]>
granicus.if.org Git - libvpx/log
Johann [Fri, 30 Sep 2016 00:57:51 +0000 (17:57 -0700)]
Refine vp8_refining_search_sadx4 targeting
This uses the same sdx4df pointers as vp8_diamond_search_sadx4 and
should therefore target the same optimizations.
See
e4ddf9db6a37eee59c079f5ae427643ae3424fcf
Change-Id: Ic298e9b25c34bbe6b7a0799509355b0addb56675
Johann [Tue, 1 Nov 2016 18:02:09 +0000 (11:02 -0700)]
Optimize idct32x32_135_add for NEON
BUG=webm:1295
Change-Id: I7f80ef4d29813fcb401fc6075babf19e3c195462
Linfeng Zhang [Tue, 8 Nov 2016 16:53:41 +0000 (16:53 +0000)]
Merge "Add high bitdepth intra prediction NEON optimization (mode dc)"
James Zern [Tue, 8 Nov 2016 01:05:16 +0000 (01:05 +0000)]
Merge "partial_idct_test: set MinSupportedCoeff for NEON"
Johann Koenig [Tue, 8 Nov 2016 00:44:16 +0000 (00:44 +0000)]
Merge "ads2gas: remove RN stanza"
Johann [Tue, 8 Nov 2016 00:19:15 +0000 (16:19 -0800)]
ads2gas: remove RN stanza
The matching on ads2gas_apple.pl is too liberal and catches
CONFIG_EXTERNAL_BUILD and CONFIG_INTERNAL_STATS because they have RN in
the names.
The RN renaming feature is not used in any existing assembly files. It
was used in some armv6 files but they were removed.
Change-Id: Ib65abf1947d3e89f0d1584e2a5de399d24008f95
James Zern [Mon, 7 Nov 2016 23:27:02 +0000 (15:27 -0800)]
partial_idct_test: set MinSupportedCoeff for NEON
vpx_idct4x4_16_add_neon fails with INT16_MIN, +1 is all right
BUG=webm:1335
Change-Id: I25830c8ab0782822fc3c9db6cc669c2e65f2700e
Linfeng Zhang [Fri, 4 Nov 2016 16:44:22 +0000 (09:44 -0700)]
Rename vpx_highbd_idct8x8_10{*}() to vpx_highbd_idct8x8_12{*}()
Also update its trigger threshold from 10 to 12.
Change-Id: Ib8dddd87a5a22a12ca66e7084d342fbb027b0a2f
Linfeng Zhang [Mon, 7 Nov 2016 16:55:00 +0000 (16:55 +0000)]
Merge "Replace highbd_dct_const_round_shift with dct_const_round_shift"
Johann Koenig [Sun, 6 Nov 2016 00:13:05 +0000 (00:13 +0000)]
Merge "idct test: use coeff consistently"
Johann Koenig [Sun, 6 Nov 2016 00:12:59 +0000 (00:12 +0000)]
Merge "partial_idct_test: Add large coefficient test"
Johann Koenig [Sun, 6 Nov 2016 00:12:12 +0000 (00:12 +0000)]
Merge "Update vp9_fdct8x8_quant_ssse3 for highbitdepth"
James Zern [Sat, 5 Nov 2016 02:47:11 +0000 (02:47 +0000)]
Merge "vp9-svc: Add unittest for svc-decoding."
Johann [Sat, 5 Nov 2016 01:41:56 +0000 (18:41 -0700)]
idct test: use coeff consistently
Change-Id: I913a13066993a3315a0ff8310b3cad1572d4cdd7
Johann [Tue, 1 Nov 2016 19:09:46 +0000 (12:09 -0700)]
partial_idct_test: Add large coefficient test
Two functions do not pass this test:
vpx_idct8x8_64_add_ssse3
vpx_idct8x8_12_add_ssse3
The test has been modified to avoid triggering an issue with those
functions but they still must be investigated.
BUG=webm:1332
Change-Id: I52569a81e8e6e0b33c4a4d060d0b69c3fc4f578e
Marco [Thu, 3 Nov 2016 19:50:44 +0000 (12:50 -0700)]
vp9-svc: Add unittest for svc-decoding.
To test the VP9_DECODE_SVC_SPATIAL_LAYER decoder control
introduced in
86b0042 .
Change-Id: I3d164a41d7bbab14c0aee80fd890870704a18f6e
Johann [Thu, 3 Nov 2016 17:47:06 +0000 (10:47 -0700)]
Update vp9_fdct8x8_quant_ssse3 for highbitdepth
Borrow transition functions from fdct.h nee vpx_quantize_b_sse2
BUG=webm:1304
Change-Id: I9c88c3eec3ff8bb461411d98c26c3c236ea28ef1
Linfeng Zhang [Wed, 2 Nov 2016 22:20:30 +0000 (15:20 -0700)]
Replace highbd_dct_const_round_shift with dct_const_round_shift
They are identical.
Change-Id: I1ccaf03c81c3cbf88e82d77ffeb8204f5b063c61
Linfeng Zhang [Fri, 4 Nov 2016 22:40:03 +0000 (22:40 +0000)]
Merge "Cosmetics of inv_txfm.c"
Johann Koenig [Fri, 4 Nov 2016 21:03:17 +0000 (21:03 +0000)]
Merge "Extract high bit depth helper functions"
Linfeng Zhang [Wed, 2 Nov 2016 21:22:00 +0000 (14:22 -0700)]
Cosmetics of inv_txfm.c
Unify code of 8-bit and high bitdepth.
Change-Id: I3fe441577af0249030ca3a1ef769eb9030711434
Johann [Thu, 3 Nov 2016 22:21:32 +0000 (15:21 -0700)]
Extract high bit depth helper functions
These can be used in the vp9 fdct as well.
Change-Id: I4f3875e0cba1b8cad209c3a0581e121deba7675e
James Zern [Fri, 4 Nov 2016 17:48:24 +0000 (17:48 +0000)]
Merge "configure: disable tools for armv7-win32-vs1[24]"
Martin Storsjo [Fri, 4 Nov 2016 10:20:07 +0000 (12:20 +0200)]
Add a missing END directive in idct_neon.asm
This fixes building with MS armasm.
Change-Id: I2629eeed859b775ca667a65ba109f8d1bf7b0e03
Martin Storsjo [Fri, 4 Nov 2016 10:18:42 +0000 (12:18 +0200)]
Fix producing vcxproj files with *.S arm assembly files
These cases were leftover in
1ddb4c03622a654452a5e76b5a0e4db715dc350d .
Change-Id: Ie058fb6c78580e60205c47a1d314bd66e794cde4
James Zern [Fri, 4 Nov 2016 05:54:35 +0000 (22:54 -0700)]
configure: disable tools for armv7-win32-vs1[24]
this shares the same prohibition as the examples
Change-Id: I17d65e4f26847af8cbb1d1a3c4a114ed021a8b9f
Marco Paniconi [Thu, 3 Nov 2016 23:12:05 +0000 (23:12 +0000)]
Merge "vp9: Non-rd pickmode: fix logic in reference masking."
Marco [Fri, 28 Oct 2016 23:26:30 +0000 (16:26 -0700)]
vp9-svc: Add decoder control to decode up to x spatial layers.
Change-Id: I85536473b8722424785c84c5b5520960b4e5744a
Marco [Thu, 3 Nov 2016 17:21:12 +0000 (10:21 -0700)]
vp9: Non-rd pickmode: fix logic in reference masking.
Add condition that usable_ref_frame > LAST.
This is to avoid potentially skipping all last-nonzero mv modes,
if golden is used as a reference but skipped completely for the
current block.
This has no effect currenty, as we always consider testing golden
mode for each block.
Change-Id: I3182cf44664081935a90ed43aa7b32e710e60e22
Debargha Mukherjee [Thu, 3 Nov 2016 17:04:17 +0000 (17:04 +0000)]
Merge "Speed-up recode loop for extreme bitrate diffs"
Jerome Jiang [Thu, 3 Nov 2016 04:31:35 +0000 (04:31 +0000)]
Merge "pp_filter_test.cc,cosmetics:adjust name convention"
Jerome Jiang [Wed, 2 Nov 2016 16:28:58 +0000 (09:28 -0700)]
pp_filter_test.cc,cosmetics:adjust name convention
Change-Id: I81b6fc9b83f0febbb12975aef92768bbd273fd61
Johann Koenig [Wed, 2 Nov 2016 18:09:45 +0000 (18:09 +0000)]
Merge "arm idct: move to-be-shared code to header"
Paul Wilkins [Wed, 2 Nov 2016 17:10:07 +0000 (17:10 +0000)]
Merge "Fixed bug in formatting of debug stats."
paulwilkins [Wed, 2 Nov 2016 09:38:18 +0000 (09:38 +0000)]
Fixed bug in formatting of debug stats.
Fixed formatting bug introduced by the fix to BUG=webm:1322
( Iedc4477aef1746aa0a4f84d88a1156296fd3ba87)
Change-Id: I715ee446c0e8584967ab87ba4e355759dd394187
James Zern [Wed, 2 Nov 2016 01:45:50 +0000 (18:45 -0700)]
vp9,tile_worker_hook: correctly set jmp target
vp9_init_macroblockd() resets the error_info to cm's global copy; this
needs to be set to the thread-level target to avoid jumping to the
incorrect stack, resulting in hang or crash.
broken since:
1f4a6c8 vp9/tile_worker_hook: add multiple tile decoding
includes v1.5.0, v1.6.0
BUG=629481
Change-Id: Icbf1696b25ba8c479e845fbf227b3c3ca73542f5
Linfeng Zhang [Tue, 25 Oct 2016 21:48:32 +0000 (14:48 -0700)]
Add high bitdepth intra prediction NEON optimization (mode dc)
BUG=webm:1316
Change-Id: I984d6004ea2445e86f213fb6fa4d794a9955af8f
Johann [Tue, 1 Nov 2016 18:01:40 +0000 (11:01 -0700)]
arm idct: move to-be-shared code to header
Change-Id: I67458cd358b4dc4434bbdbfcdd571769561b619e
James Zern [Tue, 1 Nov 2016 22:34:59 +0000 (22:34 +0000)]
Merge "idct32x32_1_add_neon: clear a couple conv warnings"
James Zern [Tue, 1 Nov 2016 21:43:13 +0000 (21:43 +0000)]
Merge changes I08af3a54,If5959a25,I6763e62e
* changes:
build/make/Android.mk: s/armv8/arm64/
build/make/Android.mk: fix armeabi-v7a build
use .S suffix rather than .s for NEON asm
Linfeng Zhang [Tue, 25 Oct 2016 21:24:16 +0000 (14:24 -0700)]
Add high bitdepth intra prediction optimization speed test
BUG=webm:1316
Change-Id: I99feec867d5b8ea06b43cdd3fcd7c90238f5efdb
Linfeng Zhang [Tue, 1 Nov 2016 19:38:06 +0000 (19:38 +0000)]
Merge "Refine 8-bit intra prediction NEON optimization (mode tm)"
Linfeng Zhang [Fri, 21 Oct 2016 19:38:49 +0000 (12:38 -0700)]
Refine 8-bit intra prediction NEON optimization (mode tm)
Change-Id: I98b9577ec51367df5e5d564bedf7c3ea0606de4c
Paul Wilkins [Tue, 1 Nov 2016 09:29:30 +0000 (09:29 +0000)]
Merge "Change to KF boost calculation."
James Zern [Tue, 1 Nov 2016 01:56:34 +0000 (18:56 -0700)]
idct32x32_1_add_neon: clear a couple conv warnings
int16_t -> uint8_t
Change-Id: I3c5e0985bc3584dce289c35b5973de24cdc73b76
James Zern [Tue, 1 Nov 2016 01:35:23 +0000 (18:35 -0700)]
build/make/Android.mk: s/armv8/arm64/
the configure target is arm64-android-gcc which generates .mk files of
the same form
Change-Id: I08af3a54ef203b1496d185a0f8c8fe702881a173
James Zern [Tue, 1 Nov 2016 01:34:16 +0000 (18:34 -0700)]
build/make/Android.mk: fix armeabi-v7a build
vpx_config.asm and idct_neon.asm.S are required since:
3ae2597 idct,NEON: add a tran_low_t->s16 load adapter
Change-Id: If5959a25edb370dd7dcdca71c96e9a5aad0840ce
James Zern [Mon, 31 Oct 2016 23:18:51 +0000 (16:18 -0700)]
use .S suffix rather than .s for NEON asm
for compatibility with other build systems
Change-Id: I6763e62e3126850ad4f8ad29e388b8dad0bbc4c3
Marco Paniconi [Mon, 31 Oct 2016 23:16:35 +0000 (23:16 +0000)]
Merge "vp9-svc: Fix some stats in vp9_spatial_svc_encoder."
Marco [Mon, 31 Oct 2016 22:23:23 +0000 (15:23 -0700)]
vp9-svc: Fix some stats in vp9_spatial_svc_encoder.
Correction to rate control stats output under -rcstat.
Change-Id: I46fa5d2a66ed657121ee3d685608a148bb9a7bb3
James Zern [Mon, 31 Oct 2016 21:59:11 +0000 (21:59 +0000)]
Merge "idct,NEON: add a tran_low_t->s16 load adapter"
Peter Boström [Mon, 31 Oct 2016 20:01:11 +0000 (20:01 +0000)]
Merge "Add vp9_spatial_svc_encoder to .gitignore."
Linfeng Zhang [Mon, 31 Oct 2016 19:57:22 +0000 (19:57 +0000)]
Merge "Refine 8-bit intra prediction NEON optimization (mode h and v)"
Peter Boström [Mon, 31 Oct 2016 18:27:02 +0000 (14:27 -0400)]
Add vp9_spatial_svc_encoder to .gitignore.
Change-Id: I3c90d657cca533264dd62bb7749c53a862d0352a
Marco Paniconi [Mon, 31 Oct 2016 18:23:13 +0000 (18:23 +0000)]
Merge "vp9-svc: Add checks to layer bitrates in vp9_spatial_svc_encoder."
James Zern [Tue, 18 Oct 2016 19:30:43 +0000 (12:30 -0700)]
idct,NEON: add a tran_low_t->s16 load adapter
enable idct4x4* and idct8x8* which are compatible for 8-bit decodes in
high-bitdepth mode. the adapter narrows 32-bit input to 16, whether the
expansion can be avoided at all in this case remains a TODO. roughly
matches sse2.
BUG=webm:1294
Change-Id: I3ea94e5a2070dfd509b5de0c555aab4e1f4da036
Linfeng Zhang [Fri, 21 Oct 2016 18:44:49 +0000 (11:44 -0700)]
Refine 8-bit intra prediction NEON optimization (mode h and v)
Change-Id: I45e1454c3a85e081bfa14386e0248f57e2a91854
Marco [Mon, 31 Oct 2016 05:20:04 +0000 (22:20 -0700)]
vp9-svc: Add checks to layer bitrates in vp9_spatial_svc_encoder.
Add some checks to the layer bitrates passed in through the command line.
Change-Id: I16f270035a6034d63313fe3019aa90dca9a3eefb
James Bankoski [Mon, 31 Oct 2016 13:58:14 +0000 (13:58 +0000)]
Merge "vpxdec.c : don't double count corrupted frames"
Jim Bankoski [Fri, 28 Oct 2016 12:53:26 +0000 (05:53 -0700)]
vpxdec.c : don't double count corrupted frames
A past patch made it so that every frame that had a decode error
caused a corrupted frame to be counted. Unfortunately it was possible
to get both a decode error and a corrupt frame for the same frame
and thus double count an error. This code makes that impossible.
Change-Id: Iea973727422a3bf093ffda72fa358a285736048b
James Zern [Sat, 29 Oct 2016 20:02:50 +0000 (13:02 -0700)]
tiny_ssim: fix visual studio build
s/inttypes.h/vpx_integer.h/
clear a uint64_t -> double conversion warning
Change-Id: I58d108b083787a754152eb79ef6df61c2c5f95b1
Peter Boström [Sat, 29 Oct 2016 01:25:01 +0000 (01:25 +0000)]
Merge "Add temporal-layer support to tiny_ssim."
Johann Koenig [Sat, 29 Oct 2016 00:34:16 +0000 (00:34 +0000)]
Merge "partial_idct_test: add _add_ test"
Peter Boström [Fri, 28 Oct 2016 18:50:20 +0000 (14:50 -0400)]
Add temporal-layer support to tiny_ssim.
Permits skipping 0, 1/2 or 3/4 of the frames, corresponding to
temporal layers 2, 1 and 0 of a 3-temporal-layer encoding. 1/2
corresponds to TL0 in a 2-layer encoding.
Change-Id: I7f6d131f63707e5262fc67d111bfb3a751ede90d
Marco Paniconi [Fri, 28 Oct 2016 18:32:40 +0000 (18:32 +0000)]
Merge "vp9: Updates to SVC sample encoder."
Marco [Thu, 27 Oct 2016 22:38:02 +0000 (15:38 -0700)]
vp9: Updates to SVC sample encoder.
Allow for passing in the layer bitrates at command line.
Fix to allow passing in bitrate for each spatial-temporal layer.
Change to some default values for 1 pass cbr mode:
spatial scale and qp-max/min.
Small fixes to some build warnings.
Change-Id: I3f9a776262712480a6570bb863a835b2fc49935a
Yaowu Xu [Fri, 28 Oct 2016 17:21:47 +0000 (17:21 +0000)]
Merge "Add tools/tiny_ssim for generating SSIM/PSNR."
Peter Boström [Thu, 27 Oct 2016 23:07:44 +0000 (19:07 -0400)]
Add tools/tiny_ssim for generating SSIM/PSNR.
Change-Id: Icc3e5aaa6636ffe17dc9da5f7a80afaccbde509a
Linfeng Zhang [Fri, 28 Oct 2016 15:57:55 +0000 (15:57 +0000)]
Merge "Refine 8-bit intra prediction NEON optimization (mode d45 and d135)"
Paul Wilkins [Wed, 12 Oct 2016 19:50:08 +0000 (20:50 +0100)]
Change to KF boost calculation.
This change is a step in a larger change to the way boost and interval are
determined for ARF and Key frames.
This patch contains some pluming for the general case but focuses on the
key frame boost calculation. This now relies more heavily on the rate at
which the error score increases between the primary and secondary reference
frame. This seems to be less fragile when dealing with different frame sizes.
For example larger image formats tend in the first pass to see a higher
% of intra coded blocks and the use of this number in calculating the frame
decay factor was leading to much lower boost numbers for 4K, for example,
than the same clip coded at 2K.
This change does give overall gains but they are MUCH larger for the 4K Netflix
set. For the 4K Netflix set the average gain is around 3% with some clips > 20%
whereas for the same set at 2K the average gain is 0.5-1%.
In general for small image formats the boost is most often reduced a little whereas
4K clips the boost is increased. There are some -ve cases such as Akiyo at 352x288
where the reduced boost hurts the metrics, especially for SSIM, even while
the set as a whole improves. This is most notable at very low Q and may be the
subject of a future patch.
Some common code for KF and ARF was separated in this patch for the purposes of
tuning but may later be re-merged if appropriate.
Change-Id: Iaa15ac5a58d2be89181100d95cef6a8dc4b12d0d
Debargha Mukherjee [Fri, 28 Oct 2016 04:21:43 +0000 (04:21 +0000)]
Merge "Force recode if framesize exceeds max allowed size"
Linfeng Zhang [Fri, 21 Oct 2016 17:44:22 +0000 (10:44 -0700)]
Refine 8-bit intra prediction NEON optimization (mode d45 and d135)
dst += stride behaving better with gcc/clang.
Unroll loops.
Change-Id: I83f85df2bc9f17c6159542f57680b509395db2b1
Debargha Mukherjee [Thu, 27 Oct 2016 16:55:36 +0000 (09:55 -0700)]
Speed-up recode loop for extreme bitrate diffs
Adjusts the q adjustement step depending on how far the
projected and target rates differ.
Change-Id: I498d03523ca233a270512ca3972c372daa4ca2a8
Debargha Mukherjee [Thu, 27 Oct 2016 05:40:59 +0000 (22:40 -0700)]
Force recode if framesize exceeds max allowed size
Fixes a case where recode is not triggered based on the value
of maxq passed into the recode loop test function.
BUG=b/
32375284
Change-Id: I15ad985d0525c68e0443cfaf842440d2754b2266
Johann Koenig [Thu, 27 Oct 2016 15:41:43 +0000 (15:41 +0000)]
Merge "partial_idct_test: consolidate block size"
Paul Wilkins [Thu, 27 Oct 2016 10:21:23 +0000 (10:21 +0000)]
Merge "Changes to KF boost calculation."
Paul Wilkins [Thu, 27 Oct 2016 10:21:05 +0000 (10:21 +0000)]
Merge "Removal of a couple of two pass adjustments."
Johann [Thu, 27 Oct 2016 04:24:46 +0000 (21:24 -0700)]
partial_idct_test: add _add_ test
The result of the transform is added to the destination buffers. In the
existing tests the destination buffer is always empty so that portion of
the code was never exercised.
Change-Id: I1858c4fed2274f1b9faf834d2ba4186a4510492a
Johann [Thu, 27 Oct 2016 04:19:36 +0000 (21:19 -0700)]
partial_idct_test: consolidate block size
Use *input_block_ for sizeof() calculation like the other test
Change-Id: I1e4bd227131662056405af78c5052ad6ef769e9f
Johann [Fri, 21 Oct 2016 22:09:53 +0000 (15:09 -0700)]
Refactor partial idct test
Switch to using correctly sized inputs and outputs. This simplifies
adding tests with varying strides.
Change-Id: I716a0d8173dcf6a86d56656ac9d3101b7ec27642
Paul Wilkins [Wed, 12 Oct 2016 18:49:19 +0000 (19:49 +0100)]
Changes to KF boost calculation.
Remove double counting of decay. Limit maximum KF boost.
Change-Id: I0fb2344d0f78b5e95bb899dfad12b0ca84034b2c
paulwilkins [Fri, 30 Sep 2016 13:51:39 +0000 (14:51 +0100)]
Removal of a couple of two pass adjustments.
Removed a couple of adjustments that no longer move the needle
much but complicate the process of tuning.
Change-Id: Ie320f5cf155e6aac14a4757ea9ada2cd59f27590
Linfeng Zhang [Wed, 26 Oct 2016 16:51:44 +0000 (16:51 +0000)]
Merge "Refine 8-bit intra prediction NEON optimization (mode dc)"
Johann [Wed, 12 Oct 2016 22:27:47 +0000 (15:27 -0700)]
Optimize idct32x32_34_add for NEON
Approximately 3 times faster than the 1024 version which was used
previously.
BUG=webm:1295
Change-Id: Id15fb3d096029ec38ef01c53e5f6eb08254347c9
James Zern [Tue, 25 Oct 2016 21:59:29 +0000 (21:59 +0000)]
Merge "Update vp9_intrapred_test.cc to support 8-bit"
Yunqing Wang [Tue, 25 Oct 2016 20:54:25 +0000 (20:54 +0000)]
Merge "Modify the encoder multi-thread unit test"
Yunqing Wang [Tue, 25 Oct 2016 17:47:21 +0000 (10:47 -0700)]
Modify the encoder multi-thread unit test
Modified the encoder multi-thread test so that it included cpu-used=0 and
frame-parallel=0.
frame_parallel_decoding_mode is 1 by default, which disables probability
updating and gives lower encoding quality. Current VP9 multi-threading
encoder and decoder support probability updating. To test this part, we
should turn on it in the unit test, namely, setting frame-parallel to 0.
Change-Id: Ia1f86e01f0de628f50d819ae31509de3e1b6c755
James Bankoski [Tue, 25 Oct 2016 18:34:06 +0000 (18:34 +0000)]
Merge "vpxdec: return fail if frame fails to decode."
Yunqing Wang [Tue, 25 Oct 2016 16:00:58 +0000 (09:00 -0700)]
Change 2 motion search counts to be tile data
This patch modified the motion search counts used in:
https://chromium-review.googlesource.com/#/c/305640/
These 2 counts were originally added as thread data, and used to
make decisions in motion search. The tile encoding order can be
inconsistent while using different number of threads, which can
cause bitstream mismatch. Here moved them to tile data to solve
the issue.
BUG=webm:1322
Change-Id: Iedc4477aef1746aa0a4f84d88a1156296fd3ba87
Linfeng Zhang [Wed, 19 Oct 2016 20:37:26 +0000 (13:37 -0700)]
Refine 8-bit intra prediction NEON optimization (mode dc)
dst += stride behaving better with gcc/clang
Expanding inline function dc_SIZExSIZE() save intructions for
vpx_dc_predictor_SIZExSIZE_neon().
Change-Id: Id0ccbd58b6a31df539141fd33bdf28633339150d
Linfeng Zhang [Wed, 19 Oct 2016 19:58:05 +0000 (12:58 -0700)]
Update vp9_intrapred_test.cc to support 8-bit
BUG=webm:1316
Change-Id: Ic9309bbeeef52e9d07fb4a4c95c12efa813cbf8c
Jim Bankoski [Mon, 24 Oct 2016 19:05:59 +0000 (12:05 -0700)]
vpxdec: return fail if frame fails to decode.
A failure to decode is most likely equivalent to a corrupt
frame for the purpose of returning a failure.
Change-Id: Ie53db2b8130b40b725841f5f7a299d63aa56913d
James Zern [Sat, 22 Oct 2016 02:29:55 +0000 (02:29 +0000)]
Merge "remove idct32x32*_add_neon.asm"
James Zern [Sat, 22 Oct 2016 02:28:15 +0000 (02:28 +0000)]
Merge "vpx_highbd_convolve_copy_neon: use multi reg loads"
Vignesh Venkatasubramanian [Sat, 22 Oct 2016 02:23:05 +0000 (02:23 +0000)]
Merge "vp9_bitstream: Encode tiles in parallel"
Vignesh Venkatasubramanian [Wed, 19 Oct 2016 18:07:41 +0000 (11:07 -0700)]
vp9_bitstream: Encode tiles in parallel
Re-use the tile worker threads to pack the bitstream in parallel
on a per-tile basis. Restricting this to real-time only for now
(further testing is needed to ensure this does not make 2-pass
worse in any case).
BUG=webm:1309
Change-Id: I8a80da7c5089b837d0df79a5c49d5e3022dfc8ec
Marco [Fri, 21 Oct 2016 18:11:34 +0000 (11:11 -0700)]
vp9: Nonrd variance partition: increase threshold for using 4x4 avg.
In variance partition low resolutions may use varianace based on
4x4 average for better partitioning.
Increase the threshold for doing this at speed = 8.
Improves speed by ~5%, with little loss, < 1%, on RTC_derf set.
Change-Id: Ib5ec420832ccff887a06cb5e1d2c73199b093941
James Zern [Thu, 20 Oct 2016 04:04:12 +0000 (21:04 -0700)]
remove idct32x32*_add_neon.asm
the intrinsics are neutral to ~20% faster on cros/android
devices when using gcc-4.9/clang-3.8.1 and gcc-4.9/clang-3.8.x from the
r13 ndk. neutral results typically came with gcc-4.9 while larger
positive gains were achieved with clang 3.8.x.
BUG=webm:1303
Change-Id: I4d31f9c017944681b881493525d4573a7a5b1e16
Marco [Fri, 21 Oct 2016 00:02:16 +0000 (17:02 -0700)]
vp9: Speed=8 real-time: Keep the bias_golden feature on.
Small/no change in metrics on RTC set, speed increase by 2-3%.
Change-Id: Iee997bd7433e8e508216e9267b1c31c5a9aa5121
Marco Paniconi [Thu, 20 Oct 2016 22:38:55 +0000 (22:38 +0000)]
Merge "vp8: Apply gf target-size boost only when refresh_golden_frame = 1."