On Nexus 7 speed -5, -6, -7, and -8 saw about a 1% increase
in perf for 480p. Speeds -5, -6, -7, and -8 saw about a 1.5%
increase in perf for 720p.
Tested on Nexus 7, built with ndk r10d, gcc 4.9.
Change-Id: Ibf17ebfd952a6aec941719bd8306df8ec4574bee
Currently, VP9 supports column-tile encoding, which allows a frame
to be encoded in multiple column tiles independently. The number of
column tiles are set by encoder option "--tile-columns". This
provides a way to encode a frame in parallel.
Based on previous set of patches, this patch implemented the tile-
based multi-threaded encoder. Each thread processes one or more
tiles.
Usage:
For HD clips:
--tile-columns=2 --threads=1/2/3/4
While using 4 threads, tests showed that the encoder achieved
2.3X - 2.5X speedup at good-quality speed 3, and 2X speedup at
realtime speed 5.
Change-Id: Ied987f8f2618b1283a8643ad255e88341733c9d4
Also removes some spurious changes in common/vp9_blockd.h which
was introduced by a rebase issue between nextgen and master branches.
Change-Id: If359f0e9a71bca9c2ba685a87a355873536bb282
(cherry picked from commit 005d80cd05269a299cd2f7ddbc3d4d8b791aebba)
(cherry picked from commit 08d2f548007fd8d6fd41da8ef7fdb488b6485af3)
(cherry picked from commit 4230c2306c194c058f56433a5275aa02a2e71d56)
This change is made in preparation for a
subsequent patch which adds acceleration
for the highbitdepth transform functions.
The highbitdepth transform functions attempt
to use 16/32bit sse instructions where possible,
but fallback to using the C implementations if
potential overflow is detected. For this reason
the dct routines are made global so they can be
called from the acceleration functions in the
subsequent patch.
Change-Id: Ia921f191bf6936ccba4f13e8461624b120c1f665
(cherry picked from commit 454342d4e77dbb67f4a3c10f97a57a6fcb46d9a0)
Also includes block error.
(This patch is mostly cherry picked from
commit db7192e0b014a331a1dcb102c8a1148e9f0e1081)
Change-Id: Idef18f90b111a0d0c9546543d3347e551908fd78
All sad function that process above 32 consecutive elements are optimized
for AVX2:
vp9_sad64x64
vp9_sad64x32
vp9_sad32x64
vp9_sad32x32
vp9_sad32x16
vp9_sad64x64_avg
vp9_sad64x32_avg
vp9_sad32x64_avg
vp9_sad32x32_avg
vp9_sad32x16_avg
The functions that appeared as a hotspot is vp9_sad32x32 and vp9_sad64x64
vp9_sad32x32 was optimized by 68% and vp9_sad64x64 was optimized by 90%
both of them gave and overall ~2.3% user level gain
Change-Id: Iccf86b375a2b54c5fbbe685902ead0c9a561b9fd
The concept:
There's too much noise in source pixels for variance and at low bitrate
the reconstructed looks nothing like the source so we have problems
getting good partitionings with either. This skirts the issue by using
a box blur scaled down version for variance calculations. To compare
against source_var_ moved keyframe to be rd based like source_var.
Change-Id: Ie3babdbfadae324b7b5a76bea192893af27f0624
This SSE2 is based on VP8 denoiser's SSE2 code. In VP8, there are
only 16x16 blocks in denoiser, while in VP9, there are 13 different
block sizes.
By adding this SSE2 code, the improvement of encoder speed is around
20%(using C code vs using SSE2 code), vary for different clips.
The unit test for VP9 denoiser is to confirm that the SSE2 code is
bit-exact with the C code. The unit test covers all block size.
Change-Id: Ic8d8ac26db4ea40a5f146b5678a065af07eaaa3d
vp9_variance16x16(), and vp9_get16x16var().
On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~16.7%.
Change-Id: Ib163aa99f56e680194aabe00dacdd7f0899a4ecb
and vp9_sad16x16_neon()
On a Nexus 7, vpxenc (in realtime mode, speed -6)
reported a performance improvement of ~17%.
Change-Id: I91e070cde2973451083d3f3d63b49b7886de9a85
vp9_rdopt is for making rd optimal mode decisions. vp9_rd is for all
other rd related routines. Anything used outside of making an rd optimal
decision belongs in rd.
Change-Id: I772a3073f7588bdf139f551fb9810b6864d8e64b
same reasoning as:
9f3a0db vp9_rtcd: correct avx2 references
these are all intrinsics, so don't depend on x86inc.asm
Change-Id: I915beaef318a28f64bfa5469e5efe90e4af5b827
Renames all x86_64 specific assembly files to consistently
end in _x86_64.asm. This will be useful for build systems to
handle these files differently.
All new 64-bit specific assembly files should use the new
naming convention.
Change-Id: I36c89584967c82ffc4088b1b5044ac15d2bb7536
vp9_block_error_sse2 can only handle 16 bytes at a time but
the function requires to handle a sequence of 32 bytes at a time
so each 16 bytes is handled in a different register.
With AVX2 optimization the 32 bytes can be handled in one register instead
of two in the SSE2
The vp9_block_error was optimized by 85%.
The user level was optimized by 1.2%
Change-Id: Ia8fffe60e61eff7432a5fbd538757894f6c319fd
Includes changes that are not compatible with VS windows builds.
Amongst other things stdint.h is not supported in VS.
This reverts commit 89fbf3de50.
Change-Id: Ifa86d7df250578d1ada9b539c9ff12ed0c523cdd
Assembly implementation of ssse3 8x8 forward 2D-DCT. The current
version is turned on only for x86_64. The average unit runtime
goes from 157 cycles down to 136 cycles, i.e., about 12.8% faster.
This translates into about 1.5% speed-up for pedestrian_area 1080p
at speed 2.
Change-Id: I0f12435857e9425ed7ce12541344dfa16837f4f4
This patch sets up a quad_tree structure (pc_tree) for holding all of
pick_mode_context data we use at any square block size during encoding
or picking modes. That includes contexts for 2 horizontal and 2 vertical
splits, one none, and pointers to 4 sub pc_tree nodes corresponding
to split. It also includes a pointer to the current chosen partitioning.
This replaces code that held an index for every level in the pick
modes array including: sb_index, mb_index,
b_index, ab_index.
These were used as stateful indexes that pointed to the current pick mode
contexts you had at each level stored in the following arrays
array ab4x4_context[][][],
sb8x4_context[][][], sb4x8_context[][][], sb8x8_context[][][],
sb8x16_context[][][], sb16x8_context[][][], mb_context[][], sb32x16[][],
sb16x32[], sb32_context[], sb32x64_context[], sb64x32_context[],
sb64_context
and the partitioning that had been stored in the following:
b_partitioning, mb_partitioning, sb_partitioning, and sb64_partitioning.
Prior to this patch before doing an encode you had to set the appropriate
index for your block size ( switch statement), update it ( up to 3
lookups for the index array value) and then make your call into a recursive
function at which point you'd have to call get_context which then
had to do a switch statement based on the blocksize, and then up to 3
lookups based upon the block size to find the context to use.
With the new code the context for the block size is passed around directly
avoiding the extraneous switch statements and multi dimensional array
look ups that were listed above. At any level in the search all of the
contexts are local to the pc_tree you are working on (in?).
In addition in most places code that used to call sub functions and
then check if the block size was 4x4 and index was > 0 and return
now don't preferring instead to call the right none function on the inside.
Change-Id: I06e39318269d9af2ce37961b3f95e181b57f5ed9
We don't use declarations from this file. The real declarations
(differently named) are in vp9_rtcd_defs.pl, e.g. vp9_full_search_sad.
Change-Id: I73cbf064305710ba20747233cfdbe67366f069a0
2 functions were optimized for avx2 by using full 256 bit register
In order to handle 32 elements in parallel instead of only 16 in parallel:
1. vp9_sad32x32x4d
2. vp9_sad64x64x4d
The function level gain is 66% and the user level gain is ~1%.
Change-Id: I4efbb3bc7d8bc03b64b6c98f5cd5c4a9dd3212cb
Removing all copies of identical vp8_mse2psnr/vp9_mse2psnr functions.
Using vpx_sse_to_psnr() instead in all places.
Change-Id: I15beef9834d43d8fc8a8a7a2d1fc5de3d658fed8
Optimizing 2 functions to process 32 elements in parallel instead of 16:
1. vp9_sub_pixel_variance64x64
2. vp9_sub_pixel_variance32x32
both of those function were calling vp9_sub_pixel_variance16xh_ssse3
instead of calling that function, it calls vp9_sub_pixel_variance32xh_avx2
that is written in avx2 and process 32 elements in parallel.
This Optimization gave 70% function level gain and 2% user level gain
Change-Id: I4f5cb386b346ff6c878a094e1c3b37e418e50bde
CONFIG_USE_X86INC is available to every makefile, there's no need to
duplicate its value with USE_X86INC
Change-Id: Id12bd5f09cba78abba56ab5a8f56351562e5b8b6
Adds an arbitrary-size resize library for use in scaling of input
frames in a non-normative manner in the vp9 encoder. The method
used is as follows:
Downsampling - Uses a 8 tap filter for factor of 2 decimation upto
a size just higher than the desired size. Then interpolates pixels
at a precision of 1/32 pel using a set of 8-tap filters.
Upsampling - Interpolates pixels at a precision of 1/32 pel using
a set of 8-tap filters.
There is no assembly optimization yet.
Change-Id: Ib5b81e174fc139da322bb97c8214d52289d60d8a
This commit setups a test framework for real-time coding. It enables
a light motion search for non-RD mode decision purpose.
Change-Id: I8bec656331539e963c2b685a70e43e0ae32a6e9d
Optimizing the variance functions: vp9_variance16x16, vp9_variance32x32,
vp9_variance64x64, vp9_variance32x16, vp9_variance64x32,
vp9_mse16x16 by migrating to AVX2
some of the functions were optimized by processing 32 elements instead of 16.
some of the functions were optimized by processing 2 loop strides of 16
elements in a single 256 bit register
This optimization gives between 2.4% - 2.7% user level performance gain
and 42% function level gain.
Change-Id: I265ae08a2b0196057a224a86450153ef3aebd85d
Since they used in encoder only. This commit also re-order includes
for the files that include vp9_extend.h
Change-Id: I929fc113f2135d3198cd1fc6a17434e5a2f8a459
Adding these functions to encapsulate tx_type check. Changing TX_TYPE to
int to match the declaration in vo9_rtch.h.
Change-Id: I6f3a2df6e35595ca73b6aaa9e3909ee7bc3fd16f
This should be similar to what x264 does with --aq-mode 1.
It works well with clips like parkjoy and touhou
(http://x264.nl/developers/Dark_Shikari/LosslessTouhou.mkv).
At low bitrates, the segmentation signaling overhead may negate the
benefits of this feature.
(PGW) Default changed to feature OFF to allow provisional merge.
Change-Id: I938abf9bb487e1d4ad3b0264ea03d9826275c70b
Enable use_x86inc as a commandline option. Fix Bug with sse2 when
x86inc is disabled. Adds Sad asm protection to x86inc protection
Change-Id: Iee0f9dd235ea10e8ace512eb362ba9bebe8c9df6
This is in preparation for the SSE2 version of the high-precision
32x32 forward DCT which will share a lot of code with the existing
low precision version used for rate-distortion search.
Change-Id: I7084b6bdfb480b1fabb8493fb14e3f7fcc7888c0
Support enabling it or disabling it. Moved read out to configure.sh
so that its done once instead of in make and in config.
Change-Id: I73a9190cf31de9f03e8a577f478fa522f8c01c8b
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.
Change-Id: I2df53770c2e850813ffa713e1a91b45b0082b904
Change vp9_block_error() to return a 64bit error variable, change all
callers to expect a 64bit return value (this will prevent overflows,
which we basically don't check for at all right now). Remove duplicate
block_error() function, which fixed that through truncation. Remove
old (incompatible) mmx/sse2 block_error SIMD versions and replace with
a new one that returns a 64bit value.
Encoding time of first 50 frames of bus @ 1500kbps goes from 3min29 to
3min23, i.e. a 3% overall speedup.
Change-Id: Ib71ac5508b5ee8a80f1753cd85d72df1629abe68
This seems to only be used in the encoder. Also remove an empty wrapper
file that contained forward declarations for this function, but didn't
actually define any actual functions.
Change-Id: Ifc561eef7ebe374a7d03698055e51e105f6d614b
Adding API to read/write uncompressed frame header bits (it is not final
yet). Separate functions to read/write uncompressed header. Moving
clr_type, error_resilient_mode, refresh_frame_context,
frame_parallel_decoding_mode, frame_context_idx from compressed partition
to uncompressed frame header.
Change-Id: Id3ed8a387980c652ae147549412f4ec24a0a5bd0
The API is not final yet and can be changed. Actual layout of
uncompressed frame part will be finalized later. Right now moving
clr_type, error_resilient_mode, refresh_frame_context,
frame_parallel_decoding_mode from first compressed partition to
uncompressed frame part.
Change-Id: I3afc5d4ea92c5a114f4c3d88f96858cccc15b76e
Conflicts:
vp9/common/vp9_findnearmv.c
vp9/common/vp9_rtcd_defs.sh
vp9/decoder/vp9_decodframe.c
vp9/decoder/x86/vp9_dequantize_sse2.c
vp9/encoder/vp9_rdopt.c
vp9/vp9_common.mk
Resolve file name changes in favor of master. Resolve rdopt changes in
favor of experimental, preserving the newer experiments.
Change-Id: If51ed8f457470281c7b20a5c1a2f4ce2cf76c20f
vp9_dequantize_x86 has only sse2 functions.
vp9_dct_sse2_intrinsics has no namespace collision and can drop
_intrinsics.
vp9_idct_mmx.h is unused.
Change-Id: Ic16e31fb372a1d1e841a62ecb4189fe8f95808ec
Frank Galligan
Yunqing Wang
Peter de Rivaz
Debargha Mukherjee
Jingning Han
levytamar82
Alex Converse
Jim Bankoski
JackyChen
Dmitry Kovalev
Scott LaVarnway
Tim Kopp
James Zern
levytamar82
Paul Wilkins
Marco Paniconi
Yaowu Xu
Guillaume Martres
Christian Duvivier
Ronald S. Bultje
Johann
John Koleszar