu, 30 Jun 2016 20:33:55 +0000 (13:33 -0700)]
Remove reference frame buffer update for show_exsiting_frame
Originally we need to send the refresh flag and the virtual indices
mapping for the reference frame buffer update for show_existing_frame to
have the BWDREF_FRAME replace the LAST_FRAME.
To remove sending this information, we update the the virtual indices
of the reference frame buffer after the last_bipred_frame is encoded,
and therefore the decoder will receive the updated reference mapping
at the next non-show-existing frame.
As a result, we can save 4 bytes per show-existing frame, and get 0.12,
0.2, and 0.07 BDRATE improvement in lowres, derf, and midref test set
respectively.
Change-Id: I63d41ee6ea99884798f0778b789d2701e2f2d3e0
Geza Lore [Fri, 1 Jul 2016 08:00:05 +0000 (09:00 +0100)]
Minor cleanup of inter mode search.
Change-Id: I523a3b30eb80fc6c6ed83638fdb82cf65c22b2e5
Debargha Mukherjee [Thu, 30 Jun 2016 18:18:53 +0000 (18:18 +0000)]
Merge "Reject ext-inter compound modes based on modelled RD." into nextgenv2
Jingning Han [Thu, 30 Jun 2016 16:26:24 +0000 (16:26 +0000)]
Merge "Remove unused BITDEPTH_10 definition" into nextgenv2
Jingning Han [Thu, 30 Jun 2016 16:26:18 +0000 (16:26 +0000)]
Merge "Fix shift value in dist_block with hbd" into nextgenv2
Debargha Mukherjee [Thu, 30 Jun 2016 16:03:49 +0000 (16:03 +0000)]
Merge "Various cosmetics on the new_quant experiment" into nextgenv2
Geza Lore [Tue, 28 Jun 2016 11:59:19 +0000 (12:59 +0100)]
Reject ext-inter compound modes based on modelled RD.
Reject ext-inter compound modes before doing full rate distortion
evaluation, if the corresponding single reference modes had a lower
modelled RD.
ext-inter speedup up to TBD.
Coding performance: TBD
Change-Id: I358bfb879c5ebe5e7afbf6f540cc784f8de14857
James Zern [Tue, 28 Jun 2016 03:43:04 +0000 (20:43 -0700)]
tests: remove redundant round() definition
use vpx_ports/msvc.h for compatibility
BUG=b/
29583530
Change-Id: I9433d8586cd0b790e7f4d697304298feafe801f1
(cherry picked from commit
0a64929f19cc1ce89f993aa5c9d61a29679eb961 )
James Zern [Tue, 28 Jun 2016 03:29:01 +0000 (20:29 -0700)]
remove visual studio < 2010 workarounds
BUG=b/
29583530
Change-Id: Iafd05637eb65f4da54a9c857e79204a77646858a
(cherry picked from commit
c125f4a594815ad63b50e4b684ada4b44c545932 )
James Zern [Tue, 28 Jun 2016 03:08:12 +0000 (20:08 -0700)]
configure: remove old visual studio support (<2010)
BUG=b/
29583530
Change-Id: If08ce6ca352f377ac4db6b9b1909b507bba6d872
(cherry picked from commit
078dff72ca7bff079cb3c56d98c588c6ea6d2814 )
(cherry picked from commit
046226376533b610ddc700f14409f195aa6abd51 )
Jingning Han [Wed, 29 Jun 2016 23:43:22 +0000 (16:43 -0700)]
Remove unused BITDEPTH_10 definition
Change-Id: Ic11f32db352e1ff7b3ed140654ee1a6016ba516f
Jingning Han [Wed, 29 Jun 2016 23:40:47 +0000 (16:40 -0700)]
Fix shift value in dist_block with hbd
This offset value related to the bit depth has been taken care of
inside the function vp10_highbd_block_error.
Change-Id: I58dd8a53380ba4529d59837e56a951bc81a2962e
Debargha Mukherjee [Wed, 29 Jun 2016 18:59:27 +0000 (11:59 -0700)]
Remove use_quant_fp speed feature
Change-Id: I22f1299545d4c75d80e72d479be66f66ea142ef1
Debargha Mukherjee [Wed, 29 Jun 2016 20:18:52 +0000 (13:18 -0700)]
Various cosmetics on the new_quant experiment
Also extends quant profiles to include quality range.
Change-Id: Ia96e45b6425e1d42ca61fc401f63d4fd7214e448
Debargha Mukherjee [Wed, 29 Jun 2016 17:52:39 +0000 (17:52 +0000)]
Merge "Remove skip_txfm optimization." into nextgenv2
Sarah Parker [Wed, 29 Jun 2016 03:29:12 +0000 (03:29 +0000)]
Merge "Fix compiler warnings in yv12extend.c" into nextgenv2
Sarah Parker [Wed, 29 Jun 2016 02:03:10 +0000 (02:03 +0000)]
Merge "Fix compiler warnings in vp10_convolve_optimz_test.cc" into nextgenv2
Sarah Parker [Wed, 29 Jun 2016 00:23:16 +0000 (17:23 -0700)]
Fix compiler warnings in yv12extend.c
Change-Id: I1f6f5b8861c4081b1f4b85c531c5d7ef0cb67bab
Sarah Parker [Wed, 29 Jun 2016 00:13:03 +0000 (17:13 -0700)]
Fix compiler warnings in vp10_convolve_optimz_test.cc
Change-Id: I11b717e1652dff440a54f6977527d544b0c5ed29
Alex Converse [Tue, 28 Jun 2016 21:21:13 +0000 (14:21 -0700)]
ethread_test: Remove vp10 as test parameter.
Change-Id: I043418cde5a2562520ff37cdf81436abc2c9821a
Geza Lore [Mon, 27 Jun 2016 12:54:30 +0000 (13:54 +0100)]
Remove skip_txfm optimization.
Commit
0d6980d7a1caa592058f8d5d618b012c160772f7 removed some use
of the skip_txfm optimization, and the rest are not productive.
The current use of this optimization is only used with --good
and --cpu-used >= 3, however the overhead of this is higher than the
speedup it yields.
Removing this, and subsequently simplifying model_rd_for_sb yields
a net encoder speedup:
--cpu-used=0 ~1.5% faster
--cpu-used=3 ~2.0% faster
The code simplification is also significant.
Change-Id: I1dd668c32de15a2e912c59c42379d0f9e1032ff8
Sarah Parker [Tue, 28 Jun 2016 02:21:35 +0000 (02:21 +0000)]
Merge "Quantization fix for new-quant/var-tx" into nextgenv2
Hui Su [Tue, 28 Jun 2016 00:37:48 +0000 (00:37 +0000)]
Merge "Rename the initial MV search candidate" into nextgenv2
Hui Su [Tue, 28 Jun 2016 00:24:01 +0000 (00:24 +0000)]
Merge "Refactor vp10_pattern_search" into nextgenv2
Debargha Mukherjee [Thu, 23 Jun 2016 21:49:00 +0000 (14:49 -0700)]
Quantization fix for new-quant/var-tx
Also use the fp quantizer consistently
lowres: -0.07 BDRATE improvement
Change-Id: I9174f6ad54a74d38541004b99cb3689d0c09be55
Hui Su [Mon, 27 Jun 2016 23:09:10 +0000 (23:09 +0000)]
Merge "Fix a bug in vp10_pattern_search()" into nextgenv2
Jingning Han [Mon, 27 Jun 2016 21:34:42 +0000 (21:34 +0000)]
Merge "Disable trellis optimized quantization in the first-pass" into nextgenv2
Yi Luo [Mon, 27 Jun 2016 21:33:44 +0000 (21:33 +0000)]
Merge "Fix bugs in convolution filter optimization" into nextgenv2
Jingning Han [Mon, 27 Jun 2016 17:15:36 +0000 (10:15 -0700)]
Disable trellis optimized quantization in the first-pass
This resolves the use of uninitialized value in the first-pass
encoding.
Change-Id: I78bc19214a1bfde5c5641424550cbbe4e52cae99
Sarah Parker [Mon, 27 Jun 2016 18:46:25 +0000 (18:46 +0000)]
Merge "Add multiple quantization profiles to new_quant experiment" into nextgenv2
Yi Luo [Sat, 25 Jun 2016 00:29:21 +0000 (17:29 -0700)]
Fix bugs in convolution filter optimization
- Fix the over-writing bug in horizontal filtering as width = 2.
- Fix 10-tap vertical filtering which no longer reads one row of
pixel above the block.
- Fix 10-tap filter zero padding.
- Encoder speed slow down ~4.0%, compared to,
81ad953 Convolution vertical filter SSSE3 optimization
Change-Id: I9bb294a4529300081c29bf284e6bc6eb081cc536
hui su [Fri, 24 Jun 2016 19:59:01 +0000 (12:59 -0700)]
Rename the initial MV search candidate
Its old name "ref_mv" is confusing.
Change-Id: I7ac8f346c468bcf3c0e7692582d423fb7a1f113a
hui su [Fri, 24 Jun 2016 18:06:50 +0000 (11:06 -0700)]
Refactor vp10_pattern_search
Combine it with vp10_pattern_search_sad
Change-Id: I47a3b34dfefad9fc8abd23fcc197f6aea3419873
hui su [Fri, 24 Jun 2016 18:47:56 +0000 (11:47 -0700)]
Fix a bug in vp10_pattern_search()
Should use sub-pel MV instead of full-pixel MV as input parameter
to calc_int_cost_list().
Change-Id: I054d94220a090ca54c8d24df265193ee345cd439
Debargha Mukherjee [Sat, 25 Jun 2016 00:32:21 +0000 (00:32 +0000)]
Merge "Turn on ActiveMapRefreshTest for Vp10" into nextgenv2
Sarah Parker [Fri, 10 Jun 2016 22:29:10 +0000 (15:29 -0700)]
Add multiple quantization profiles to new_quant experiment
Add the ability to pick between 3 quantization profiles.
The profile is chosen based on the entropy context at the
block level.
Change-Id: Iaea0485798441b7d635962c2563f3a477f582dac
Debargha Mukherjee [Fri, 24 Jun 2016 17:31:07 +0000 (10:31 -0700)]
Turn on ActiveMapRefreshTest for Vp10
Also reduce number of frames coded for VP10.
Change-Id: I7de908861620b6f4f08513516110fd584660d994
Yi Luo [Fri, 24 Jun 2016 18:47:21 +0000 (18:47 +0000)]
Merge "Change register loading to fix stack overflow issue" into nextgenv2