platform/upstream/libvpx.git
10 years agoOnly build neon assembly for armv7 targets
Johann [Wed, 7 May 2014 18:01:31 +0000 (11:01 -0700)]
Only build neon assembly for armv7 targets

Allow selectively building just the intrinsics for armv8

Change-Id: I2f29b2e4508b8b8e5649c2906b3159ad1d4ec477

10 years agoMerge "Fix check of debug counts for corrupt frame"
Adrian Grange [Sun, 11 May 2014 16:13:55 +0000 (09:13 -0700)]
Merge "Fix check of debug counts for corrupt frame"

10 years agoMerge "vp9_dx_iface: subtract ptrs to validate frame_size"
James Zern [Sat, 10 May 2014 18:05:31 +0000 (11:05 -0700)]
Merge "vp9_dx_iface: subtract ptrs to validate frame_size"

10 years agoMerge "Apply constrained partition search range to non-RD mode decision"
Jingning Han [Sat, 10 May 2014 17:39:26 +0000 (10:39 -0700)]
Merge "Apply constrained partition search range to non-RD mode decision"

10 years agoMerge "Removing VP9DecoderConfig."
Dmitry Kovalev [Sat, 10 May 2014 02:13:08 +0000 (19:13 -0700)]
Merge "Removing VP9DecoderConfig."

10 years agoApply constrained partition search range to non-RD mode decision
Jingning Han [Wed, 23 Apr 2014 21:54:16 +0000 (14:54 -0700)]
Apply constrained partition search range to non-RD mode decision

This commit enables a chessboard pattern for partition search. All
the black blocks run regular partition search ranging from 8x8 to
32x32. The rest white blocks take the nearby blocks' information
to adaptively decide the effective search range.

The compression performance for rtc set at speed -5 is down by 1.5%.
For pedestrian 1080p at speed -5, the runtime goes from 41594 ms to
39697 ms, i.e., about 5% faster.

Change-Id: Ia4b96e237abfaada487c743bca08fe1afd298685

10 years agoAdd another test vector
Yaowu Xu [Fri, 9 May 2014 23:20:49 +0000 (16:20 -0700)]
Add another test vector

The test vector has segment enabled with different quantizer used for
different segments for bot the first frame(key) frame and the rest of
non-key frames.

Change-Id: I7e21122183050ee046219caba483c18cbc34afe7

10 years agoMerge "Added a test vector"
Yaowu Xu [Fri, 9 May 2014 22:56:45 +0000 (15:56 -0700)]
Merge "Added a test vector"

10 years agoMerge "Remove bitrotted example code: example_xma"
Tom Finegan [Fri, 9 May 2014 21:11:22 +0000 (14:11 -0700)]
Merge "Remove bitrotted example code: example_xma"

10 years agoMerge "Add an x86inc MMX fwht4x4."
Alex Converse [Fri, 9 May 2014 20:48:49 +0000 (13:48 -0700)]
Merge "Add an x86inc MMX fwht4x4."

10 years agoMerge "Allow speed 5 lossless"
Alex Converse [Fri, 9 May 2014 20:48:42 +0000 (13:48 -0700)]
Merge "Allow speed 5 lossless"

10 years agoRemove bitrotted example code: example_xma
Tom Finegan [Fri, 9 May 2014 17:55:55 +0000 (10:55 -0700)]
Remove bitrotted example code: example_xma

Doesn't work. Doesn't build. Not supported.

Change-Id: I5a2f9965afdda82370049c10ebcbe77ac4c6c9b9

10 years agoFix check of debug counts for corrupt frame
Adrian Grange [Wed, 7 May 2014 17:31:55 +0000 (10:31 -0700)]
Fix check of debug counts for corrupt frame

Fixes the idecoder in the case where:
  cm->error_resilient_mode == 0, and
  cm->frame_parallel_decoding_mode == 0, but
  new_fb->corrupted == 1.

The assert in debug_check_frame_counts fails to
take into account the case of a corrupt frame.

Change-Id: Idf318a68458cc88d65d6f3f408a10d8ffe87e43f

10 years agoMerge changes I3edd4b95,I4514f974,Ie7fa4386
Jingning Han [Fri, 9 May 2014 16:58:39 +0000 (09:58 -0700)]
Merge changes I3edd4b95,I4514f974,Ie7fa4386

* changes:
  Turn on unit tests for SSSE3 8x8 forward and inverse 2D-DCT
  Change eob threshold for partial inverse 8x8 2D-DCT to 12
  SSSE3 8x8 inverse 2D-DCT with first 10 coeffs non-zero

10 years agoAllow speed 5 lossless
Alex Converse [Fri, 9 May 2014 01:28:16 +0000 (18:28 -0700)]
Allow speed 5 lossless

tx_mode supercedes whatever mechanism is used to push for 16x16
allowing for the use of the 4x4 transform.

Change-Id: I6c3f05ab9fe52050e40cc6303de9334653763289

10 years agoMerge "Skip testing large tx sizes when encoding lossless."
Yaowu Xu [Fri, 9 May 2014 16:28:05 +0000 (09:28 -0700)]
Merge "Skip testing large tx sizes when encoding lossless."

10 years agovp9_dx_iface: subtract ptrs to validate frame_size
James Zern [Fri, 9 May 2014 03:20:20 +0000 (20:20 -0700)]
vp9_dx_iface: subtract ptrs to validate frame_size

Change-Id: Ic5a6a4a2fec802d9c9c7a71dbae59d5b4d3a8b23

10 years agovp9_dx_iface: fix implicit ptrdiff_t -> uint32_t conversion
James Zern [Fri, 9 May 2014 03:19:49 +0000 (20:19 -0700)]
vp9_dx_iface: fix implicit ptrdiff_t -> uint32_t conversion

Change-Id: Ib773aac06b22009408cdd29a73190b276a53fcf6

10 years agoSkip testing large tx sizes when encoding lossless.
Alex Converse [Fri, 9 May 2014 01:04:56 +0000 (18:04 -0700)]
Skip testing large tx sizes when encoding lossless.

Change-Id: Ib7a17893f6dc0f501b1252ae32804ede144dde37

10 years agoMerge "Revert "Removing redundant variables from variance_test.cc.""
James Zern [Thu, 8 May 2014 23:28:55 +0000 (16:28 -0700)]
Merge "Revert "Removing redundant variables from variance_test.cc.""

10 years agoRemoving VP9DecoderConfig.
Dmitry Kovalev [Thu, 8 May 2014 23:24:36 +0000 (16:24 -0700)]
Removing VP9DecoderConfig.

We only used two members from that struct: max_threads and inv_tile_order.
Moving them directly to VP9Decoder struct.

Change-Id: If696a4e5b5b41868a55f3cc971e1d7c1dd9d5f69

10 years agoMerge "Remove decode_with_partial_drops example."
Tom Finegan [Thu, 8 May 2014 22:54:46 +0000 (15:54 -0700)]
Merge "Remove decode_with_partial_drops example."

10 years agoMerge "Add test for the vp8cx_set_ref example."
Tom Finegan [Thu, 8 May 2014 22:54:35 +0000 (15:54 -0700)]
Merge "Add test for the vp8cx_set_ref example."

10 years agoMerge "tools_common.sh: Fix vlog()"
Tom Finegan [Thu, 8 May 2014 22:54:26 +0000 (15:54 -0700)]
Merge "tools_common.sh: Fix vlog()"

10 years agoMerge "Simplifying decoder_decode() function."
Dmitry Kovalev [Thu, 8 May 2014 22:01:40 +0000 (15:01 -0700)]
Merge "Simplifying decoder_decode() function."

10 years agoRevert "Removing redundant variables from variance_test.cc."
James Zern [Thu, 8 May 2014 21:32:32 +0000 (14:32 -0700)]
Revert "Removing redundant variables from variance_test.cc."

This reverts commit 4725ab7e51a69bf890856e524c6d78aadd97d64e.

The constants are necessary to avoid breakage in vs9 builds:
 warning C4180: qualifier applied to function type has no meaning; ignored
 error C2436: 'f2_' : member function or nested class in constructor initializer list
 while compiling class template member function 'std::tr1::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>::tuple(const int &,const int &,unsigned int (__cdecl &))'
 ..\test\variance_test.cc : see reference to class template instantiation 'std::tr1::tuple<T0,T1,T2,T3,T4,T5,T6,T7,T8,T9>' being compiled

Change-Id: Ia218b74fc473d40f02fee84cb7009adfbe82e5a7

10 years agoMerge "webmdec: Fix visual studio warnings."
James Zern [Thu, 8 May 2014 19:43:54 +0000 (12:43 -0700)]
Merge "webmdec: Fix visual studio warnings."

10 years agoAdd test for the vp8cx_set_ref example.
Tom Finegan [Thu, 8 May 2014 19:35:36 +0000 (12:35 -0700)]
Add test for the vp8cx_set_ref example.

Change-Id: I68850699ecaebbbb7962cbc1c9bc85e92a0ea5ca

10 years agotools_common.sh: Fix vlog()
Tom Finegan [Thu, 8 May 2014 19:06:08 +0000 (12:06 -0700)]
tools_common.sh: Fix vlog()

Abusing '[] && echo' resulted in the --verbose flag being required to
actually run tests.

Change-Id: I39edaa30a05272928f9f4a4bb8581f91d6fe824b

10 years agoAdd an x86inc MMX fwht4x4.
Alex Converse [Wed, 7 May 2014 19:51:11 +0000 (12:51 -0700)]
Add an x86inc MMX fwht4x4.

Change-Id: Ib0a73d4863478f9b8a00976379d25d2f6ebbb197

10 years agoUse __asm__ __volatile__ with gcc
Johann [Thu, 8 May 2014 18:16:07 +0000 (11:16 -0700)]
Use __asm__ __volatile__ with gcc

Some version of clang refuse 'asm volatile'

Change-Id: I79d909ac8ae3c04b608f20c6f10fa79b7f9fc8e0

10 years agoTurn on unit tests for SSSE3 8x8 forward and inverse 2D-DCT
Jingning Han [Thu, 8 May 2014 16:56:38 +0000 (09:56 -0700)]
Turn on unit tests for SSSE3 8x8 forward and inverse 2D-DCT

Change-Id: I3edd4b956a1273d65547771bf43c5cdaea25e5d6

10 years agoChange eob threshold for partial inverse 8x8 2D-DCT to 12
Jingning Han [Thu, 8 May 2014 16:42:26 +0000 (09:42 -0700)]
Change eob threshold for partial inverse 8x8 2D-DCT to 12

The scanning order has the first 12 coefficients of the 8x8 2D-DCT
sitting in the top left 4x4 block. Hence the partial inverse 8x8
2D-DCT allows to handle cases with eob below 12.

The overall runtime of the inverse 8x8 2D-DCT unit is reduced from
166 cycles (using SSE2) to 150 cycles (using SSSE3).

Change-Id: I4514f9748042809ac84df4c14382c00f313f1cd2

10 years agoMerge "vp9_pickmode: Fix signed/unsigned mismatch."
Frank Galligan [Thu, 8 May 2014 15:49:12 +0000 (08:49 -0700)]
Merge "vp9_pickmode: Fix signed/unsigned mismatch."

10 years agoMerge "vp9_firstpass: Fix MSVC data loss warning."
Frank Galligan [Thu, 8 May 2014 15:49:04 +0000 (08:49 -0700)]
Merge "vp9_firstpass: Fix MSVC data loss warning."

10 years agoMerge "vp9_svc_layercontext: Clean up function definition mismatch."
Frank Galligan [Thu, 8 May 2014 15:48:54 +0000 (08:48 -0700)]
Merge "vp9_svc_layercontext: Clean up function definition mismatch."

10 years agoMerge "Removing redundant variables from variance_test.cc."
Dmitry Kovalev [Thu, 8 May 2014 02:00:18 +0000 (19:00 -0700)]
Merge "Removing redundant variables from variance_test.cc."

10 years agovp9_svc_layercontext: Clean up function definition mismatch.
Tom Finegan [Thu, 8 May 2014 01:12:13 +0000 (18:12 -0700)]
vp9_svc_layercontext: Clean up function definition mismatch.

vp9_is_upper_layer_key_frame() definition does not match declaration--
it was missing the second const.

Change-Id: I71312579eb443be1924b8b06d8b3177c3dcb40f3

10 years agovp9_firstpass: Fix MSVC data loss warning.
Tom Finegan [Thu, 8 May 2014 01:27:29 +0000 (18:27 -0700)]
vp9_firstpass: Fix MSVC data loss warning.

Change-Id: I5307fe85ee081fbcfdf55a974f7eba1e0070c723

10 years agovp9_pickmode: Fix signed/unsigned mismatch.
Tom Finegan [Thu, 8 May 2014 01:20:29 +0000 (18:20 -0700)]
vp9_pickmode: Fix signed/unsigned mismatch.

Avoids getting reverted on the next libvpx chrome roll attempt.

Change-Id: I2431bd601111a78320af5f211f6413009674f1c8

10 years agoSSSE3 8x8 inverse 2D-DCT with first 10 coeffs non-zero
Jingning Han [Fri, 2 May 2014 23:29:08 +0000 (16:29 -0700)]
SSSE3 8x8 inverse 2D-DCT with first 10 coeffs non-zero

This commit enables ssse3 assembly implementation of the 8x8
inverse 2D-DCT with only first 10 coefficients non-zero. The
average runtime for this unit goes down from 198 cycles to 129
cycles (34.8% faster).

Change-Id: Ie7fa4386f6d3a2fe0d47a2eb26fc2a6bbc592ac7

10 years agoMerge "Tune rate-distortion modeling to account for frame light change"
Jingning Han [Thu, 8 May 2014 00:36:00 +0000 (17:36 -0700)]
Merge "Tune rate-distortion modeling to account for frame light change"

10 years agoMerge "Remove separate arf and gf minq tables."
Yaowu Xu [Thu, 8 May 2014 00:04:49 +0000 (17:04 -0700)]
Merge "Remove separate arf and gf minq tables."

10 years agoMerge "Rate control adjustment."
Yaowu Xu [Thu, 8 May 2014 00:04:41 +0000 (17:04 -0700)]
Merge "Rate control adjustment."

10 years agoMerge "Relax rc test threshold"
Yaowu Xu [Wed, 7 May 2014 23:52:01 +0000 (16:52 -0700)]
Merge "Relax rc test threshold"

10 years agoMerge "Revert "Remove struct params from vp8_denoiser_filter""
Frank Galligan [Wed, 7 May 2014 23:21:46 +0000 (16:21 -0700)]
Merge "Revert "Remove struct params from vp8_denoiser_filter""

10 years agoRevert "Remove struct params from vp8_denoiser_filter"
Frank Galligan [Wed, 7 May 2014 23:19:20 +0000 (16:19 -0700)]
Revert "Remove struct params from vp8_denoiser_filter"

This reverts commit e516a42527098a26798dbb3663a5bcdd38793839

Change-Id: I7c78712acc737ad5f580181cdab3aa76b23f3ca5

10 years agoRemoving redundant variables from variance_test.cc.
Dmitry Kovalev [Wed, 7 May 2014 21:40:21 +0000 (14:40 -0700)]
Removing redundant variables from variance_test.cc.

Change-Id: Icd44bce1c9d292f6e6f4d5157b694f6170b7b289

10 years agoMerge "Cleaning up vp9_encode_frame() function."
Dmitry Kovalev [Wed, 7 May 2014 21:10:07 +0000 (14:10 -0700)]
Merge "Cleaning up vp9_encode_frame() function."

10 years agoMerge "Using SPEED_FEATURES instead of VP9_COMP in vp9_init_search_range()."
Dmitry Kovalev [Wed, 7 May 2014 21:07:33 +0000 (14:07 -0700)]
Merge "Using SPEED_FEATURES instead of VP9_COMP in vp9_init_search_range()."

10 years agoRemove separate arf and gf minq tables.
Paul Wilkins [Wed, 7 May 2014 15:19:50 +0000 (16:19 +0100)]
Remove separate arf and gf minq tables.

Merged minq tables for arf and gf cases.
These tables were almost the same and for
VBR the arf table was not used at all.

Change-Id: Ie3c87e91dab613cf06f6945ac1ace0e0e4213d34

10 years agoRate control adjustment.
Paul Wilkins [Wed, 23 Apr 2014 00:09:47 +0000 (17:09 -0700)]
Rate control adjustment.

Small adjustment to the active Q range calculations.
These changes should slightly extend the available Q range
for KF/GF/ARF and narrow it for other frames.

The results for this change in isolation are broadly positive
for SSIM and average PSNR and slightly up but mixed for opsnr.

derf +0.293% opsnr, +1.286% SSIM
std-hd + 0.528% opsnr, + 1.746% SSIM
yt +0.056% opsnr, +0.457% SSIM
yt-hd -0.147% opsnr, + 0.226% SSIM

Change-Id: If065280342027ecc5d44b49fc1d440dfef041002

10 years agoRelax rc test threshold
Yaowu Xu [Wed, 7 May 2014 20:49:36 +0000 (13:49 -0700)]
Relax rc test threshold

For multilayer coding

Change-Id: I830a0176c020658b836e3f5d2ce6bc5932736c0c

10 years agoMerge "Moving pair_set_epi32 macro into vp9_dct32x32_sse2.c."
Dmitry Kovalev [Wed, 7 May 2014 20:34:05 +0000 (13:34 -0700)]
Merge "Moving pair_set_epi32 macro into vp9_dct32x32_sse2.c."

10 years agoAdded a test vector
Yaowu Xu [Fri, 25 Apr 2014 23:15:45 +0000 (16:15 -0700)]
Added a test vector

The test vector is produced to have a single key frame, with segment
map enabled and transmitted. Yet no segment feature is active.

Change-Id: I365d62f00d05c07098b9a76fc8d3a991e427ec1a

10 years agoRemove decode_with_partial_drops example.
Tom Finegan [Wed, 30 Apr 2014 23:41:16 +0000 (16:41 -0700)]
Remove decode_with_partial_drops example.

Bitrotted and non-functional. Does one of two things: Crashes or reports
errors and quits.

Change-Id: Ia9c391ecd6e716b183d925247bf3dc0509ce4586

10 years agoMerge "Removing MODE_TEST_HIT_STATS."
Dmitry Kovalev [Wed, 7 May 2014 17:40:31 +0000 (10:40 -0700)]
Merge "Removing MODE_TEST_HIT_STATS."

10 years agoMerge "Revert "Add an MMX fwht4x4""
Alex Converse [Wed, 7 May 2014 17:09:04 +0000 (10:09 -0700)]
Merge "Revert "Add an MMX fwht4x4""

10 years agoMerge "Mark VP8E_SET_ARNR_TYPE as being deprecated"
Adrian Grange [Wed, 7 May 2014 16:03:02 +0000 (09:03 -0700)]
Merge "Mark VP8E_SET_ARNR_TYPE as being deprecated"

10 years agoMerge "Android NDK support for x86 and mips"
Johann [Wed, 7 May 2014 15:54:13 +0000 (08:54 -0700)]
Merge "Android NDK support for x86 and mips"

10 years agoMerge "Clean up full-pixel search calling code"
Yunqing Wang [Wed, 7 May 2014 15:33:03 +0000 (08:33 -0700)]
Merge "Clean up full-pixel search calling code"

10 years agoMerge "Revert "VP8 for ARMv8 by using NEON intrinsics 10""
Johann [Wed, 7 May 2014 13:47:48 +0000 (06:47 -0700)]
Merge "Revert "VP8 for ARMv8 by using NEON intrinsics 10""

10 years agoMerge "arm: Use a correct neon vector type for 64 bit integers"
Johann [Wed, 7 May 2014 13:34:25 +0000 (06:34 -0700)]
Merge "arm: Use a correct neon vector type for 64 bit integers"

10 years agoMerge "arm: Add a no-op define of __builtin_prefetch for MSVC"
Johann [Wed, 7 May 2014 13:02:22 +0000 (06:02 -0700)]
Merge "arm: Add a no-op define of __builtin_prefetch for MSVC"

10 years agoMerge "Fix building for arm with Visual Studio 2013"
Johann [Wed, 7 May 2014 13:00:54 +0000 (06:00 -0700)]
Merge "Fix building for arm with Visual Studio 2013"

10 years agoMerge "Don't try to use getenv on windows phone/rt"
Johann [Wed, 7 May 2014 13:00:04 +0000 (06:00 -0700)]
Merge "Don't try to use getenv on windows phone/rt"

10 years agoRevert "Add an MMX fwht4x4"
Paul Wilkins [Wed, 7 May 2014 11:53:27 +0000 (12:53 +0100)]
Revert "Add an MMX fwht4x4"

Includes changes that are not compatible with VS windows builds.
Amongst other things stdint.h is not supported in VS.

This reverts commit 89fbf3de501b5d7fd90047192521eae3198705cd.

Change-Id: Ifa86d7df250578d1ada9b539c9ff12ed0c523cdd

10 years agoarm: Add a no-op define of __builtin_prefetch for MSVC
Martin Storsjo [Wed, 7 May 2014 07:26:32 +0000 (10:26 +0300)]
arm: Add a no-op define of __builtin_prefetch for MSVC

Both GCC and RVCT/ARMCC support __builtin_prefetch, but MSVC
doesn't.

Change-Id: I44e1eecead61bc88d8fdfd3fef03d76d4f5afe08

10 years agoarm: Use a correct neon vector type for 64 bit integers
Martin Storsjo [Wed, 7 May 2014 07:22:18 +0000 (10:22 +0300)]
arm: Use a correct neon vector type for 64 bit integers

This fixes building with MSVC.

Change-Id: I763ba8855c8083d82c8b477d3a297e310e93a335

10 years agoTune rate-distortion modeling to account for frame light change
Jingning Han [Tue, 6 May 2014 23:26:30 +0000 (16:26 -0700)]
Tune rate-distortion modeling to account for frame light change

When the variance is far less than sse, the block is considered to
be under light change. All the energy is compacted into DC coeff
and can be coded at low cost. In such situation, switch the rate-
distortion modeling from sse+var based back to variance based.

Note that this is a temporary solution to handle the rare situations
where the scene light changes.

Change-Id: I1ee0fe2b9eda6b5fac40152e1841bf23f4d229fd

10 years agoRevert "VP8 for ARMv8 by using NEON intrinsics 10"
Johann [Tue, 6 May 2014 19:26:21 +0000 (12:26 -0700)]
Revert "VP8 for ARMv8 by using NEON intrinsics 10"

This reverts commit c500fc22c1bb2a3ae5c318bfb806f7e9bd57ce25

There is an issue with gcc 4.6 in the Android NDK:
loopfiltersimpleverticaledge_neon.c: In function 'vp8_loop_filter_bvs_neon':
loopfiltersimpleverticaledge_neon.c:176:1: error: insn does not satisfy its constraints:

Change-Id: I95b6509d12f075890308914cc691b813d2e5cd9f

10 years agoRevert "VP8 for ARMv8 by using NEON intrinsics 08"
Johann [Tue, 6 May 2014 19:25:33 +0000 (12:25 -0700)]
Revert "VP8 for ARMv8 by using NEON intrinsics 08"

This reverts commit a5d79f43b963ced59b462206faf3b7857bdeff7b

There is an issue with gcc 4.6 in the Android NDK:
loopfilter_neon.c: In function 'vp8_loop_filter_vertical_edge_y_neon':
loopfilter_neon.c:394:1: error: insn does not satisfy its constraints:

Change-Id: I2b8c6ee3fa595c152ac3a5c08dd79bd9770c7b52

10 years agowebmdec: Fix visual studio warnings.
Vignesh Venkatasubramanian [Wed, 30 Apr 2014 22:15:47 +0000 (15:15 -0700)]
webmdec: Fix visual studio warnings.

Fix a couple of visual studio warnings.

Change-Id: I161edf4b37f5ac2defbe6eb5b7e34d219b1f52d9

10 years agoMerge "third_party/libwebm: pull from upstream"
Vignesh Venkatasubramanian [Tue, 6 May 2014 18:15:57 +0000 (11:15 -0700)]
Merge "third_party/libwebm: pull from upstream"

10 years agoMerge "Add an MMX fwht4x4"
Alex Converse [Tue, 6 May 2014 18:12:27 +0000 (11:12 -0700)]
Merge "Add an MMX fwht4x4"

10 years agoMerge "Add test for 4x4 WHT."
Alex Converse [Tue, 6 May 2014 18:12:23 +0000 (11:12 -0700)]
Merge "Add test for 4x4 WHT."

10 years agoMerge "SSSE3 implementation of full inverse 8x8 2D-DCT"
Jingning Han [Tue, 6 May 2014 16:17:22 +0000 (09:17 -0700)]
Merge "SSSE3 implementation of full inverse 8x8 2D-DCT"

10 years agoMerge "[spatial svc] Fix one extra frame count during flush"
Minghai Shang [Tue, 6 May 2014 06:22:09 +0000 (23:22 -0700)]
Merge "[spatial svc] Fix one extra frame count during flush"

10 years agoMerge "[spatial svc] Use VPX_DL_GOOD_QUALITY for vp9_spatial_scalable_encoder"
Minghai Shang [Tue, 6 May 2014 06:22:03 +0000 (23:22 -0700)]
Merge "[spatial svc] Use VPX_DL_GOOD_QUALITY for vp9_spatial_scalable_encoder"

10 years agoMerge "[spatial svc] No need to code full width and height for non key frame"
Minghai Shang [Tue, 6 May 2014 06:21:53 +0000 (23:21 -0700)]
Merge "[spatial svc] No need to  code full  width and height for non key frame"

10 years agoMerge "Cleaning up vp9_full_range_search_c() function."
Dmitry Kovalev [Tue, 6 May 2014 05:01:02 +0000 (22:01 -0700)]
Merge "Cleaning up vp9_full_range_search_c() function."

10 years agoMaking vp9_get_sse_sum_{8x8, 16x16} static.
Dmitry Kovalev [Fri, 2 May 2014 01:40:24 +0000 (18:40 -0700)]
Making vp9_get_sse_sum_{8x8, 16x16} static.

Change-Id: Ifb7937c977308c682986f0ce9645a0807d2aa46a

10 years agoMerge "Moving costs from MACROBLOCK to VP9_COMP."
Dmitry Kovalev [Tue, 6 May 2014 02:10:25 +0000 (19:10 -0700)]
Merge "Moving costs from MACROBLOCK to VP9_COMP."

10 years agoMerge "Cleaning up vp9_variance.c"
Dmitry Kovalev [Tue, 6 May 2014 02:09:57 +0000 (19:09 -0700)]
Merge "Cleaning up vp9_variance.c"

10 years agoClean up full-pixel search calling code
Yunqing Wang [Tue, 6 May 2014 00:35:54 +0000 (17:35 -0700)]
Clean up full-pixel search calling code

Removed repetitive code.

Change-Id: Ib6adb6eaf7d4e3feeabb71651f4cc447974a925d

10 years agoMerge changes Ibfa447fe,I771809ba
Adrian Grange [Mon, 5 May 2014 22:41:10 +0000 (15:41 -0700)]
Merge changes Ibfa447fe,I771809ba

* changes:
  Fix generic-gnu target build
  Fix rounding in ARNR calculation

10 years agoAdd an MMX fwht4x4
Alex Converse [Mon, 5 May 2014 21:10:41 +0000 (14:10 -0700)]
Add an MMX fwht4x4

7% faster encoding a desktop lossless at RT speed 4.

Change-Id: I41627f5b737752616b6512bb91a36ec45995bf64

10 years agoAdd test for 4x4 WHT.
Alex Converse [Mon, 5 May 2014 20:50:12 +0000 (13:50 -0700)]
Add test for 4x4 WHT.

Change-Id: I97ba3152f9b44d23a74c6144809142810d07e368

10 years agothird_party/libwebm: pull from upstream
Vignesh Venkatasubramanian [Fri, 2 May 2014 18:44:35 +0000 (11:44 -0700)]
third_party/libwebm: pull from upstream

Pulling libwebm from upstream

Changes from upstream:
249629d make Mkv(Reader|Writer)(FILE*) explicit
7f3cda4 mkvparser: fix a bunch of windows warnings
5c06178 Merge "clang-format on mkvparser.[ch]pp"
4df111e clang-format on mkvparser.[ch]pp
7b24501 clang-format re-run.
c6767b9 Change AlignTrailingComments to false in .clang-format
9097a06 Merge "muxer: Reject file if TrackType is never specified"
eddf974 Merge "clang-format on mkvmuxertypes.hpp and webmids.hpp"
def325c muxer: Reject file if TrackType is never specified
41f869c Merge "clang-format on webvttparser.(cc|h)"
fd0be37 clang-format on webvttparser.(cc|h)
207d8a1 Merge "clang-format on mkvmuxerutil.[ch]pp"
02429eb Merge "clang-format on mkvwriter.[ch]pp"
0cf7b1b Merge "clang-format on mkvreader.[ch]pp"
2e80fed Merge "clang-format on sample.cpp"
3402e12 Merge "clang-format on sample_muxer.cpp"
1a685db Merge "clang-format on sample_muxer_metadata.(cc|h)"
6634c7f Merge "clang-format on vttreader.cc"
7566004 Merge "clang-format on vttdemux.cc"
9915b84 clang-format on mkvreader.[ch]pp
7437254 clang-format on mkvmuxertypes.hpp and webmids.hpp
0d5a98c clang-format on sample_muxer.cpp
e3485c9 clang-format on vttdemux.cc
46cc823 clang-format on dumpvtt.cc
5218bd2 clang-format on vttreader.cc
1a0130d clang-format on sample_muxer_metadata.(cc|h)
867f189 clang-format on sample.cpp
4c7bec5 clang-format on mkvwriter.[ch]pp
9ead078 clang-format on mkvmuxerutil.[ch]pp
fb6b6e6 clang-format on mkvmuxer.[ch]pp
ce77592 Update .clang-format to allow short functions in one line
0a24fe4 Merge "Add support for DateUTC and DefaultDuration in MKV Muxer."
11d5b66 Merge "Add .clang-format"
a1a3b14 Add .clang-format
0fcec38 Add support for DateUTC and DefaultDuration in MKV Muxer.

Change-Id: Ia0ed161ffc3d63c2eba8ed145707ffe543617976

10 years agoMerge "VP8 for ARMv8 by using NEON intrinsics 16"
Johann [Mon, 5 May 2014 20:09:43 +0000 (13:09 -0700)]
Merge "VP8 for ARMv8 by using NEON intrinsics 16"

10 years agoMerge "VP8 for ARMv8 by using NEON intrinsics 15"
Johann [Mon, 5 May 2014 20:09:12 +0000 (13:09 -0700)]
Merge "VP8 for ARMv8 by using NEON intrinsics 15"

10 years agoMark VP8E_SET_ARNR_TYPE as being deprecated
Adrian Grange [Mon, 5 May 2014 18:31:55 +0000 (11:31 -0700)]
Mark VP8E_SET_ARNR_TYPE as being deprecated

In a future release we plan to remove the
option of setting the ARNR filter type.

This patch marks this control as being deprecated
as advance warning that it will be removed from
the API at some point.

Change-Id: I5dcca804b44c7c93b1a10da7d69d19ba6061869c

10 years agoAdd test for postproc example.
Tom Finegan [Wed, 30 Apr 2014 21:33:24 +0000 (14:33 -0700)]
Add test for postproc example.

Change-Id: I3446d2ebc23e6f1ed650dd5c82648041febe3c4d

10 years agoFix generic-gnu target build
Adrian Grange [Mon, 5 May 2014 17:44:48 +0000 (10:44 -0700)]
Fix generic-gnu target build

Added macro to conditionally compile some of the
post-processing functions only when CONFIG_POSTPROC
is defined.

This was causing the build for the generic-gnu target
to fail.

Change-Id: Ibfa447feceb7a0528135025f105be48f97e9965c

10 years agoFix rounding in ARNR calculation
Adrian Grange [Mon, 5 May 2014 16:39:24 +0000 (09:39 -0700)]
Fix rounding in ARNR calculation

The rounding of the ARNR filter output prior to
normalization by the filter strength was incorrect
when strength = 0.

In this case 1 << (strength - 1) would not create the
required rounding of 0, rather it would outrange. This
patch fixes this issue.

Change-Id: I771809ba34d6052b17d34c870ea11ff67b418dab

10 years agoSSSE3 implementation of full inverse 8x8 2D-DCT
Jingning Han [Fri, 2 May 2014 01:34:46 +0000 (18:34 -0700)]
SSSE3 implementation of full inverse 8x8 2D-DCT

This commit enables SSSE3 version full inverse 8x8 2D-DCT and
reconstruction. It makes the runtime of vp9_idct8x8_64_add down
from 256 cycles (SSE2) to 246 cycles.

Change-Id: I0600feac894d6a443a3c9d18daf34156d4e225c3

10 years agoMerge "VP8 for ARMv8 by using NEON intrinsics 14"
Johann [Mon, 5 May 2014 14:08:08 +0000 (07:08 -0700)]
Merge "VP8 for ARMv8 by using NEON intrinsics 14"

10 years agoMerge changes Iaf7d6b0a,Iece0bf56
Johann [Mon, 5 May 2014 12:36:54 +0000 (05:36 -0700)]
Merge changes Iaf7d6b0a,Iece0bf56

* changes:
  Use INLINE and include vpx_config.h instead of plain 'inline'
  Use vreinterpret instead of casting neon vector types

10 years agoMerge "Remove struct params from vp8_denoiser_filter"
Scott LaVarnway [Mon, 5 May 2014 12:31:24 +0000 (05:31 -0700)]
Merge "Remove struct params from vp8_denoiser_filter"

10 years agoFix building for arm with Visual Studio 2013
Martin Storsjo [Sun, 4 May 2014 20:24:59 +0000 (23:24 +0300)]
Fix building for arm with Visual Studio 2013

The microsoft build tools explicitly disallow building for arm in
the "desktop" target configuration; one has to target "Windows
Store" apps (aka WinRT/Metro) or Windows Phone. In Visual Studio
2012, one could just pick the v110_wp80 toolset which made the
vcxproj files buildable. In Visual Studio 2013, picking the v120_wp81
toolset isn't enough - one has to configure the vcxproj files
as an "AppContainerApplication". This has the implication that
you can't just build a plain .exe (such as the examples) - an .exe
project would need to have an AppxManifest file. Therefore we can
only build the library itself.

If loaded into Visual Studio for Windows (the Windows Store/Phone
version of Visual Studio, not the Desktop one), the obj_int_extract
project is omitted since it's treated as incompatible. Building
from the command line with msbuild works fine though.

The armv7-win32-vs12 target was added as part of a638bdf4 even
though actual use of it hadn't been tested.

Change-Id: Iee8088252cf790317aeb6b417d29058225f1f629