James Bankoski [Sat, 10 Dec 2016 00:01:32 +0000 (00:01 +0000)]
Merge "vp8: adds multithread testing."
Marco Paniconi [Fri, 9 Dec 2016 23:47:01 +0000 (23:47 +0000)]
Merge "vp9: Fix to crash in svc code."
Jim Bankoski [Fri, 9 Dec 2016 23:05:59 +0000 (15:05 -0800)]
vp8: adds multithread testing.
The test is disabled because of TSAN errors until we resolve
BUG=webm:851
Change-Id: I0b21c8d815bc1ea365da024b1e2ee5e1fc5715c2
Johann [Fri, 9 Dec 2016 20:54:55 +0000 (12:54 -0800)]
Move load_and_transpose to transpose_neon.h
Allows for use outside the idcts without pulling in idct_neon.h
Change-Id: I4a94c1af3dac3e1b5bc8296ec9eab0ddcc8cfecf
Marco [Wed, 7 Dec 2016 17:50:27 +0000 (09:50 -0800)]
vp9: Fix to crash in svc code.
use_base_mv assumes 2x2 scaling, so fix is to shutoff
this feature unless spatial scale factors are 2.
Added svc unittest for 2 spatial layers with 5x5 scaling,
which generates the issue without this fix.
Also fix some settings in svc unittest:
let the speed setting vary (from 5 to 8), and enable static threshold.
BUG=webm:1344
Change-Id: Idfd0a6c633c21b49a0479601506302cfe974e30e
James Zern [Fri, 9 Dec 2016 03:19:16 +0000 (03:19 +0000)]
Merge "idct16x16_add_neon: fix arm visual studio builds"
Marco [Thu, 8 Dec 2016 23:25:07 +0000 (15:25 -0800)]
vp8 multi_res_encoder: Ajust some settings in sample encoder.
Set #threads to default 1 for all streams, change bit allocaton
for 3 temporal layers, and enable denoiser on middle resolution layer.
Change-Id: I4a57adbfdb2c319002b8f3cf359613842dc00d75
James Zern [Thu, 8 Dec 2016 21:02:30 +0000 (13:02 -0800)]
idct16x16_add_neon: fix arm visual studio builds
after:
2d3d95f enable vpx_idct16x16_256_add_neon in hbd builds
reorder INCLUDEs and fix indent of IF/ENDIFs
remove vpx_config.asm to avoid multiple symbol definitions in windows
builds and shift idct_neon.asm.S to the top to allow use of
CONFIG_VP9_HIGHBITDEPTH in the export list.
Change-Id: I0dacfbae62a6ec8fe4a26940c1a52da2dfad2029
Yunqing Wang [Thu, 8 Dec 2016 22:46:44 +0000 (22:46 +0000)]
Merge "Remove an unused first pass statistic"
Yunqing Wang [Wed, 7 Dec 2016 18:00:36 +0000 (10:00 -0800)]
Remove an unused first pass statistic
One of the first pass stats "new_mv_count" is no longer used in VP9,
and is removed. This also makes it easy to implement a multi-threaded
first pass. This change doesn't affect the coding performance, which
has been verified by borg tests.
Change-Id: I4c7c7bf9465fda838eb230814ef0c631c068c903
Marco Paniconi [Wed, 7 Dec 2016 22:06:01 +0000 (22:06 +0000)]
Merge "vp9: Fix some TODOs in svc code."
Linfeng Zhang [Wed, 7 Dec 2016 21:05:05 +0000 (21:05 +0000)]
Merge "Update TEST_P(PartialIDctTest, RunQuantCheck)"
Linfeng Zhang [Wed, 7 Dec 2016 21:03:21 +0000 (21:03 +0000)]
Merge "Update idct NEON optimization to not use narrowing saturating shift"
Marco [Wed, 7 Dec 2016 21:02:10 +0000 (13:02 -0800)]
vp9: Fix some TODOs in svc code.
Change-Id: Ie9f441245987ade9dab38af69adf4dd1fb38ca3f
James Zern [Wed, 7 Dec 2016 20:26:43 +0000 (20:26 +0000)]
Merge "enable vpx_idct16x16_256_add_neon in hbd builds"
Linfeng Zhang [Wed, 7 Dec 2016 19:34:00 +0000 (11:34 -0800)]
Update TEST_P(PartialIDctTest, RunQuantCheck)
1. Use correct projections when copying real dct/quant outputs.
2. Remove local random number generator and combine loops.
3. Quantization with minimum allowed step sizes instead of maximum.
This may generate larger inputs.
Change-Id: I154afc26230c894d564671cff4b8fd5485b69598
Marco Paniconi [Wed, 7 Dec 2016 19:31:13 +0000 (19:31 +0000)]
Merge "vp9: Adjust the weight factor for segment rate cost for aq-mode=3."
Linfeng Zhang [Tue, 22 Nov 2016 22:21:36 +0000 (14:21 -0800)]
Update idct NEON optimization to not use narrowing saturating shift
Change-Id: Iae517017217dbacd638d40fcfeeb0f4bba7b8b8b
James Zern [Tue, 29 Nov 2016 04:20:28 +0000 (20:20 -0800)]
enable vpx_idct16x16_256_add_neon in hbd builds
BUG=webm:1294
Change-Id: Ib421c150b0d29dee0a81390a612bf01a4a28cff1
James Zern [Wed, 7 Dec 2016 01:40:28 +0000 (01:40 +0000)]
Merge changes Ibad079f2,I7858a0a1
* changes:
enable vpx_idct16x16_10_add_neon in hbd builds
idct16x16,NEON: rm output_stride from pass1 fns
James Zern [Thu, 24 Nov 2016 04:48:00 +0000 (20:48 -0800)]
enable vpx_idct16x16_10_add_neon in hbd builds
BUG=webm:1294
Change-Id: Ibad079f25e673d4f5181961896a8a8333a51e825
James Zern [Wed, 30 Nov 2016 03:47:50 +0000 (19:47 -0800)]
idct16x16,NEON: rm output_stride from pass1 fns
vpx_idct16x16_256_add_neon_pass1, vpx_idct16x16_10_add_neon:
this was a constant 8 in all cases meaning the results are stored
contiguously, this allows the number of stores to be reduced.
Change-Id: I7858a0a15a284883ef45c13dfd97c308df9ea09e
Linfeng Zhang [Wed, 16 Nov 2016 21:22:50 +0000 (13:22 -0800)]
Refine 8-bit 8x8 idct NEON intrinsics
Change-Id: I4ec4ad1928ec2ed87f596f52f097bc52065278dd
Marco [Mon, 5 Dec 2016 20:05:35 +0000 (12:05 -0800)]
vp9: Adjust the weight factor for segment rate cost for aq-mode=3.
Use the segment weight factor based on the target (cr->percent_refresh)
if it less than the current estimate (avergae of past usage and target).
Small improvement at low bitrates.
Change-Id: Iba8fd909e203f94458901366d3a991f7ea854d49
Linfeng Zhang [Fri, 7 Oct 2016 00:01:30 +0000 (17:01 -0700)]
Check in vpx_lpf_vertical_4_dual_neon() assembly
This replaces its C version.
Change-Id: Ie39e9324305fdc0fff610ced608a037e44a85a1a
James Zern [Fri, 2 Dec 2016 04:01:27 +0000 (04:01 +0000)]
Merge changes I4afc130e,Iaa64d23f
* changes:
Add high bitdepth 4x4 idct NEON intrinsics
Update idct x86 intrinsics to not use saturated add and sub
Linfeng Zhang [Tue, 8 Nov 2016 23:20:18 +0000 (15:20 -0800)]
Add high bitdepth 4x4 idct NEON intrinsics
Change-Id: I4afc130effa05b8be2e9f982967216b1beb2ce4b
Linfeng Zhang [Wed, 23 Nov 2016 00:59:17 +0000 (16:59 -0800)]
Update idct x86 intrinsics to not use saturated add and sub
Change-Id: Iaa64d23fdb45ca1f235b0ea57e614516e548eca4
James Zern [Thu, 24 Nov 2016 01:21:26 +0000 (17:21 -0800)]
idct16x16,NEON,cosmetics: normalize fn signatures
+ remove unused parameters from vpx_idct16x16_10_add_neon_pass2
Change-Id: Ie5912a4abdd308fab589380bca054a2e7234a2c4
James Zern [Tue, 29 Nov 2016 00:12:45 +0000 (00:12 +0000)]
Merge changes Ide6d3994,I164cfcbe
* changes:
enable vpx_idct32x32_135_add_neon in hbd builds
idct_neon: rename load_tran_low_to_s16 -> ...s16q
James Zern [Mon, 28 Nov 2016 23:39:37 +0000 (23:39 +0000)]
Merge "build/make/Android.mk: correct rtcd template var refs"
James Bankoski [Mon, 28 Nov 2016 22:27:25 +0000 (22:27 +0000)]
Merge "svc_test: fix two warnings"
Jim Bankoski [Mon, 28 Nov 2016 20:53:39 +0000 (12:53 -0800)]
svc_test: fix two warnings
Use of possibly uninitialized variable and missing test initializer.
Change-Id: I2192c81c39ef4239cc11a309850c0ee8781ef17e
Jerome Jiang [Mon, 28 Nov 2016 18:08:12 +0000 (10:08 -0800)]
Cosmetic changes to variable names in deblocker tests.
Change kExpectedOutput to expected_output in function parameters in
the deblocker test.
Change-Id: I5baf8d1285ac47922950887406c7aa519ddc512a
James Zern [Fri, 25 Nov 2016 01:51:10 +0000 (17:51 -0800)]
build/make/Android.mk: correct rtcd template var refs
the expansion of findstring and rtcd_dep_template_CONFIG_ASM_ABIS needs
to be deferred until the block is parsed as makefile syntax rather than
eval time where rtcd_dep_template_CONFIG_ASM_ABIS will be unset. this
ensures vpx_config.asm is properly created.
Change-Id: I7c38c6c082da78397936467482789dd468adc316
James Zern [Thu, 24 Nov 2016 07:22:04 +0000 (07:22 +0000)]
Merge changes I6b4cd56e,I88f91b92
* changes:
Android.mk,armv7: fix idct_neon.asm.S creation
build/make/Android.mk: set/use qexec appropriately
James Zern [Wed, 23 Nov 2016 01:47:38 +0000 (17:47 -0800)]
enable vpx_idct32x32_135_add_neon in hbd builds
BUG=webm:1294
Change-Id: Ide6d3994fe01c4320c9d143e6d059b49568048e4
James Zern [Sat, 19 Nov 2016 04:25:17 +0000 (20:25 -0800)]
idct_neon: rename load_tran_low_to_s16 -> ...s16q
BUG=webm:1294
Change-Id: I164cfcbe9bc4511d1d04af9206cf351a0ec2957b
James Zern [Thu, 24 Nov 2016 03:31:25 +0000 (03:31 +0000)]
Merge changes Icc4ead05,Ib019964b,I3b5fd3b3,Ieedadee2
* changes:
Update vpx_idct4x4_16_add_neon() to pass SingleExtremeCoeff test
Refine 8-bit 4x4 idct NEON intrinsics
Add idct speed test.
Update partial_idct_test.cc to support high bitdepth
Jerome Jiang [Thu, 24 Nov 2016 01:56:34 +0000 (01:56 +0000)]
Merge "Change C/MSA post proc to match SSE2."
Jerome Jiang [Thu, 24 Nov 2016 01:56:22 +0000 (01:56 +0000)]
Merge "Cover more filter levels in unit tests for post proc."
James Zern [Thu, 24 Nov 2016 00:49:19 +0000 (16:49 -0800)]
Android.mk,armv7: fix idct_neon.asm.S creation
force this to be created before any other .S files. this change
additionally removes the file from the source list as it doesn't need to
be compiled on its own.
Change-Id: I6b4cd56ef6059d08f75f06fb749cddf76e0e165e
James Zern [Thu, 24 Nov 2016 00:46:50 +0000 (16:46 -0800)]
build/make/Android.mk: set/use qexec appropriately
commands are echo'd when V=1; libs.mk depends on this variable as well
Change-Id: I88f91b9260f16686cfccdf6bd3f29d246521b62e
Marco [Tue, 22 Nov 2016 00:37:32 +0000 (16:37 -0800)]
vp9: Adjust cyclic refresh parameters for low bitrates.
Increase the motion threshold and qp-delta for segment#2 boost.
This can increase the frame-drop at low bitrates, but generally
better spatial quality.
Only affects real-time mode with aq-mode=3, at very low bitrates.
Change-Id: I5ccb784667f70d0c27d369806b93b1f93d5605d1
James Zern [Wed, 23 Nov 2016 19:27:20 +0000 (19:27 +0000)]
Merge "use storage.googleapis for testdata download"
Jerome Jiang [Sat, 19 Nov 2016 01:11:59 +0000 (17:11 -0800)]
Change C/MSA post proc to match SSE2.
BUG=webm:1321
Change-Id: I719023375dc48cf7d8ed72188853f0f1ccc4ad7f
Jerome Jiang [Sat, 19 Nov 2016 01:07:20 +0000 (17:07 -0800)]
Cover more filter levels in unit tests for post proc.
For some filter level, the C/MSA doesn't match SSE2. Part of unit tests
are disabled. They will be re-enabled when C/MSA funcs are fixed.
BUG=webm:1321
Change-Id: Ib16b98b5eecb15d2252aa4ea267b782ee2b27533
Marco Paniconi [Wed, 23 Nov 2016 18:15:58 +0000 (18:15 +0000)]
Merge "vp9: Use more aggressive skip when short_circuit_low_temp_var = 1."
James Zern [Wed, 23 Nov 2016 07:03:12 +0000 (23:03 -0800)]
use storage.googleapis for testdata download
replace downloads.webmproject.org with the canonical
storage.googleapis.com/... form. this appears less likely to fail when
dealing with multiple concurrent connections.
Change-Id: I0dcbd04df9e4057fa851f458b3ef7e3589f1f2f1
James Zern [Wed, 23 Nov 2016 00:43:21 +0000 (00:43 +0000)]
Merge "avoid redefining WIN32_LEAN_AND_MEAN"
James Zern [Wed, 23 Nov 2016 00:42:24 +0000 (00:42 +0000)]
Merge "vp9,read_inter_block_mode_info: quiet msan warning"
James Zern [Tue, 22 Nov 2016 02:20:33 +0000 (18:20 -0800)]
vp9,read_inter_block_mode_info: quiet msan warning
best_sub8x8[1] won't be used meaningfully when is_compound is false, but
may trigger an msan warning as the value is copied around and later
clamped.
BUG=667044
Change-Id: Icc24c3b72cdb550bebea44d4aaa4ff8bf3fbab56
Linfeng Zhang [Tue, 22 Nov 2016 23:20:53 +0000 (23:20 +0000)]
Merge "Add 32x32 d45 and 8x8, 16x16, 32x32 d135 NEON intra prediction"
James Zern [Tue, 22 Nov 2016 23:15:53 +0000 (15:15 -0800)]
avoid redefining WIN32_LEAN_AND_MEAN
fixes redef errors when the macro is supplied elsewhere, e.g., the
command line
Change-Id: Ic15726817a43e30595d50562ef1f077060c193cf
Marco [Tue, 22 Nov 2016 18:10:06 +0000 (10:10 -0800)]
vp9: Use more aggressive skip when short_circuit_low_temp_var = 1.
Use the same feature as https://chromium-review.googlesource.com/#/c/411327/,
but allow it to be used for speed = 6 and 7, where
short_circuit_low_temp_var = 1.
Speed up of ~2-3% for speed 7, with little/no loss in compression.
Change-Id: I263a0f261ad9929034392d68f0153dc6376fdb5f
Jerome Jiang [Tue, 22 Nov 2016 21:35:12 +0000 (13:35 -0800)]
Cosmetic changes to post proc unit tests.
Remove unnecessary "virtual" before some functions. Change *_btm_* in
variable names to *_bottom_*.
Change-Id: Ifd4ce667537617f451cdfed47dd8c48817fd983b
James Zern [Tue, 22 Nov 2016 20:14:12 +0000 (20:14 +0000)]
Merge "build/make/Android.mk: use -fPIC w/ENABLE_SHARED=1"
Linfeng Zhang [Tue, 22 Nov 2016 19:35:05 +0000 (11:35 -0800)]
Update vpx_idct4x4_16_add_neon() to pass SingleExtremeCoeff test
Change-Id: Icc4ead05506797d12bf134e8790443676fef5c10
James Bankoski [Tue, 22 Nov 2016 19:34:16 +0000 (19:34 +0000)]
Merge "vp9-tests : split VpxEncoderThreadTest into two tests."
Linfeng Zhang [Tue, 15 Nov 2016 18:21:35 +0000 (10:21 -0800)]
Refine 8-bit 4x4 idct NEON intrinsics
Change-Id: Ib019964bfcbce7aec57d8c3583127f9354d3c11f
Linfeng Zhang [Tue, 8 Nov 2016 01:23:38 +0000 (17:23 -0800)]
Add idct speed test.
Change-Id: I3b5fd3b36cac1fb3a93e27fd8fd0781c91d412ce
Linfeng Zhang [Fri, 4 Nov 2016 18:24:47 +0000 (11:24 -0700)]
Update partial_idct_test.cc to support high bitdepth
BUG=webm:1301
Change-Id: Ieedadee221ce539e39bf806c41331f749f891a3c
Jim Bankoski [Tue, 22 Nov 2016 15:31:04 +0000 (07:31 -0800)]
vp9-tests : split VpxEncoderThreadTest into two tests.
VpxEncoderThreadTest was taking a very long time for some runs and
timing out a lot. This is an attempt to split the test into runs
that can be run nightly ( speeds 2 through 9) and runs that can
be run weekly ( speeds 0-1 ).
Change-Id: Iee6f61a561006d3a30381dd3b52b9a4dce07a70c
Kaustubh Raste [Tue, 22 Nov 2016 12:19:17 +0000 (17:49 +0530)]
Fix mips dspr2 build warning
Change-Id: Ia8fb3ed124f01384e7896e309c9ff22c05b40719
Yaowu Xu [Mon, 21 Nov 2016 18:49:56 +0000 (10:49 -0800)]
Add validation of frame_parallel_decoding_mode
This is a boolean value that is written into bitstream, any value other
than 0 or 1 could have led to unexpected behavior. This commit fix the
issue by adding validation of the value to make sure it is boolean.
BUG=webm:1339
Change-Id: I2d3e69e8dbefcab9a0db9cb39a91a40ce531c5a1
Jingning Han [Sat, 19 Nov 2016 04:12:55 +0000 (04:12 +0000)]
Merge "Enable asymptotic closed-loop encoding decision"
Kaustubh Raste [Sat, 19 Nov 2016 03:37:29 +0000 (03:37 +0000)]
Merge "Fix SingleLargeCoeff idct test"
James Zern [Sat, 19 Nov 2016 03:11:55 +0000 (19:11 -0800)]
vpx_temporal_svc_encoder.sh: fix comment (// -> #)
Change-Id: Ib13152a9ff523b1c29e8519e4f7ed01ad9874525
James Zern [Sat, 19 Nov 2016 02:53:22 +0000 (18:53 -0800)]
build/make/Android.mk: use -fPIC w/ENABLE_SHARED=1
fixes reloc errors like:
R_X86_64_PC32
vpx_dsp/x86/deblock_sse2.o:
requires dynamic R_X86_64_PC32 reloc against 'vpx_rv' which may overflow
at runtime
Change-Id: I218fc0e7c8258197f890d395f335e5a4fe82dccb
James Zern [Sat, 19 Nov 2016 01:29:37 +0000 (01:29 +0000)]
Merge "Clean horizontal intra prediction NEON optimization"
James Zern [Fri, 18 Nov 2016 20:02:43 +0000 (20:02 +0000)]
Merge "partial_idct_test: s/SingleLargeCoef/SingleExtremeCoeff/"
Jerome Jiang [Fri, 18 Nov 2016 00:23:45 +0000 (00:23 +0000)]
Merge "Change *_xmm to *_sse2 in deblocker assembly functions."
Jerome Jiang [Thu, 17 Nov 2016 19:09:29 +0000 (11:09 -0800)]
Change *_xmm to *_sse2 in deblocker assembly functions.
Some cosmetic changes because xmm is an anachronism.
Change-Id: I436a5b78a3c52776c20d6640939311f2a84a9bc7
James Zern [Thu, 17 Nov 2016 20:28:57 +0000 (12:28 -0800)]
partial_idct_test: s/SingleLargeCoef/SingleExtremeCoeff/
tests with 'Large' in the name are reserved for slow running tests which
may not be run on all platforms
Change-Id: I2a7d6dd46b29b50469893e46433844132fb727c2
Marco Paniconi [Thu, 17 Nov 2016 20:22:32 +0000 (20:22 +0000)]
Merge "vpx_temporal_svc_encoder.sh: Run all tests for 1-4 threads for vp8/vp9."
Kaustubh Raste [Thu, 17 Nov 2016 11:35:53 +0000 (17:05 +0530)]
Fix SingleLargeCoeff idct test
Updated idct code to handle single large coefficient (-32768)
Change-Id: Ia13ab1ab434a9a1b9954a5914088977a88841cc7
Jerome Jiang [Thu, 17 Nov 2016 05:16:27 +0000 (05:16 +0000)]
Merge "Change C and msa to match results from sse2."
Jerome Jiang [Wed, 2 Nov 2016 16:55:49 +0000 (09:55 -0700)]
Change C and msa to match results from sse2.
Re-enable the tests to check CvsAssembly.
BUG=webm:1321
Change-Id: Id7f7d74b06c469fb6c8f5d04e91359e9cd9097a6
Marco [Wed, 16 Nov 2016 21:03:26 +0000 (13:03 -0800)]
vpx_temporal_svc_encoder.sh: Run all tests for 1-4 threads for vp8/vp9.
Change-Id: I079ee87cb32e36a1486c377c0037945b4bb89626
Jim Bankoski [Mon, 14 Nov 2016 14:17:19 +0000 (06:17 -0800)]
stress.sh: Runs multiple libvpx encodes and decodes in parallel
This runs multiple encodes and decodes of vp8 and vp9 in parallel,
with so many threads that problems with synchronization can show up.
Change-Id: I2b297e7f43d1e741323c7ad9f50a3931ae609f16
James Zern [Tue, 15 Nov 2016 23:41:48 +0000 (23:41 +0000)]
Merge "build/make/Android.mk: fix cpufeatures import"
James Zern [Tue, 15 Nov 2016 23:41:18 +0000 (23:41 +0000)]
Merge changes I3950c883,I2b679b04
* changes:
partial_idct_test: use <limits> for int16_min/max
vpx_timer.h,x86.h: define NOMINMAX for windows.h
Linfeng Zhang [Tue, 15 Nov 2016 23:30:48 +0000 (23:30 +0000)]
Merge "Add high bitdepth intra prediction NEON optimization (mode tm)"
Jerome Jiang [Tue, 15 Nov 2016 22:50:58 +0000 (22:50 +0000)]
Merge "vp9: Speed 8: More aggresive golden skip for low res."
Linfeng Zhang [Fri, 28 Oct 2016 22:04:45 +0000 (15:04 -0700)]
Add high bitdepth intra prediction NEON optimization (mode tm)
BUG=webm:1316
Change-Id: Ib014de06836ac12726f4a2c9f0833ec4eb4d233b
Jerome Jiang [Tue, 15 Nov 2016 18:37:12 +0000 (10:37 -0800)]
vp9: Speed 8: More aggresive golden skip for low res.
Add a new, more aggresive short circuit: short_circuit_low_temp_var = 3 to skip
golden of any mode when variance is lower than threshold for low res.
This change only affects speed = 8, low resolution.
Metrics for avgPSNR/SSIM on rtc_derf (low resolution) show loss of
0.27/0.31%.
On Nexus 6, the encoding time is reduced by ~2.3% on average across all
low-res clips.
Visually little change on rtc_derf clips.
Change-Id: Ia8f7366fc2d49181a96733a380b4dbd7390246ec
James Zern [Sat, 12 Nov 2016 20:59:34 +0000 (12:59 -0800)]
partial_idct_test: use <limits> for int16_min/max
this removes the need for __STDC_LIMIT_MACROS which is defined in
vpx_integer.h, but may be preceded by earlier includes of stdint.h;
fixes build with the r13 ndk
Change-Id: I3950c8837cf90d5584a20ce370ae370581c2182c
James Zern [Tue, 15 Nov 2016 20:16:14 +0000 (12:16 -0800)]
vpx_timer.h,x86.h: define NOMINMAX for windows.h
avoids the definition of min/max macros in headers that may appear in
c++ unit tests. the codebase uses VPXMIN/MAX for this purpose in any
case
Change-Id: I2b679b045d64fb34fd8780f704e3caf10a758d82
James Zern [Mon, 14 Nov 2016 21:21:27 +0000 (13:21 -0800)]
build/make/Android.mk: fix cpufeatures import
use 'android/cpufeatures' rather than 'cpufeatures'; this matches the
documentation, fixes compilation with r12b/r13 and still works with
older ndks.
Change-Id: I2f34233c164e6d4d46428f8905d5502cea4288a2
Jerome Jiang [Mon, 14 Nov 2016 18:22:00 +0000 (10:22 -0800)]
vp9: Speed 8: Turn off 4x4avg for low-res non-key frames.
Changes only affects speed = 8 for low resolutions.
Metrics for avgPSNR/SSIM on rtc_derf (low resolutions) show loss of
0.5/0.6%.
On Nexus 6, the encoding time is reduced by ~5.9% on average across all
low-res clips.
Visually little/no change on rtc_derf clips.
Change-Id: I68dd50e558d72dcc1af8317d224bfae5e3bd872d
Jingning Han [Sat, 12 Nov 2016 00:10:01 +0000 (16:10 -0800)]
Enable asymptotic closed-loop encoding decision
This commit enables asymptotic closed-loop encoding decision for
the key frame and alternate reference frame. It follows the regular
rate control scheme, but leaves out additional iteration on the
updated frame level probability model. It is enabled for speed 0.
The compression performance is improved:
lowres 0.2%
midres 0.35%
hdres 0.4%
Change-Id: I905ffa057c9a1ef2e90ef87c9723a6cf7dbe67cb
Linfeng Zhang [Fri, 28 Oct 2016 16:42:11 +0000 (09:42 -0700)]
Add high bitdepth intra prediction NEON optimization (h and v)
BUG=webm:1316
Change-Id: I47eeac698a98a31d1af5f72441052302e9fa4f46
Jerome Jiang [Sat, 12 Nov 2016 04:38:26 +0000 (04:38 +0000)]
Merge "Add unit tests for post proc."
Jerome Jiang [Wed, 2 Nov 2016 16:52:41 +0000 (09:52 -0700)]
Add unit tests for post proc.
Some tests are disabled since C and msa don't match sse2.
BUG=webm:1321
Change-Id: I61f303348e5292844a822612f100dbe006489e3e
Marco Paniconi [Fri, 11 Nov 2016 17:11:19 +0000 (17:11 +0000)]
Merge "vp9: Adjust thresholds for limiting cyclic refresh for noisy content."
James Zern [Thu, 10 Nov 2016 20:53:22 +0000 (20:53 +0000)]
Merge "*ppflags.h: remove unused *_DEBUG_* enum values"
James Zern [Thu, 10 Nov 2016 05:02:39 +0000 (05:02 +0000)]
Merge changes I339088b2,Iaade219e,If142afb1,I4257c4b3
* changes:
fdct8x8_test: add vpx_idct8x8_64_add_neon in hbd
fdct4x4_test: add vpx_idct4x4_16_add_neon in hbd
partial_idct_test,NEON: add missing idct variants
enable vpx_idct32x32_34_add_neon in hbd builds
James Zern [Wed, 9 Nov 2016 02:02:24 +0000 (18:02 -0800)]
fdct8x8_test: add vpx_idct8x8_64_add_neon in hbd
this was enabled in:
3ae2597 idct,NEON: add a tran_low_t->s16 load adapter
+ enable it for all NEON configs, both intrisincs and assembly versions
exist
BUG=webm:1294
Change-Id: I339088b2a398200f95658d040034fb9b2a7c8ce0
Linfeng Zhang [Thu, 27 Oct 2016 23:06:07 +0000 (16:06 -0700)]
Add high bitdepth intra prediction NEON optimization (mode d45 and d135)
BUG=webm:1316
Change-Id: I6a330874348df04df24a6d9efdc06f567e04bf8e
James Zern [Wed, 9 Nov 2016 05:09:16 +0000 (21:09 -0800)]
*ppflags.h: remove unused *_DEBUG_* enum values
usage of the vp8 versions was removed in:
3f72509 vp8: remove VP8_SET_DBG* control support
vp9 had the usage stripped even earlier.
Change-Id: I978142eb6492552cd29c9c6feb1e89acfc5f7b84