]>
granicus.if.org Git - libvpx/log
Jingning Han [Tue, 3 Jun 2014 23:43:26 +0000 (16:43 -0700)]
Merge "Rework unit test for 8x8 transformation"
Jingning Han [Mon, 2 Jun 2014 23:40:01 +0000 (16:40 -0700)]
Rework unit test for 8x8 transformation
This commit reworks the unit test for 8x8 forward/inverse
transformation. It adds extreme input value test to detect overflow
issues in the intermediate steps.
It temporarily disables unit test for the SSSE3 version, which
showed overflow failure in the new test conditions.
Change-Id: I7caf10bba4b6db031add65d8c0eb99426b38aa42
Dmitry Kovalev [Tue, 3 Jun 2014 20:56:46 +0000 (13:56 -0700)]
Adding buffer levels to RATE_CONTROL struct.
Change-Id: Ib35ff854378764dc3c6745844c67a33dee545663
Dmitry Kovalev [Tue, 3 Jun 2014 17:22:35 +0000 (10:22 -0700)]
Merge "Cleaning up full_pixel_search()."
Dmitry Kovalev [Tue, 3 Jun 2014 17:05:38 +0000 (10:05 -0700)]
Merge "Moving first pass related functions to vp9_firstpasss.c."
Dmitry Kovalev [Tue, 3 Jun 2014 17:04:27 +0000 (10:04 -0700)]
Merge "Reusing existing vp9_get{8x8, 16x16}var() instead of new ones."
Paul Wilkins [Tue, 3 Jun 2014 12:03:49 +0000 (13:03 +0100)]
Fix AQ mode 2 bug where delta causes Q 0.
In Aq mode 2 for kf/arf/gf the segment q delta
is calculated and then applied by re-quantization without
going through the rd loop again. If the base Q != 0
but the segment Q == 0 (lossless) this can could give rise
to a situation where we have an illegal combination of
transform size and Q. (Q == 0 requires that all blocks
are coded 4x4 WHT).
Change-Id: I241a58c6494ed442e9e4630070b0cde0fb99ae45
Deb Mukherjee [Tue, 3 Jun 2014 05:39:17 +0000 (22:39 -0700)]
Merge "Remove Wextra warnings from vp9_sad.c"
Alex Converse [Tue, 3 Jun 2014 01:50:40 +0000 (18:50 -0700)]
Merge "Remove an attempt to handle SEG_LVL_SKIP sub8x8."
Deb Mukherjee [Tue, 13 May 2014 17:11:42 +0000 (10:11 -0700)]
Remove Wextra warnings from vp9_sad.c
As a side-effect, the sad unit tests for VP8 and VP9
had to be separated.
Fixes a bug in original patch:
(https://gerrit.chromium.org/gerrit/#/c/70163/8)
that was reverted due to a nightly test failure.
Change-Id: Ia2a4e9e278fd3c89d6c3c82fcc6381320ec2a8a6
Dmitry Kovalev [Mon, 2 Jun 2014 18:14:12 +0000 (11:14 -0700)]
Fixing failed ARM build.
Change-Id: I3f74418f07c2dfdd7725a5b4a8ef5c5f4aca6289
Tim Kopp [Mon, 2 Jun 2014 18:08:19 +0000 (11:08 -0700)]
Merge "Fixed OUTPUT_YUV_SRC behavior for VP8"
Yaowu Xu [Mon, 2 Jun 2014 16:02:32 +0000 (09:02 -0700)]
Merge "seeing a 10x slowing down, revert now for investigation"
Yaowu Xu [Mon, 2 Jun 2014 16:02:19 +0000 (09:02 -0700)]
seeing a 10x slowing down, revert now for investigation
Revert "Fix a problem of using an uninitialized parameter"
This reverts commit
538af7db5ffe2ab560511a3bc0ae74dae9316c14
Change-Id: I071aa9b7068ef515abb8ae9584df15067706ccb5
Frank Galligan [Sun, 1 Jun 2014 23:58:11 +0000 (16:58 -0700)]
Merge "Revert "Remove Wextra warnings from vp9_sad.c""
Frank Galligan [Sun, 1 Jun 2014 23:20:26 +0000 (16:20 -0700)]
Revert "Remove Wextra warnings from vp9_sad.c"
This reverts commit
916550428db803c54c993ff9d3c34b9b0bcebb7c
Change-Id: I500822b03f09c64ff6ec5396c68edee9ca3b75cb
Dmitry Kovalev [Sat, 31 May 2014 07:08:45 +0000 (00:08 -0700)]
Merge "Converting disable_inter_mode_mask to inter_mode_mask."
Jingning Han [Fri, 30 May 2014 22:53:17 +0000 (15:53 -0700)]
Merge "Add overflow check unit test for 16x16 inverse DCT/ADST transform"
Jingning Han [Fri, 30 May 2014 22:52:53 +0000 (15:52 -0700)]
Merge "Fix a potential overflow issue in inverse 16x16 full 2D-DCT"
James Zern [Fri, 30 May 2014 22:09:43 +0000 (15:09 -0700)]
Merge "Fix unused variable warning."
hkuang [Fri, 30 May 2014 20:38:36 +0000 (13:38 -0700)]
Merge "Refactor the vp9_get_frame code for frame parallel."
Jingning Han [Thu, 29 May 2014 19:50:54 +0000 (12:50 -0700)]
Add overflow check unit test for 16x16 inverse DCT/ADST transform
This commit applies quantization process with coarse quantization
step size to the forward transform coefficients and tests all the
inverse 16x16 DCT and ADST implementation versions with the
dequantized coefficients as input, to verify that the outcomes
match the prototype.
Change-Id: I68034a6126b45192c87d8c642155290e89bff8fa
Yaowu Xu [Fri, 30 May 2014 18:37:04 +0000 (11:37 -0700)]
Merge "Fix a problem of using an uninitialized parameter"
Dmitry Kovalev [Fri, 30 May 2014 18:24:25 +0000 (11:24 -0700)]
Merge "Removing unused ref_frame_mask local var."
hkuang [Fri, 23 May 2014 22:18:41 +0000 (15:18 -0700)]
Refactor the vp9_get_frame code for frame parallel.
In frame parallel decoding mode, there will be still several frames inside
the decoder when application stop calling vpx_codec_decode to decode frames.
The application will need to keep calling vpx_codec_get_frame to get all the
remaining decoded frames in the decoder.
Change-Id: I2ce8260a91282f045bb9a6093ff8a606b1990f14
Yaowu Xu [Fri, 30 May 2014 17:33:29 +0000 (10:33 -0700)]
Merge "Do not export non-existant symbols"
Christian Duvivier [Fri, 30 May 2014 17:03:46 +0000 (10:03 -0700)]
Fix unused variable warning.
Change-Id: I52f73943194039697da6880d2da54821b2498336
Yaowu Xu [Fri, 30 May 2014 17:15:30 +0000 (10:15 -0700)]
Fix a problem of using an uninitialized parameter
This commit added a call to set speed feature before initializing
motion search, fixed the problem where unintialized search method
is used before its value being set.
Change-Id: I537e4612bf0d00fd6f51396fd222d4b3bd6fde58
Tim Kopp [Fri, 30 May 2014 15:46:08 +0000 (08:46 -0700)]
Fixed OUTPUT_YUV_SRC behavior for VP8
By enabling the OUTPUT_YUV_SRC compiler flag, the encoder will write the raw
input to bd.yuv.
The functionality was mostly implemented, but in its previous state did not
compile.
Change-Id: Ia331ad0f4c6e6f9f51e8d42cd33ba8cc146b3dbf
Paul Wilkins [Fri, 30 May 2014 13:14:06 +0000 (06:14 -0700)]
Merge "Re-factor some duplicate code."
Sean McGovern [Thu, 29 May 2014 02:16:13 +0000 (22:16 -0400)]
Do not export non-existant symbols
Change-Id: Ic0fb8e5016d064c2227cf5d87ffba86a4303c82b
Dmitry Kovalev [Thu, 29 May 2014 22:03:02 +0000 (15:03 -0700)]
Removing unused ref_frame_mask local var.
Change-Id: Ie11558c076a0161cc9608788e050b1b16e31c490
Dmitry Kovalev [Thu, 29 May 2014 20:23:31 +0000 (13:23 -0700)]
Merge "Consistent names for intra mask flags."
Alex Converse [Thu, 29 May 2014 19:46:20 +0000 (12:46 -0700)]
Merge "Don't update encoder skip count for SEG_LVL_SKIP."
Dmitry Kovalev [Thu, 29 May 2014 19:25:41 +0000 (12:25 -0700)]
Converting disable_inter_mode_mask to inter_mode_mask.
Making this consistent with intra mode masks: you need to specify
allowed inter/intra modes to use.
Change-Id: Iaecd28bf79047259707d8e7a59a57bb7b856383e
Dmitry Kovalev [Thu, 29 May 2014 19:11:02 +0000 (12:11 -0700)]
Consistent names for intra mask flags.
Change-Id: Ibdd5255d37200fb8a1d50f71a2a49c6089ae21e7
Alex Converse [Thu, 29 May 2014 18:27:07 +0000 (11:27 -0700)]
Remove an attempt to handle SEG_LVL_SKIP sub8x8.
SEG_LEVEL_SKIP requires the block size to be at least 8x8. Attempting to
use it on smaller partitions causes the decoder to reject the bitstream.
Change-Id: Ia7188cdf8ae5ac1df6bd29f3f80dbb0610e1f7b1
Dmitry Kovalev [Thu, 29 May 2014 18:58:42 +0000 (11:58 -0700)]
Merge "Making speed checks consistent in set_rt_speed_feature()."
Jingning Han [Thu, 29 May 2014 18:41:55 +0000 (11:41 -0700)]
Fix a potential overflow issue in inverse 16x16 full 2D-DCT
An overflow issue could potentially happen in the second round 1-D
transform of the SSSE3 full inverse 16x16 2D-DCT. This commit fixes
this issue.
Change-Id: Ia19e4888fda1cc929a28a5f89a5beec612d628dc
Alex Converse [Thu, 29 May 2014 01:44:17 +0000 (18:44 -0700)]
Don't update encoder skip count for SEG_LVL_SKIP.
This aligns the encoder behavior with the decoder.
Change-Id: Ifa0840e4b07b19309e0bf1d1182498883249ec45
Dmitry Kovalev [Thu, 29 May 2014 18:16:39 +0000 (11:16 -0700)]
Merge "Moving itxm_add pointer from MACROBLOCKD to MACROBLOCK."
Dmitry Kovalev [Wed, 28 May 2014 21:07:13 +0000 (14:07 -0700)]
Reusing existing vp9_get{8x8, 16x16}var() instead of new ones.
Change-Id: I87b7c657d8813d7fb383ab519d150c0ffb1dd377
Dmitry Kovalev [Thu, 29 May 2014 18:11:50 +0000 (11:11 -0700)]
Making speed checks consistent in set_rt_speed_feature().
Change-Id: Id3d0a49836fe996b806707d29a8130acf9d7ea0e
Yaowu Xu [Thu, 29 May 2014 16:09:58 +0000 (09:09 -0700)]
Merge "Fixing -Wextra warnings in vp9_{cx, dx}_iface.c."
Yaowu Xu [Thu, 29 May 2014 15:16:19 +0000 (08:16 -0700)]
Merge "Fixed a crash windows build"
Yaowu Xu [Wed, 28 May 2014 21:17:46 +0000 (14:17 -0700)]
Fixed a crash windows build
Change-Id: I58baa1da1f3bfc8a6da454399139fe6a7473ff10
Marco Paniconi [Wed, 28 May 2014 19:32:44 +0000 (12:32 -0700)]
Merge "vp8 denoiser: fix to zero_mv mode selection."
Marco Paniconi [Tue, 27 May 2014 23:44:17 +0000 (16:44 -0700)]
vp8 denoiser: fix to zero_mv mode selection.
In the current logic, if the sse for zero motion is smaller
than the sse for new_mv (i.e., best_sse), we may still end up
using the non-zero mv for denoising (if the magnitude of new_mv is above threshold).
This can happen for very noisy content, and can lead to artifacts.
This change ensures that we always use zero_mv (over new_mv) for
denoisng if sse_zero_mv <= best_sse.
Change-Id: I8ef9294d837b077013b77a46c9a71d17c648b48a
Dmitry Kovalev [Wed, 28 May 2014 18:15:43 +0000 (11:15 -0700)]
Fixing -Wextra warnings in vp9_{cx, dx}_iface.c.
Change-Id: I0abad32551dc534d3db27424c118e4b2f6b50f37
Dmitry Kovalev [Wed, 28 May 2014 17:57:40 +0000 (10:57 -0700)]
Merge "Using 2 instead of 3 elements for last_q array."
Dmitry Kovalev [Wed, 28 May 2014 17:55:12 +0000 (10:55 -0700)]
Merge "Removing redundant vp9_zero() call."
Dmitry Kovalev [Wed, 28 May 2014 17:50:46 +0000 (10:50 -0700)]
Merge "Cleaning up vp9_variance_sse2.c."
Alex Converse [Wed, 28 May 2014 17:19:49 +0000 (10:19 -0700)]
Merge "Always allow ZEROMV when SEG_LVL_SKIP is on."
Alex Converse [Wed, 28 May 2014 17:19:18 +0000 (10:19 -0700)]
Merge "Fix the all intra modes mask constant."
Paul Wilkins [Wed, 28 May 2014 15:07:50 +0000 (08:07 -0700)]
Merge "Removing this_frame_stats member from TWO_PASS struct."
Paul Wilkins [Tue, 27 May 2014 14:40:33 +0000 (15:40 +0100)]
Re-factor some duplicate code.
Change-Id: I89a1dbea39c50c7633f746d9c93fec3a289f1b42
Paul Wilkins [Wed, 28 May 2014 09:04:29 +0000 (02:04 -0700)]
Merge "Remove brightness weighting in two pass."
Deb Mukherjee [Wed, 28 May 2014 06:24:06 +0000 (23:24 -0700)]
Removing undeclared identifier - build fix
Fixes build with --enable-internal-stats
Change-Id: I137169c859f561478e45891defe976d595454166
Dmitry Kovalev [Wed, 28 May 2014 01:38:57 +0000 (18:38 -0700)]
Cleaning up full_pixel_search().
Change-Id: Ie517ac06385133ffb3bbc449d9f23240f245976d
Dmitry Kovalev [Wed, 28 May 2014 01:19:19 +0000 (18:19 -0700)]
Using 2 instead of 3 elements for last_q array.
Change-Id: I2c6950e7d79fc89c6f97e6dcf47317ef66c453a5
Alex Converse [Wed, 28 May 2014 01:14:14 +0000 (18:14 -0700)]
Always allow ZEROMV when SEG_LVL_SKIP is on.
Change-Id: I6db1dc82f66438ac48f571d2f1a2ac7c39a97a1a
Alex Converse [Wed, 28 May 2014 01:11:21 +0000 (18:11 -0700)]
Fix the all intra modes mask constant.
The new constant expands to 0x3fc00808.
Change-Id: Ib5109e4faf035fe0402b59f8a8d2e412628b9276
Dmitry Kovalev [Wed, 28 May 2014 00:35:38 +0000 (17:35 -0700)]
Merge "Removing ctrl_id parameter from vpx_codec_control_fn_t."
Dmitry Kovalev [Wed, 28 May 2014 00:09:28 +0000 (17:09 -0700)]
Removing this_frame_stats member from TWO_PASS struct.
Change-Id: Id8877fad1f1e88b145e7c40c43174109b9c4f373
Jingning Han [Tue, 27 May 2014 23:58:00 +0000 (16:58 -0700)]
Merge "Fix compiling error in MSVS"
Dmitry Kovalev [Tue, 27 May 2014 23:45:58 +0000 (16:45 -0700)]
Removing ctrl_id parameter from vpx_codec_control_fn_t.
Change-Id: I2b61c8c17ded1074dea92b4f6ad9be84d128b52a
Dmitry Kovalev [Tue, 27 May 2014 23:24:01 +0000 (16:24 -0700)]
Removing redundant vp9_zero() call.
rd.tx_select_threshes is cleared in encode_frame_internal().
Change-Id: Ie03776a41c585f13b392a9b62d4e91ef26ebeaf0
Jingning Han [Tue, 27 May 2014 18:30:14 +0000 (11:30 -0700)]
Fix compiling error in MSVS
Need to include math.h before tmmintrin.h in some versions of MSVS.
Change-Id: Ia6b83ae599316887ecf30c4e4b9e4355fb8a4219
Yaowu Xu [Tue, 27 May 2014 22:52:50 +0000 (15:52 -0700)]
Merge "vp9_rdopt.c: Removed 2 unused parameters"
Marco Paniconi [Tue, 27 May 2014 22:32:16 +0000 (15:32 -0700)]
Merge "Fix to reduce block artifacts from vp8 temporal denoiser."
Dmitry Kovalev [Tue, 27 May 2014 22:02:21 +0000 (15:02 -0700)]
Merge "Converting target_bandwidth to Bit/s at very beginning."
Yaowu Xu [Tue, 27 May 2014 21:53:49 +0000 (14:53 -0700)]
Merge "Revert "Making vp9_get_sse_sum_{8x8, 16x16} static.""
Yaowu Xu [Tue, 27 May 2014 21:45:19 +0000 (14:45 -0700)]
vp9_rdopt.c: Removed 2 unused parameters
Change-Id: I935ec0e78570ce3d3585f972350e39043eefa30a
Erik Niemeyer [Tue, 27 May 2014 21:00:39 +0000 (14:00 -0700)]
Merge "Turn on unit tests for AVX2 convolve functions"
Dmitry Kovalev [Tue, 27 May 2014 20:53:19 +0000 (13:53 -0700)]
Cleaning up vp9_variance_sse2.c.
Change-Id: I5ec336848f6489c31cf2b645026fa2025db07466
Yunqing Wang [Tue, 27 May 2014 20:26:40 +0000 (13:26 -0700)]
Revert "Making vp9_get_sse_sum_{8x8, 16x16} static."
This reverts commit
e8bbb3d9db797dab7c2f947cc43e8d0f168e4953 .
Change-Id: Ie368d36fd249d323d859d208609c711f04537bbc
Deb Mukherjee [Tue, 27 May 2014 18:54:05 +0000 (11:54 -0700)]
Merge "Remove Wextra warnings from vp9_sad.c"
Yunqing Wang [Tue, 27 May 2014 17:52:16 +0000 (10:52 -0700)]
Merge "Fix decoder mismatch in sub-pixel AVX2 intrinsic filters"
Dmitry Kovalev [Tue, 27 May 2014 17:50:55 +0000 (10:50 -0700)]
Merge "Reusing rd_less_than_thresh() function."
Dmitry Kovalev [Tue, 27 May 2014 17:50:31 +0000 (10:50 -0700)]
Merge "Removing vp8/common/pragmas.h."
Yunqing Wang [Tue, 27 May 2014 17:36:56 +0000 (10:36 -0700)]
Turn on unit tests for AVX2 convolve functions
This patch turned on unit tests for AVX2 convolve functions.
Change-Id: I51b8bfdaa290fb22862c68af61abf2394d00d47c
Scott LaVarnway [Tue, 27 May 2014 16:01:48 +0000 (09:01 -0700)]
Merge "neon matches "C" when using increase_denoising"
Paul Wilkins [Fri, 23 May 2014 18:01:43 +0000 (19:01 +0100)]
Remove brightness weighting in two pass.
This code dates from the ancient past and
applied an error score weighting based on pixel
brightness. This not seem to be providing any
benefit metrics wise and could be making some
visual issues in dark frames worse.
The field is left in place in the FIRSTPASS_STATS data
structure in this patch, pending changes to unit tests that
use a pre-defined first pass file.
Change-Id: Id50f04205230234858e7548ce523f11acaf3567d
Scott LaVarnway [Mon, 26 May 2014 15:06:25 +0000 (08:06 -0700)]
neon matches "C" when using increase_denoising
If increase_denoising is set,
vp8_denoiser_filter_neon() produced incorrect results.
Change-Id: I645f78e48b8f6657fa8a4b69d2c4d3488a0581dc
Paul Wilkins [Sun, 25 May 2014 21:48:36 +0000 (14:48 -0700)]
Merge "Further first pass allocation changes."
Paul Wilkins [Sun, 25 May 2014 21:47:35 +0000 (14:47 -0700)]
Merge "Re-factor bit allocation in first pass."
Dmitry Kovalev [Sat, 24 May 2014 01:11:07 +0000 (18:11 -0700)]
Converting target_bandwidth to Bit/s at very beginning.
Change-Id: I1d8c9fe4228e2f1ef67a66883694842a9545e7b9
levytamar82 [Fri, 23 May 2014 23:42:10 +0000 (16:42 -0700)]
Fix decoder mismatch in sub-pixel AVX2 intrinsic filters
The subpixel SSSE3 was fixed in this patch:
https://gerrit.chromium.org/gerrit/#/c/70283/
So the equivalent AVX2 is fixed accordingly.
Change-Id: Ieebbc1949c99d34b12b8b47692df71aca5001f3a
Jingning Han [Fri, 23 May 2014 23:02:09 +0000 (16:02 -0700)]
Merge changes I4f0f4378,I02250d13
* changes:
Re-enable unit test on neon implementation of 16x16 idct
Turn on unit test for inverse 16x16 2D-DCT ssse3
Jingning Han [Fri, 23 May 2014 23:01:22 +0000 (16:01 -0700)]
Merge "Inverse 16x16 2D-DCT SSSE3 implementation"
Jingning Han [Thu, 22 May 2014 18:36:00 +0000 (11:36 -0700)]
Re-enable unit test on neon implementation of 16x16 idct
The previous change only tunes forward transform. It doesn't affect
the neon implementation of the inverse transform. Hence turn the
unit test on.
Change-Id: I4f0f43783b98814d1eee53182209f9669d538140
Jingning Han [Thu, 22 May 2014 18:31:43 +0000 (11:31 -0700)]
Turn on unit test for inverse 16x16 2D-DCT ssse3
Change-Id: I02250d1364ca2e637585a8db020cba9fd4a95664
Jingning Han [Thu, 22 May 2014 16:44:40 +0000 (09:44 -0700)]
Inverse 16x16 2D-DCT SSSE3 implementation
This commit enables the SSSE3 implementation of full inverse 16x16
2D-DCT. The unit runtime goes down from 1642 cycles to 1519 cycles,
about 7% speed-up.
Change-Id: I14d2fdf9da1fb4ed1e5db7ce24f77a1bfc8ea90d
Yaowu Xu [Fri, 23 May 2014 21:45:30 +0000 (14:45 -0700)]
Merge "Use extreme values for input in convovle tests"
Yunqing Wang [Fri, 23 May 2014 21:24:48 +0000 (14:24 -0700)]
Merge "Fix decoder mismatch in sub-pixel SSSE3 intrinsic filters"
Yaowu Xu [Fri, 23 May 2014 19:23:29 +0000 (12:23 -0700)]
Use extreme values for input in convovle tests
The intepolation filter functions can be better tested withe extreme
values, especially given the optimization functions are prone to
overflow signed 16 bit intermediate value when operation order is
wrong.
Change-Id: I712142b0bc1e5969c692c0486a57ffa37c9742b5
Dmitry Kovalev [Fri, 23 May 2014 20:03:15 +0000 (13:03 -0700)]
Removing vp8/common/pragmas.h.
Change-Id: I80630a7350e884ebc4fef73fb5b52ec25f908523
Dmitry Kovalev [Fri, 23 May 2014 19:58:00 +0000 (12:58 -0700)]
Merge "Removing vp9_pragmas.h."
Paul Wilkins [Wed, 21 May 2014 12:17:00 +0000 (13:17 +0100)]
Further first pass allocation changes.
Further changes to first pass allocation for gf/arf groups.
Three variables removed from TWO_PASS structure as only
now used locally. Dont adjust gf_group_bits in the post
encode update as this will no longer have any effect.
Change-Id: Iff89b225db923fc856f5d2aedbc899f1d7d68b55
Yunqing Wang [Fri, 23 May 2014 18:52:20 +0000 (11:52 -0700)]
Fix decoder mismatch in sub-pixel SSSE3 intrinsic filters
In 8-tap filtering, to guarantee the intermediate results fit in
16 bits, the order of accumulating the products needs to be done
correctly, and the largest product should be added last. This
patch fixed the problem using the method in commit "Correct ssse3
8/16-pixel wide sub-pixel filter calculation".
Change-Id: I79d0ad60c057b15011ece84cda9648eee0809423