Jingning Han [Tue, 2 Jul 2013 23:48:15 +0000 (16:48 -0700)]
Enable early termination in rd search
This commit allows encoder to detect the cumulative rate-distortion
cost per transformed block inside a partition. If the cumulative
rd cost is already above the best rd value, it terminates the rest
operations and continue to next prediction mode test.
It reduces the runtime of bus at target bit-rate 2000 from 308 second
to 266 second, i.e., about 13% speed-up at no performance penalty.
Paul Wilkins [Mon, 1 Jul 2013 15:27:12 +0000 (16:27 +0100)]
Added two new skip experiments.
sf->unused_mode_skip_lvl. Tests modes as normal for all
sizes at or below the given level. At larger sizes it skips
all modes that were not chosen at any smaller size.
Hence setting BLOCK_SIZE_SB64X64 is in effect off.
Setting BLOCK_SIZE_AB4X4 will only consider modes that
were chosen for one or more 4x4 blocks at larger sizes.
sf->reference_masking.
Do a test encode of the NONE partition at one size and create
a reference frame mask based on the best rd choice. In the
full search only allow this reference frame.
Currently it is testing 64x64 and repeats this in the full search.
This does not work well with Jim's Partition code just now and
is disabled by default.
Yaowu Xu [Thu, 27 Jun 2013 19:07:07 +0000 (12:07 -0700)]
Added a speed feature use_square_partition_only
This commit adds a speed feature where only squared partition are
evaluated in partition picking. Enable this feature in cpu-used 2
reduces encoding time by ~30%.
loss of compression:
-0.9% on cif set
-1.23% on stdhd
Use pmovmskb to skip quantize loops over empty coefficients.
If none of the 16 coefficients that we quantize per loop iteration
are larger than the zbin, directly skip to the next round of coeffs,
rather than doing a full quantize loop that will eventually result
in 16 zeroes. This incurs a jump cost, but saves a lot of other work.
32x32 quant goes from 1349 -> 1184 cycles. The same approach yielded
no significantly positive results for smaller transforms, so is not
used there (8x8: 103 -> 101 cycles; 16x16: 302 -> 306 cycles).
Deb Mukherjee [Tue, 2 Jul 2013 18:18:00 +0000 (11:18 -0700)]
Speed feature to binary search dir intramodes
This speed feature will skip searching the directional intra prediction
modes D63, D117, D27, D153 if the best intra mode so far is not one of
the diagonal, horizontal or vertical directions closest to the respective
directions being tested. In other words, this implements a sort of
binary search in the angular domain.
Speedup: about 9-10%
Results: -0.05% only on derfraw300.
Deb Mukherjee [Fri, 21 Jun 2013 23:31:12 +0000 (16:31 -0700)]
Tx size selection enhancements
(1) Refines the modeling function and uses that to add some speed
features. Specifically, intead of using a flag use_largest_txfm as
a speed feature, an enum tx_size_search_method is used, of which
two of the types are USE_FULL_RD and USE_LARGESTALL. Two other
new types are added:
USE_LARGESTINTRA (use largest only for intra)
USE_LARGESTINTRA_MODELINTER (use largest for intra, and model for
inter)
(2) Another change is that the framework for deciding transform type
is simplified to use a heuristic count based method rather than
an rd based method using txfm_cache. In practice the new method
is found to work just as well - with derf only -0.01 down.
The new method is more compatible with the new framework where
certain rd costs are based on full rd and certain others are
based on modeled rd or are not computed. In this patch the existing
rd based method is still kept for use in the USE_FULL_RD mode.
In the other modes, the count based method is used.
However the recommendation is to remove it eventually since the
benefit is limited, and will remove a lot of complications in
the code
(3) Finally a bug is fixed with the existing use_largest_txfm speed feature
that causes mismatches when the lossless mode and 4x4 WH transform is
forced.
Results on derf:
USE_FULL_RD: +0.03% (due to change in the tables), 0% encode time reduction
USE_LARGESTINTRA: -0.21%, 15% encode time reduction (this one is a
pretty good compromise)
USE_LARGESTINTRA_MODELINTER: -0.98%, 22% encode time reduction
(currently the benefit of modeling is limited for txfm size selection,
but keeping this enum as a placeholder) .
USE_LARGESTALL: -1.05%, 27% encode-time reduction (same as existing
use_largest_txfm speed feature).
Yunqing Wang [Sun, 30 Jun 2013 00:34:51 +0000 (17:34 -0700)]
Add speed feature to disable splitmv
Added a speed feature in speed 1 to disable splitmv for HD (>=720)
clips. Test result on stdhd set: 0.3% psnr loss and 0.07% ssim
loss. Encoding speedup is 36%.
(For reference: The test result on derf set showed 2% psnr loss
and 1.6% ssim loss. Encoding speedup is 34%. SPLITMV should be
enabled for small resolution videos.)
Jingning Han [Mon, 1 Jul 2013 23:50:58 +0000 (16:50 -0700)]
Calculate rd cost per transformed block
Compute the rate-distortion cost per transformed block, and cumulate
the cost through all blocks inside a partition. This allows encoder
to detect if the cumulative rd cost is already above the best rd cost,
thereby enabling early termination in the rate-distortion optimization
search.
This should significantly speedup cost_coeffs(). Basically what the
patch does is to make the neighbour arrays padded by one item to
prevent an eob check in get_coef_context(), then it populates each
col/row scan and left/top edge coefficient with two times the same
neighbour - this prevents a single/double context branch in
get_coef_context(). Lastly, it populates neighbour arrays in pixel
order (rather than scan order), so we don't have to dereference the
scantable to get the correct neighbours.
Total encoding time of first 50 frames of bus (speed 0) at 1500kbps
goes from 2min10.1 to 2min5.3, i.e. a 2.6% overall speed increase.
Total encoding time for first 50 frames of bus (speed 0) @ 1500kbps
goes 2min34.8 to 2min14.4, i.e. a 10.4% overall speedup. The code is
x86-64 only, it needs some minor modifications to be 32bit compatible,
because it uses 15 xmm registers, whereas 32bit only has 8.
James Zern [Sat, 29 Jun 2013 01:07:37 +0000 (18:07 -0700)]
fix test compile error
since: 92479d9 Make update_partition_context faster
fixes:
vp9/common/vp9_blockd.h:408:22: error:
non-constant-expression cannot be narrowed from type 'int' to 'char' in
initializer list [-Wc++11-narrowing]
char pcvalue[2] = {~(0xe << boffset), ~(0xf <<boffset)};
^~~~~~~~~~~~~~~~~
Jingning Han [Fri, 28 Jun 2013 20:37:19 +0000 (13:37 -0700)]
Enable SSE2 4x4 ADST/DCT transform
This commit enables SSE2 4x4 foward hybrid transform. The runtime
goes from 249 cycles down to 74 cycles. Overall around 2% speed-up
at no compression performance change.
Dmitry Kovalev [Fri, 28 Jun 2013 22:32:01 +0000 (15:32 -0700)]
Cleanup inside vp9_decodemv.c.
Adding read_skip_coeff function. Renaming decode_mv to read_mv for
consistency with another function names. Removing redundant function
arguments. Renaming kfread_modes to read_intra_mode_info, read_mb_modes_mv
to read_inter_mode_info, vp9_decode_mb_mode_mv to vp9_read_mode_info,
vp9_decode_mode_mvs_init to vp9_prepare_read_mode_info. Inlining function
mb_mode_mv_init inside vp9_prepare_read_mode_info.
Ronald S. Bultje [Fri, 28 Jun 2013 00:41:54 +0000 (17:41 -0700)]
Make coefficient skip condition an explicit RD choice.
This commit replaces zrun_zbin_boost, a method of biasing non-zero
coefficients following runs of zero-coefficients to be rounded towards
zero, with an explicit skip-block choice in the RD loop.
The logic is basically that if individual coefficients should be rounded
towards zero (from a RD point of view), the trellis/optimize loop should
take care of it. If whole blocks should be zero (from a RD point of
view), a single RD check is much more efficient than a complete
serialization of the quantization loop.
Quality change: derf +0.5% psnr, +1.6% ssim; yt +0.6% psnr, +1.1% ssim.
SIMD for quantize will follow in a separate patch. Results for other
test sets pending.
Yaowu Xu [Thu, 27 Jun 2013 19:07:07 +0000 (12:07 -0700)]
Optimize partition search order
This commit change the partition search order to allow checking of
rectangular partition to be done after square partitions. It also
added a speed feature to skip rectangular partition check when
NONE is better than SPLIT in RD sense.
This feature roughly speed up encoder by 1.5X with loss on compression
-0.91% on cif set
-0.56% on stdhd set
Dmitry Kovalev [Thu, 27 Jun 2013 23:15:43 +0000 (16:15 -0700)]
Decoder's code cleanup.
Using vp9_set_pred_flag function instead of custom code, adding
decode_tokens function which is now called from decode_atom,
decode_sb_intra, and decode_sb.
Frank Galligan [Fri, 21 Jun 2013 19:58:46 +0000 (12:58 -0700)]
Add Neon optimized loop filter functions.
- Added vp9_loop_filter_horizontal_edge_neon and
vp9_loop_filter_vertical_edge_neon.
- The functions are based off the vp8 loopfilter
functions.
- Matches x86 md5 checksum.
Jingning Han [Wed, 26 Jun 2013 02:41:56 +0000 (19:41 -0700)]
Make intra predictor reference buffer configurable
This commit enables configurable reference buffer pointer for intra
predictor. This allows later removal of spatial dependency between
blocks inside a 64x64 superblock in the rate-distortion optimization
loop.
projects / libvpx / log