Commit Graph

63 Commits

Author SHA1 Message Date
Yunqing Wang
10a497bd38 Make the row based multi-threaded encoder deterministic
This patch followed allow_exhaustive_searches feature modification and
continued to modify the encoder to achieve the determinism in the row
based multi-threaded encoding. While row-mt = 1 and using multiple
threads, the adaptive feature in encoder was disabled, which gave
BDRate gain(at speed 1, -0.6% ~ -0.7%; at speed 2, -0.46% ~ -0.59%),
but some encoder speed losses(7% ~ 10% at speed 1 and 3% ~ 6% at
speed 2). These speed losses were acceptable considering the speed
gains obtained from row-mt.

Change-Id: I60d87a25346ebc487a864b57d559f560b7e398bb
2017-04-24 16:28:27 -07:00
Gabriel Marin
976ddb61d3 Add a vector form of routine vp9_model_rd_from_var_lapndz
Add routine vp9_model_rd_from_var_lapndz_vec and call it from model_rd_for_sb
to model the rate and distortion for MAX_MB_PLANE Laplacian sources in
parallel. The caller ensures that all sources have non-zero variance.

Measured a 18% to 25% reduction in retired instructions, and 17% to 24%
reduction in instruction execution cost with different compilers for the
Laplacian modeling.

No change in behavior.

TEST=Verified that encoded files match bit for bit, with and without this
change.
BUG=b/33678225

Change-Id: I6b76947f21c659a349adb896e13e99f6e3f951e6
2017-03-16 22:19:44 +00:00
Ranjit Kumar Tulabandu
71061e9332 Row based multi-threading of encoding stage
(Yunqing Wang)
This patch implements the row-based multi-threading within tiles in
the encoding pass, and substantially speeds up the multi-threaded
encoder in VP9.

Speed tests at speed 1 on STDHD(using 4 tiles) set show that the
average speedups of the encoding pass(second pass in the 2-pass
encoding) is 7% while using 2 threads, 16% while using 4 threads,
85% while using 8 threads, and 116% while using 16 threads.

Change-Id: I12e41dbc171951958af9e6d098efd6e2c82827de
2017-02-15 00:49:34 +00:00
Ranjit Kumar Tulabandu
75d2443bf0 Initialize errorperbit and sabperbit in ARNR filtering
(Yunqing)
This patch added the missing initialization in temporal filter.
Borg test BDRate results:
PSNR: -0.019%(lowres); -0.013%(hdres);
SSIM: -0.001%(lowres); -0.010%(hdres).
Other q values gave comparable but no better results.

Change-Id: I7ad0c18b39e6f558342688e2fe1e12fdb133ce9b
2017-01-24 08:58:17 -08:00
clang-format
e0cc52db3f vp9/encoder: apply clang-format
Change-Id: I45d9fb4013f50766b24363a86365e8063e8954c2
2016-08-02 16:47:11 -07:00
Scott LaVarnway
c969b2b02b VP9: get_pred_context_switchable_interp() -- encoder side
Change-Id: I7217c90d5cf38c51b76759a2dc4f10070f3a40ac
2016-07-21 11:47:51 -07:00
Yury Gitman
67611119b5 cosmetics: Beautify whitespaces and line wrapping
Change-Id: I9afa02cae671bd3527cf344695e53d0cc767f549
2016-06-24 10:18:06 -07:00
Alex Converse
09f9c5d7f9 Better workaround for Bug 1089.
Don't initialize first pass costs for a number of symbols where first
pass probabilities aren't initialized.

This brings a 1.22x first pass speedup.

https://bugs.chromium.org/p/webm/issues/detail?id=1089

Change-Id: I97438c357bd88f52f5a15c697031cf0c3cc8f510
2016-02-17 14:46:26 -08:00
Alex Converse
fac947df77 Restore previous motion search bit-error scale.
The bit to error transformation got doubled as a result of going from
8-bit to 9-bit costs (change d13385c).

Use defines to derive the scale numbers and comment some of the fields.

derf: -0.023 BDRATE
hevcmr: +0.067 BDRATE
stdhd: +0.098 BDRATE
(These are substantially smaller than than the original gains from 8 to
9 bit costing.)

Change-Id: I6a2b3b029b2f1415e4f90a05709b2333ec0eea9b
2016-02-09 13:20:25 -08:00
Alex Converse
4326cffa65 Merge "Tie the bit cost scale to a define." 2016-01-21 19:17:56 +00:00
Scott LaVarnway
5232326716 VP9: Eliminate MB_MODE_INFO
Change-Id: Ifa607dd2bb366ce09fa16dfcad3cc45a2440c185
2016-01-19 16:40:20 -08:00
Alex Converse
269428e35c Tie the bit cost scale to a define.
This is a pure-refactor in preparation to potentially raise the bit-cost
resolution.

Verified at good speed 0 and rt speed -6.

Change-Id: I5347e6e8c28a9ad9dd0aae1d76a3d0f3c2335bb9
2016-01-15 15:59:31 -08:00
hui su
6ab6ac450b Use accurate bit cost for uv_mode in UV intra mode RD selection
On derflr, +0.1% for VP10; however, -0.03% on VP9.

Change-Id: I09c724232ede74254043d61d3cadc506256af0af
2015-11-06 14:45:43 -08:00
Johann
c5f11912ae Include vpx_dsp_common.h when using VPXMIN/MAX
Change-Id: I2e387a06484a06301f3cd6600c4ba2f4335b61ee
2015-08-31 14:36:35 -07:00
James Zern
5e16d397bd vpx_dsp_common: add VPX prefix to MIN/MAX
prevents redeclaration warnings;
vp8 has its own define which will be resolved in a future commit

Change-Id: Ic941fef3dd4262fcdce48b73075fe6b375f11c9c
2015-08-26 20:11:32 -07:00
hui su
697a8e6fe6 Avoid setting rate multiplier as 0
In high bitdepth setting, the rate multipier may be set as 0. In
lossless mode, the RD cost would always be 0, resulting in bad
partition and prediction mode choices.

Change-Id: I297014dd8bfa8a07ff0ab480119f75678300ff68
2015-08-24 11:47:36 -07:00
Alex Converse
a8a08ce57e Move vp9_systemdependent.h to vpx_ports bitops.h and system_state.h
Use system_state.h in vpx_dsp and remove unneeded includes of
vp9_systemdependent.h.

Change-Id: I92557ec6dd5aa790160b4f31fe7967db0d7ec3c4
2015-08-10 15:37:14 -07:00
Yaowu Xu
bf82514b54 vpx_dsp/bitreader.h: vp9_->vpx_
Replace vp9_ in names to vpx_ as they are not codec specific.

Change-Id: I2e583aa63dee769353ada4b42417aa15c4074ebb
2015-07-20 18:06:31 -07:00
Johann
6a82f0d7fb Move sub pixel variance to vpx_dsp
Change-Id: I66bf6720c396c89aa2d1fd26d5d52bf5d5e3dff1
2015-07-07 15:51:04 -07:00
Scott LaVarnway
c06d56cc7d VP9: Move ref_mvs[][] and mode_context[] from MB_MODE_INFO
to MB_MODE_INFO_EXT.  This saves 36 bytes per 8x8 area for
both the decoder and encoder. (encoder has two MODE_INFO
buffers)

Change-Id: If006abb2224acaf326df3c2be09e77e967662107
2015-06-29 12:46:47 -07:00
Scott LaVarnway
b2658ec321 Eliminated frame_type check in get_partition_probs()
Moved the frame_type check to the tile level and stored
the prob ptr in MACROBLOCKD.

Change-Id: I10b5a4abd58213dc7610e3ade1a1583c01526842
2015-06-16 05:37:54 -07:00
James Zern
18b60af27c vp9: correct some function signatures
silences missing prototype warnings

Change-Id: Idaf68d83d2cb03847f3ee002c4d00c2ac79da604
2015-05-15 10:43:47 -07:00
Johann
1d7ccd5325 Relocate memory operations for common code
With the sad functions, and hopefully the variance functions soon,
moving to the vpx_dsp location, place the defines used in the
reference C code in a common location.

Change-Id: I4c8ce7778eb38a0a3ee674d2f1c488eda01cfeca
2015-05-13 11:41:15 -07:00
James Zern
f274c2199b vpx_mem: remove vpx_memcpy
vestigial. replace instances with memcpy() which they already were being
defined to.

Change-Id: Icfd1b0bc5d95b70efab91b9ae777ace1e81d2d7c
2015-04-28 19:59:41 -07:00
Scott LaVarnway
8b17f7f4eb Revert "Remove mi_grid_* structures."
(see I3a05cf1610679fed26e0b2eadd315a9ae91afdd6)

For the test clip used, the decoder performance improved by ~2%.
This is also an intermediate step towards adding back the
mode_info streams.

Change-Id: Idddc4a3f46e4180fbebddc156c4bbf177d5c2e0d
2015-04-21 11:16:45 -07:00
Yaowu Xu
9fd8abc541 vp9_pred_mv(): misc fixes and optimizations
1. skip near if it is same as nearest
2. correct rounding for converting mv to fullpel position
3. update pred_mv_sad after new mv search.

Overall .1%~.25% compression gains on rtc set for speed 5, 6, 7, 8.

Change-Id: Ic300ca53f7da18073771f1bb993c58cde9deee89
2015-03-20 17:17:04 -07:00
James Zern
428369293d vp9_mv_pred: quiet a static analysis warning
add an assert to validate pred_mv array size

Change-Id: I532b882b71e2baff3ac76e07ed133ec5a11bd0fc
2015-03-18 14:31:58 -07:00
Yunqing Wang
6e0ec0b2d9 vp9_ethread: fix me consts initialization to support aq_mode=3 encoding
While turning on "--aq_mode=3", the quantizers are updated by each
thread. Fixed the me consts initialization function to make sure
that the correct thread data are updated.

Change-Id: Ied27bb7bae76fc3fa2cda4f8c35ac0b46271bef4
2015-03-06 16:31:46 -08:00
Adrian Grange
3807dd82ab Make encoder buffer allocation dynamic
Frame buffers are now allocated dynamically on-demand.

Entries in the reference frame map, cm->ref_frame_map,
may now be set to -1 (INVALID_IDX) to indicate that
there is not a valid reference buffer in that "slot".

All slots in the reference frame map are now initialized
to the empty state (-1) and each buffer is initialized
to have a reference count of 0.

Change-Id: Id1afe98de98db4ae8b2dfefed7889c3b28c68582
2015-03-04 07:58:32 -08:00
Jingning Han
7a4e0b2265 Update partition rate cost in rtc speed 5
The block partition rate cost should be updated when recursive
partition search is needed.

Change-Id: I7bc5ad1fc2cbd3577dee7f7e8da111a2742bdeb9
2015-02-11 12:48:29 -08:00
hkuang
be6aeadaf4 Try again to merge branch 'frame-parallel' into master branch.
In frame parallel decode, libvpx decoder decodes several frames on all
cpus in parallel fashion. If not being flushed, it will only return frame
when all the cpus are busy. If getting flushed, it will return all the
frames in the decoder. Compare with current serial decode mode in which
libvpx decoder is idle between decode calls, libvpx decoder is busy
between decode calls.

Current frame parallel decode will only speed up the decoding for frame
parallel encoded videos. For non frame parallel encoded videos, frame
parallel decode is slower than serial decode due to lack of loopfilter
worker thread.

There are still some known issues that need to be addressed. For example:
decode frame parallel videos with segmentation enabled is not right sometimes.

* frame-parallel:
  Add error handling for frame parallel decode and unit test for that.
  Fix a bug in frame parallel decode and add a unit test for that.
  Add two test vectors to test frame parallel decode.
  Add key frame seeking to webmdec and webm_video_source.
  Implement frame parallel decode for VP9.
  Increase the thread test range to cover 5, 6, 7, 8 threads.
  Fix a bug in adding frame parallel unit test.
  Add VP9 frame-parallel unit test.
  Manually pick "Make the api behavior conform to api spec." from master branch.
  Move vp9_dec_build_inter_predictors_* to decoder folder.
  Add segmentation map array for current and last frame segmentation.
  Include the right header for VP9 worker thread.
  Move vp9_thread.* to common.
  ctrl_get_reference does not need user_priv.
  Seperate the frame buffers from VP9 encoder/decoder structure.
  Revert "Revert "Revert "Revert 3 patches from Hangyu to get Chrome to build:"""
 Conflicts:
       test/codec_factory.h
       test/decode_test_driver.cc
       test/decode_test_driver.h
       test/invalid_file_test.cc
       test/test-data.sha1
       test/test.mk
       test/test_vectors.cc
       vp8/vp8_dx_iface.c
       vp9/common/vp9_alloccommon.c
       vp9/common/vp9_entropymode.c
       vp9/common/vp9_loopfilter_thread.c
       vp9/common/vp9_loopfilter_thread.h
       vp9/common/vp9_mvref_common.c
       vp9/common/vp9_onyxc_int.h
       vp9/common/vp9_reconinter.c
       vp9/decoder/vp9_decodeframe.c
       vp9/decoder/vp9_decodeframe.h
       vp9/decoder/vp9_decodemv.c
       vp9/decoder/vp9_decoder.c
       vp9/decoder/vp9_decoder.h
       vp9/encoder/vp9_encoder.c
       vp9/encoder/vp9_pickmode.c
       vp9/encoder/vp9_rdopt.c
       vp9/vp9_cx_iface.c
       vp9/vp9_dx_iface.c

This reverts commit a18da9760a.

Change-Id: I361442ffec1586d036ea2e0ee97ce4f077585f02
2015-01-30 21:00:13 -08:00
Adrian Grange
527e073163 Remove elevate_newmv_thresh from SPEED_FEATURES (unused)
Change-Id: I78ef7f89586a329787f6bc4c58ec83af210989a3
2015-01-22 16:12:50 -08:00
Jingning Han
dad89d5ca1 Enable sub8x8 inter block search for RTC coding mode
This commit enables sub8x8 inter block coding for RTC mode. The
use of sub8x8 blocks can be turned on by allowing
choose_partitioning function to select 4x4/4x8/8x4 block sizes.

Change-Id: Ifbf1fb3888fe4c094fc85158ac3aa89867d8494a
2014-12-24 17:40:31 -08:00
Jingning Han
eefe869291 Simplify rate-distortion modeling function
Use left shift to replace one multiplication. The computation
outcome remains identical.

Change-Id: I1e1737af0a245de0d2a2bde10f0c171477199fc1
2014-12-15 11:51:16 -08:00
Yunqing Wang
edbd61e136 vp9_ethread: modify VP9_COMP structure
This patch modified struct VP9_COMP. Created a struct ThreadData
to include data that need to be copied for each thread. In
multiple thread case, one thread processes one tile. all threads
share one copy of VP9_COMP,
(refer to VP9_COMP *cpi in the code)
but each thread has its own copy of ThreadData,
(refer to ThreadData *td in the code).
Therefore, within the scope of encode_tiles(), both cpi and td
need to be passed as function parameters.

In single thread case, the FRAME_COUNTS pointer in ThreadData
points to "counts" in VP9_COMMON.

Change-Id: Ib37908b2d8e2c0f4f9c18f38017df5ce60e8b13e
2014-11-24 17:57:38 -08:00
Yunqing Wang
ad7586a9e1 vp9_ethread: move max/min partition size to mb struct
The max_partition_size and max_partition_size are set at the
beginning while setting speed features, and then adjusted at
SB level. Moving them to mb struct ensures there is a local
copy for each thread.

Change-Id: I7dd08dc918d9f772fcd718bbd6533e0787720ad4
2014-11-20 09:24:50 -08:00
Adrian Grange
0d085ebc0a Prepare for dynamic frame resizing in the recode loop
Prepare for the introduction of frame-size change
logic into the recode loop.

Separated the speed dependent features into
separate static and dynamic parts, the latter being
those features that are dependent on the frame size.

Change-Id: Ia693e28c5cf069a1a7bf12e49ecf83e440e1d313
2014-11-13 11:41:20 -08:00
Jingning Han
61966b1d10 Merge "Refactor vp9_update_rd_thresh_fact" 2014-10-31 08:55:28 -07:00
Jingning Han
f7b46d8c5e Refactor vp9_update_rd_thresh_fact
Reduce the scope of function parameters.

Change-Id: Ifef2cfb559908a97498ffdbd6ea53da1cd45a73c
2014-10-30 11:09:40 -07:00
Jingning Han
afa31ab9b8 Merge "Enable mode search threshold update in non-RD coding mode" 2014-10-29 12:42:22 -07:00
Jingning Han
9349a28e80 Enable mode search threshold update in non-RD coding mode
Adaptively adjust the mode thresholds after each mode search round
to skip checking less likely selected modes. Local tests indicate
5% - 10% speed-up in speed -5 and -6. Average coding performance
loss is -1.055%.

speed -5
vidyo1 720p 1000 kbps
16533 b/f, 40.851 dB, 12607 ms -> 16556 b/f, 40.796 dB, 11831 ms

nik 720p 1000 kbps
33229 b/f, 39.127 dB, 11468 ms -> 33235 b/f, 39.131 dB, 10919 ms

speed -6
vidyo1 720p 1000 kbps
16549 b/f, 40.268 dB, 10138 ms -> 16538 b/f, 40.212 dB, 8456 ms

nik 720p 1000 kbps
33271 b/f, 38.433 dB,  7886 ms -> 33279 b/f, 38.416 dB, 7843 ms

Change-Id: I2c2963f1ce4ed9c1cf233b5b2c880b682e1c1e8b
2014-10-29 10:55:34 -07:00
Adrian Grange
4074099ed8 Simplify vp9_set_rd_speed_thresholds_sub8x8
Change-Id: I4bf0f9a38697f5aea564a47afd7f02bb8b2888b6
2014-10-29 09:09:46 -07:00
Yunqing Wang
330a6b2756 Merge "vp9_ethread: allocate frame contexts outside VP9_COMMON struct" 2014-10-22 17:10:39 -07:00
Yunqing Wang
7c7e4d4eb8 vp9_ethread: allocate frame contexts outside VP9_COMMON struct
This patch allocated frame contexts outside VP9_COMMON. This allows
multiple threads to share the same copy of frame contexts, and
reduces the overhead. It also guarantees the correct update of
these contexts during bitstream packing. This patch doesn't change
encoding result.

Change-Id: Ic181a2460b891d1d587278a6d02d8057b9dbd353
2014-10-22 15:03:12 -07:00
Jingning Han
e2612fbd70 Add init and reset functions for RD_COST struct
Change-Id: I2902de7051a883fd22e27a655209233733969cfd
2014-10-15 15:02:06 -07:00
Deb Mukherjee
00a4b20fbe rdmult data type change
To fix a VS warning.

Change-Id: I4c530c0afe8d06acdb8cc78b7995aba57a25373d
2014-10-03 00:09:41 -07:00
Deb Mukherjee
35e8fa1458 rdmult data type change to fix high bit-depth
Fixes an intermittent assert failure for highbitdepth.

Change-Id: If8cad0209a94f1184b69c7b3f1d587934f857d9b
2014-10-02 07:37:26 -07:00
Deb Mukherjee
993d10a217 Adds various high bit-depth encode functions
Change-Id: I6f67b171022bbc8199c6d674190b57f6bab1b62f
2014-09-25 01:50:36 -07:00
hkuang
c70cea97ac Remove mi_grid_* structures.
mi_grid_* are arrays of pointer to pointer. They save the pointers that point
to the MIs in cm->mi. But they are unnecessary and complicated. The original
goal was to remove MODE_INFO_t copy. But with an extra MODE_INFO_t pointer
inside MODE_INFO_t, same goal could be achieved.

This commit totally removes the mi_grid_* structures. But there are still
many dummy MODE_INFO_t inside cm->mi which are a waste of memory. Next commit
will do on-demand MODE_INFO_t allocation in order to save these memories.

Change-Id: I3a05cf1610679fed26e0b2eadd315a9ae91afdd6
2014-09-19 21:27:11 -07:00
Deb Mukherjee
5cd0aab81a Adds high bitdepth quantization functions
Adds various high bitdepth quantization functions.

Change-Id: I36fc0bf75a1bd15128ed271df8723de0ac134b0c
2014-09-16 14:55:37 -07:00