17183 Commits

Author SHA1 Message Date
Geza Lore
7de2ba3eae Fix false uninitialized warnings (GCC 5+).
Change-Id: Ia00c754ddaf22bb7f1dfcd20106db6293bf4b070
2016-06-21 12:54:17 +01:00
Jingning Han
5223a4b405 Handle two identical states in the trellis chain
When the next two states are identical, skip repeated cost table
fetch and multiplication operations. This makes the trellis unit
about 5% faster.

Change-Id: I0dbf7ad0a5732044e4e45dd59e9431a251c678f2
2016-06-20 16:59:28 -07:00
Yue Chen
474ea305ea Merge "Fix RDO issue of obmc + speed feature fast_inter_tx_type_search" into nextgenv2 2016-06-20 21:57:24 +00:00
Yaowu Xu
b4aa5c9408 Merge "Make type conversion explicit" 2016-06-20 21:34:41 +00:00
Yi Luo
f1a50db2d1 Merge "Convolution horizontal filter SSSE3 optimization" into nextgenv2 2016-06-20 20:06:02 +00:00
Yaowu Xu
aee577b691 Merge "Fix ubsan warning: vp9/encoder/vp9_encodeframe.c" 2016-06-20 19:33:33 +00:00
Yaowu Xu
543ea3eb3e Make type conversion explicit
This fixes MSVC warnings.

Change-Id: I675d8486230b2b74d7973d95720a4995c4750282
2016-06-20 12:05:29 -07:00
Yi Luo
229690a95c Convolution horizontal filter SSSE3 optimization
- Apply signal direction/4-pixel vertical/8-pixel vertical
  parallelism.
- Add unit test to verify the bit exact result.
- Overall encoding time improves ~24% on Xeon E5-2680 CPU.

Change-Id: I104dcbfd43451476fee1f94cd16ca5f965878e59
2016-06-20 11:10:30 -07:00
Jingning Han
6929ab0ba6 Merge "Refactor optimize_b for speed performance" 2016-06-20 16:48:22 +00:00
Jingning Han
85786a8ddb Merge "Port optimize_b speed-up from vp10" 2016-06-20 16:48:02 +00:00
Jingning Han
a5bcf03030 Merge "Use precise rate estimate for zero_token" into nextgenv2 2016-06-20 16:47:33 +00:00
Jingning Han
899a989d3a Merge "Optimize the use case of token_cost table" into nextgenv2 2016-06-20 16:47:20 +00:00
Debargha Mukherjee
dc5431ad4b Merge "Turn on AqSegment tests for VP10" into nextgenv2 2016-06-20 16:47:13 +00:00
Yue Chen
1273c39c03 Fix RDO issue of obmc + speed feature fast_inter_tx_type_search
Change-Id: I86a967ad2d824ca7877626eed9eb11f0e057b22d
2016-06-20 16:38:12 +00:00
Yue Chen
b37c279ab5 Merge "Re-enable ActiveMapTest for VP10" into nextgenv2 2016-06-20 16:37:53 +00:00
Johann Koenig
188420f4fd Merge "Configuration: Use 'is_in' instead of echo | grep" 2016-06-19 23:51:33 +00:00
Johann
7a3d48e56f Fix iosbuild.sh: outdated reference to ios-version.sh
File was removed in b29ebfe91c643b4bab1f24c2b55bf0a5a83cb825 but a
reference was missed.

Change-Id: I7a3d81ab196a70cb5691331705f956cc7dc9bb2a
2016-06-19 16:49:00 -07:00
Jingning Han
86b7d39a83 Merge "Fix unit test failure in obmc exp" into nextgenv2 2016-06-18 22:38:53 +00:00
James Zern
4d9e876b44 realtime_test: remove decoded frame count check
decoding is done if the decoder is available, with errors handled
accordingly. the encoded frame count should be sufficient for this test.

+ remove HandleDecodeResult() as it's redundant given the base
  implementation

BUG=webm:1233

Change-Id: I513c1c3475c58a746f4df627491bdc392fe21416
2016-06-18 11:05:55 -07:00
James Zern
d0de89a12a remove vpx_highbd_1[02]_sub_pixel_variance4x4_sse4_1
these cause ASan errors VP10/EndToEndTestLarge.EndtoEndPSNRTest

BUG=webm:1242

Change-Id: I0334e3b255b14e18f61970c3721ae748dc79727b
2016-06-17 19:46:20 -07:00
Jingning Han
887f020691 Fix unit test failure in obmc exp
Properly restore the rate cost in the inner search loop of obmc
prediction. This avoids unexpected encoding behavior. It fixes
the unit test failure in obmc experiment:

AltRefForcedKeyTestLarge.Frame1IsKey/2

Change-Id: I667b219dfcf2f2c63d9d984900ed3cfd10c354bd
2016-06-17 17:44:03 -07:00
Jingning Han
9e185ed177 Refactor optimize_b for speed performance
This commit refactors the trellis coefficient optimization process.
It saves multiplications used to generate the final dequantized
coefficients. It removes two memset operations on quantized
and dequantized coefficient sets. This improves the unit speed
by 10%.

Change-Id: I23f47c6e14582520a7f952f03ce8f72183e7f0e6
2016-06-17 17:41:09 -07:00
Jingning Han
dba1d1a63d Port optimize_b speed-up from vp10
This commit back ports the speed-up from vp10. It improves the
unit speed by 15%.

Change-Id: Ibe8c0e0974b03266d6abd16a41e89c3b91d8db2a
2016-06-17 17:41:05 -07:00
Yue Chen
0c73623d2c Merge "Make variance based partitioning compatible with SEG_LVL_SKIP" into nextgenv2 2016-06-18 00:19:42 +00:00
Yaowu Xu
40477f1f70 Fix ubsan warning: vp9/encoder/vp9_encodeframe.c
BUG=webm:1219

Change-Id: I29b896c7f61abf871d7853b1d8200aa5a9f6a95c
2016-06-17 17:07:06 -07:00
Jingning Han
1417ed28f3 Merge "Use 64-bit integer to store distortion in optimize_b" 2016-06-17 23:05:33 +00:00
Jingning Han
f99f78c7af Use 64-bit integer to store distortion in optimize_b
This fixes the overflow issue.

Bug=webm:1241

Change-Id: Ia168b7fae1ad214a6837aaa785a08bf8506987dd
2016-06-17 15:07:00 -07:00
DO NOT USE
4f367f5cd9 Merge "Parse codec options in order of occurrence." 2016-06-17 22:03:50 +00:00
Jingning Han
8293759056 Merge "Skip restore token_cache value" into nextgenv2 2016-06-17 21:59:53 +00:00
Johann
1de5ba75be Configuration: Use 'is_in' instead of echo | grep
Change-Id: I38f7684969f12a624980f4e06698379a49e61bdf
2016-06-17 14:28:19 -07:00
James Zern
9d7a12a0e7 Merge "Fix warnings from gtest under GCC 5 or newer." 2016-06-17 21:18:21 +00:00
Johann Koenig
7055ca556c Merge "Remove ios-version.sh" 2016-06-17 21:08:51 +00:00
Johann
0146fa95ef Parse codec options in order of occurrence.
Each time a codec is enabled or disabled with the umbrella
--enable-vpN flag, set the encoder and decoder configurations as well.
This was done as a post-processing step but doing that lost the order of
the arguments.

BUG=webm:1205

Change-Id: Ic629bfdd06acc04bc5a7227309f36bba54dad8b1
2016-06-17 14:04:51 -07:00
Geza Lore
dc5ae1e34c Merge "Fix warnings from gtest under GCC 5 or newer." into nextgenv2 2016-06-17 20:10:14 +00:00
Zoe Liu
5805a14ca6 Merge bi-predictive frames to EXT_REFS
This patch removed the experiment of BIDIR_PRED and merged the feature
into the experiment of EXT_REFS:

(1) Each frame now has up to 6 reference frames, namely
    LAST_FRAME, LAST2_FRAME, LAST3_FRAME, GOLDEN_FRAME, (forward) and
    BWDREF_FRAME, ALTREF_FRAME (backward);
    LAST4_FRAME has been removed;
(2) First pass still keeps the 8 updates:
    KF_UPDATE, LF_UPDATE, GF_UPDATE, ARF_UPDATE, OVERLAY_UPDATE, and
    BRF_UPDATE, LAST_BIPRED_UPDATE, BI_PRED_UPDATE;
(3) show_existing_frame==1 is supported in the experiment of EXT_REFS;
(4) New encoding modes are added for both single-ref and compound cases,
    through the use of the 2 extra forward references (LAST2 & LAST3)
    and the 1 extra backward reference (BWDREF).

RD performance wise, using Overall PSNR: Avg/BDRate
        Bipred only      Prev EXT_REFS    Current EXT_REFS with bipred
lowres: -3.474/-3.324    -1.748/-1.586    -4.613/-4.387
derflr: -2.097/-1.353    -1.439/-1.215    -3.120/-2.252
midres: -2.129/-1.901    -1.345/-1.185    -2.898/-2.636

If in vp10/encoder/firstpass.h, change BFG_INTERVAL from 2 to 3, i.e. to
use 2 bi-predictive frames than 1, a further improvement may be
obtained:
                 Current EXT_REFS with bipred
        1 bi-predictive frame    2 bi-predictive frames
lowres: -4.613/-4.387            -4.675/-4.465
derflr: -3.120/-2.252            -3.333/-2.516
midres: -2.898/-2.636            -3.406/-3.095

Change-Id: Ib06fe9ea0a5cfd7418a1d79b978ee9d80bf191cb
2016-06-17 12:43:39 -07:00
Tom Finegan
bfc75f7382 Disable neon for Xcode versions < v6.3.
vtbl1_u8 is marked unavailable prior to Xcode v6.3.

BUG=webm:1154

Change-Id: I08bd34f712ab7da72658c89d8b606658c1348923
2016-06-17 12:39:14 -07:00
Geza Lore
7172e97abe Re-enable ActiveMapTest for VP10
Change-Id: I030fdde966b9911712eca131d095015afd9b0d8a
2016-06-17 20:33:58 +01:00
Geza Lore
169431b84a Make variance based partitioning compatible with SEG_LVL_SKIP
Inter blocks that have SEG_LVL_SKIP active must be at least 8x8 in
size for bitstream conformance (see read_inter_block_mode_info in
decodemv.c).

This patch makes the variance based partitioning scheme stop at 8x8
blocks in inter frames. This satisfies the SEG_LVL_SKIP constraint
and is more in line with the original implementation of this function
(before it got extended for 128x128 superblocks).

BUG=webm:1234

Change-Id: I1fdd894569a9c0817713a77daabe4c8b8e1d00c0
2016-06-17 20:31:05 +01:00
Yaowu Xu
c5fae00227 Merge "Fix ubsan warnings: vp9/encoder/vp9_encoder.c" 2016-06-17 19:29:19 +00:00
Geza Lore
cdd83a0bb2 Fix warnings from gtest under GCC 5 or newer.
Change-Id: I9661f2fe9d315dccae69caa70d929b5d9d93b7db
(cherry picked from commit 4c83fdd3d7cacd66a2c64451db0fb64e5b2ecf48)
2016-06-17 20:11:16 +01:00
Tom Finegan
5a9f21db54 Output frames in first pass for VPX_DL_REALTIME.
Since combining VPX_DL_REALTIME with VPX_RC_FIRST_PASS is basically
nonsense, ignore the user's pass setting when this happens and
behave as if the requested encode is a single pass encode.

BUG=webm:1233

Change-Id: I5ee4c4e5838c4ca6d24988890aae490b10826db2
2016-06-17 11:25:55 -07:00
Yaowu Xu
754eb95105 Fix ubsan warnings: vp9/encoder/vp9_encoder.c
BUG=webm:1219

Change-Id: I433a688fbbc87094afa888c1e7d49f3eefc0e765
2016-06-17 11:25:11 -07:00
Jingning Han
019b750867 Use precise rate estimate for zero_token
This commit takes the precise rate estimate for zero_token rate
cost update. It improves the compression performance:

lowres 0.15%
midres 0.23%

Change-Id: I36761079f75ce43c814f8c663667e359d4ac2cd4
2016-06-17 10:57:30 -07:00
Jingning Han
90ea281f29 Optimize the use case of token_cost table
Reduce the cache footprint of the token_costs table.

Change-Id: Ie989e60c6479ac3251cadaac9c7e795ccba52f4e
2016-06-17 10:15:34 -07:00
Geza Lore
4c83fdd3d7 Fix warnings from gtest under GCC 5 or newer.
Change-Id: I9661f2fe9d315dccae69caa70d929b5d9d93b7db
2016-06-17 15:36:06 +01:00
Johann Koenig
a724477d69 Merge "vp8: correct function return types" 2016-06-17 04:19:37 +00:00
Johann Koenig
e9b6172476 Merge "VP8_COMP: make frames_since_golden signed" 2016-06-17 04:19:17 +00:00
James Zern
cf2bfcd9d8 Merge "vp9_pickmode: quiet int64->int conversion warning" 2016-06-17 03:58:46 +00:00
Jingning Han
019dbb4cdc Merge "Rework table access operations in vp10_optimize_b function" into nextgenv2 2016-06-17 00:25:50 +00:00
Johann
b29ebfe91c Remove ios-version.sh
The logic can be incorporated into configure.sh

Removes a dependency on ios-version.sh which was not part of DIST-SRCS
and removes a warning from 'make dist' sub builds:

../src/build/make/configure.sh: line 787:
../src/build/make/ios-version.sh: No such file or directory

Change-Id: Ic38314708eb278dd9d2a9769a670da32f6126637
2016-06-16 16:37:45 -07:00