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
replace with vpx_highbd_lpf_horizontal_edge_16 and
vpx_highbd_lpf_horizontal_edge_8 to avoid passing a count parameter
Change-Id: I551f8cec0fce57032cb2652584bb802e2248644d
replace with vpx_lpf_horizontal_edge_16 and vpx_lpf_horizontal_edge_8 to
avoid passing a count parameter
Change-Id: I848c95c02a3c6ebaa6c2bdf0983dce05cd645271
External dynamic resize with swapping width and height was
not handled properly.
Fix is to re-init loop-filter under certain condtions.
Modify unittest to test this case.
Without this change test will fail.
Relates to: https://bugs.chromium.org/p/webm/issues/detail?id=1140
Change-Id: I7d81ca7fe0783b3bc103a52a7b7cf073a96be26e
This patch fixes a bug that causes the loop filter search to reset to
a low value or zero after each arf overlay frame. We expect the overlay
frames to need little or no loop filtering but this should not propagate.
Change-Id: I895b28474cf200f20d82793f3de40b60b19579fd
Prior to this patch, read_inter_block_mode_info() would
find the nearmv and nearestmv for all modes. Now it does not
search for ZEROMV modes and breaks out early for NEARMV and
NEWMV modes.
Change-Id: Ifa7b1eaf58bb03b9c7792ea5012fef477527d0fd
Under --enable-better-hw-compabibility, this commit adds the asserts
that no mv clamping is applied for scaled references, so when built
with this configure option, decoder will assert if an input bitstream
triggger mv clamping for scaled reference frames.
Change-Id: I786e86a2bbbfb5bc2d2b706a31b0ffa8fe2eb0cb
This commit adds a new configure option:
--enable-better-hw-compatibility
The purpose of the configure option is to provide information on known
hardware decoder implementation bugs, so encoder implementers may
choose to implement their encoders in a way to avoid triggering these
decoder bugs.
The WebM team were made aware of that a number of hardware decoders
have trouble in handling the combination of scaled frame reference
frame and 8x4 or 4x8 partitions. This commit added asserts to vp9
decoder, so when built with above configure option, the decoder can
assert if an input bitstream triggers such decoder bug.
Change-Id: I386204cfa80ed16b50ebde57f886121ed76200bf
the final sum may use up to 26 bits
+ add a unit test
+ disable the sse2 as the result will rollover; this will be fixed in a
future commit
Change-Id: I2a49811dfaa06abfd9fa1e1e65ed7cd68e4c97ce
This change alters the nature and use of exhaustive motion search.
Firstly any exhaustive search is preceded by a normal step search.
The exhaustive search is only carried out if the distortion resulting
from the step search is above a threshold value.
Secondly the simple +/- 64 exhaustive search is replaced by a
multi stage mesh based search where each stage has a range
and step/interval size. Subsequent stages use the best position from
the previous stage as the center of the search but use a reduced range
and interval size.
For example:
stage 1: Range +/- 64 interval 4
stage 2: Range +/- 32 interval 2
stage 3: Range +/- 15 interval 1
This process, especially when it follows on from a normal step
search, has shown itself to be almost as effective as a full range
exhaustive search with step 1 but greatly lowers the computational
complexity such that it can be used in some cases for speeds 0-2.
This patch also removes a double exhaustive search for sub 8x8 blocks
which also contained a bug (the two searches used different distortion
metrics).
For best quality in my test animation sequence this patch has almost
no impact on quality but improves encode speed by more than 5X.
Restricted use in good quality speeds 0-2 yields significant quality gains
on the animation test of 0.2 - 0.5 db with only a small impact on encode
speed. On most clips though the quality gain and speed impact are small.
Change-Id: Id22967a840e996e1db273f6ac4ff03f4f52d49aa
This function now has an AVX intrinsics version which is about 80%
faster compared to the C implementation. This provides a 2-4% total
speed-up for encode, depending on encoding parameters. The function
utilizes 3 properties of the cost function lookup table, constructed
in 'cal_nmvjointsadcost' and 'cal_nmvsadcosts'.
For the joint cost:
- mvjointsadcost[1] == mvjointsadcost[2] == mvjointsadcost[3]
For the component costs:
- For all i: mvsadcost[0][i] == mvsadcost[1][i]
(equal per component cost)
- For all i: mvsadcost[0][i] == mvsadcost[0][-i]
(Cost function is even)
These must hold, otherwise the AVX version of the function cannot be used.
Change-Id: I6c2791d43022822a9e6ab43cd124a773946d0bdc
This reverts commit f1342a7b07.
This breaks 32-bit builds:
runtime error: load of misaligned address 0xf72fdd48 for type 'const
__m128i' (vector of 2 'long long' values), which requires 16 byte
alignment
+ _mm_set1_epi64x is incompatible with some versions of visual studio
Change-Id: I6f6fc3c11403344cef78d1c432cdc9147e5c1673
This function now has an AVX intrinsics version which is about 80%
faster compared to the C implementation. This provides a 2-4% total
speed-up for encode, depending on encoding parameters. The function
utilizes 3 properties of the cost function lookup table, constructed
in 'cal_nmvjointsadcost' and 'cal_nmvsadcosts'.
For the joint cost:
- mvjointsadcost[1] == mvjointsadcost[2] == mvjointsadcost[3]
For the component costs:
- For all i: mvsadcost[0][i] == mvsadcost[1][i]
(equal per component cost)
- For all i: mvsadcost[0][i] == mvsadcost[0][-i]
(Cost function is even)
These must hold, otherwise the AVX version of the function cannot be used.
Change-Id: I184055b864c5a2dc37b2d8c5c9012eb801e9daf6
The old workaround "p = 0 ? 0 : p -1" is misleading.
?: happens before =
assigning back to p truncates to one byte.
Therefore it is equivalent to (p - 1) & 0xFF, but the check just exists
to work around a first pass bug, so let's make the work around more
clear.
https://bugs.chromium.org/p/webm/issues/detail?id=1089
Change-Id: I587c44dd61c1f3767543c0126376f881889935af
This reverts commit 7f56cb2978.
It causes uninitialized reads in the first pass setting up later cost tables.
Change-Id: I2df498df3f5c03eff359f79edf045aed0c618dc9
The old workaround "p = 0 ? 0 : p -1" is misleading.
?: happens before =
assigning back to p truncates to one byte.
Therefore it is equivalent to (p - 1) & 0xFF, but the check just exists
to work around a first pass bug, so let's make the work around more
clear.
https://code.google.com/p/webm/issues/detail?id=1089
Change-Id: Ia6dcc8922e1acbac0eeca23a4d564a355c489572
A new version of vp9_highbd_error_8bit is now available which is
optimized with AVX assembly. AVX itself does not buy us too much, but
the non-destructive 3 operand format encoding of the 128bit SSEn integer
instructions helps to eliminate move instructions. The Sandy Bridge
micro-architecture cannot eliminate move instructions in the processor
front end, so AVX will help on these machines.
Further 2 optimizations are applied:
1. The common case of computing block error on 4x4 blocks is optimized
as a special case.
2. All arithmetic is speculatively done on 32 bits only. At the end of
the loop, the code detects if overflow might have happened and if so,
the whole computation is re-executed using higher precision arithmetic.
This case however is extremely rare in real use, so we can achieve a
large net gain here.
The optimizations rely on the fact that the coefficients are in the
range [-(2^15-1), 2^15-1], and that the quantized coefficients always
have the same sign as the input coefficients (in the worst case they are
0). These are the same assumptions that the old SSE2 assembly code for
the non high bitdepth configuration relied on. The unit tests have been
updated to take this constraint into consideration when generating test
input data.
Change-Id: I57d9888a74715e7145a5d9987d67891ef68f39b7
If high bit depth configuration is enabled, but encoding in profile 0,
the code now falls back on optimized SSE2 assembler to compute the
block errors, similar to when high bit depth is not enabled.
Change-Id: I471d1494e541de61a4008f852dbc0d548856484f
vp9_filter_block_plane_ss11() and vp9_filter_block_plane_non420()
are only called for the uv planes.
Change-Id: Iacd3b3242c8ce581edd37c8f06d95efc8a0f88a3
The loopfilter masks are now built in the decode loop.
This is done so we can eventually reduce the number of
MODE_INFO structs required by the decoder.
The encoder builds the masks for the entire frame prior
to calling the loopfilter.
Change-Id: Ia2146b07e0acb8c50203e586dfae0c4c5b316f11
When configured with high bitdepth enabled, the 8bit transform
stopped using optimised code. This made 8bit content decode slowly.
Change-Id: I67d91f9b212921d5320f949fc0a0d3f32f90c0ea
The name "display_*" (or "d_*") is used for non-compatible information
(that is, the cropped frame dimensions in pixels, as opposed to the
intended screen rendering surface size). Therefore, continuing to use
display_* would be confusing to end users. Instead, rename the field
to render_*, so that struct vpx_image can include it.
Change-Id: Iab8d2eae96492b71c4ea60c4bce8121cb2a1fe2d
In decoder, export (eventually) into vpx_image_t.range field. In
encoder, use oxcf->color_range to set it (same way as for
color_space).
See issue 1059.
Change-Id: Ieabbb2a785fa58cc4044bd54eee66f328f3906ce
Sync the encoder's buffer offset calculation for sub8x8 block motion
compensated prediction with scaled reference frame to match the
decoder's behavior. This resolves an enc/dec mismatch issue when
sub8x8 inter mode with scaled is turned on.
Change-Id: I4bab3672b007a5ae0c992f8a701341892d2458b0
This commit clears the function naming convention in vpx_dsp. It
replaces vp9_ prefix of global functions with vpx_ prefix. It also
removes the vp9_ prefix from static functions.
Change-Id: I6394359a63b71a51dda01342eec6a3cc08dfeedf
This commit moves the module inverse transform functions from vp9
to vpx_dsp folder. The hybrid transform wrapper functions stay in
the vp9 folder, since it involves codec-specific data structures.
Change-Id: Ib066367c953d3d024c73ba65157bbd70a95c9ef8
It in essence refactors the code for both the interpolation
filtering and the convolution. This change includes the moving
of all the files as well as the changing of the code from vp9_
prefix to vpx_ prefix accordingly, for underneath architectures:
(1) x86;
(2) arm/neon; and
(3) mips/msa.
The work on mips/drsp2 will be done in a separate change list.
Change-Id: Ic3ce7fb7f81210db7628b373c73553db68793c46
Replace the duplicate coefficient definition in neon implementations
of inverse transform with those from vpx_dsp/txfm_common.h
Change-Id: I4cd9bd9569ab1793dfdbb6f16d80bcb581599f0d
This commit replaces vp9_idct.h with txfm_common.h in many SIMD
implementation files for precise file dependency.
Change-Id: If73dd726bb16537e7494f28538b0a169810f9756
Separate the common coefficient constant into vpx_dsp/txfm_common.h.
Move the SSE2 macro definitions to vpx_dsp/x86/txfm_common_sse2.h.
This clears the use case of vp9_idct.h in vpx_dsp folder.
Change-Id: I319735a2abf42888e5080ac14cfbcde34be7b121
This commit factors out common macro definitions from the forward
and inverse transform implementations into vpx_dsp. It removes
the duplicate macro definitions from encoder and decoder folders.
Change-Id: I92301acbd3317075e9c5f03328a25abb123bca78
This commit factors the 4x4, 8x8, and 16x16 2D-DCT forward
transform operations into vpx_dsp folder.
Change-Id: I084b117b79c0925edcbcabb93f62b9f4bf8dbe7d
The SSE2 version high bit-depth forward hybrid transforms are
essentially using the C functions via cross referencing to 1-D
functions in vp9_dct.c. This commit unifies the two versions and
removes the unnecessary dependency.
Change-Id: Ib4d0702a138f8daf7d0bd97c141ee7088f293765
The following quantization functions were moved:
vp9_quantize_b
vp9_quantize_b_32x32
vp9_highbd_quantize_b
vp9_highbd_quantize_b_32x32
vp9_quantize_dc
vp9_quantize_dc_32x32
vp9_highbd_quantize_dc
vp9_highbd_quantize_dc_32x32
The purpose of doing that was to allow these functions to be shared
by multiple codecs.
Change-Id: Id8ab939f283353cdd07bd930d47db3d932a5d87f
This commit moves the loop filter dspr2 implementation from vp9 to
vpx_dsp directory. It also fixes header file format issues.
Change-Id: I09203ed4bd267d7fd76bb79a6ee84a37646206b2
The various tap loop filter operations are common functions across
codec. This commit moves them along with SIMD optimizations to
vpx_dsp folder.
Change-Id: Ia5fa0b2e5289cdb98467502a549c380b9c60e92c
Move the clamp functions to vpx_dsp_common.h file. Clear out the
dependency of vp9_loopfilter_filters.c on vp9_common.h file.
Change-Id: I9c4b928bcd7f597106b5aa96354356d3775a3431
This commit simplifies the intra block boundary condition logic.
It removes the block index from the argument set.
Change-Id: If00142512eb88992613d6609356dfd73ba390138
The encoder gets its dqcoeff from the context tree. In the decoder move
it to directly after MACROBLOCKD.
Change-Id: I46c9b76f26956a360d17de0b26ecb994dae34ecb
The decoding process handles detokenization and reconstruction per
transform block sequentially. There is no need to offset the dqcoeff
buffer according to the transform block index. This allows to
reduce the memory spill and improve cache performance.
Change-Id: Ibb8bfe532a7a08fcabaf6d42cbec1e986901d32d
inline the code directly in read_mv_component(), the only place where it
was being used; this removes a function call in a hot function
Change-Id: I66f99c0c9ce3bc310101dbca4a470f023cc6fb55
This reverts commit a42df86c03.
this change causes MSA/VP9SubpelVarianceTest.Ref and
MSA/VP9SubpelVarianceTest.ExtremeRef failures under
mips32r5el-msa-linux-gnu and mips64r6el-msa-linux-gnu
Change-Id: I40b71a0b774eaeb31f66f795733f95cf360909f7
This reverts commit 61774ad1c4.
this change causes MSA/VP9SubpelAvgVarianceTest.Ref failures under
mips32r5el-msa-linux-gnu and mips64r6el-msa-linux-gnu
Change-Id: I7fb520c12b2a3b212d5e84b7619a380a48e49bb0
The vp9_lpf_vertical_16_dual function optimized for x86 32bit target. The hot code in that function was caused by the call to the transpose8x16.
The gcc generated assembly created uneeded fills and spills to the stack. By interleaving 2 loads and unpack instructions, in addition to hoisting the consumer
instruction closer to the producer instructions, we eliminated most of the fills and spills and improve the function-level performance by 17%.
credit for writing the function as well as finding the root cause goes to Erik Niemeyer (erik.a.niemeyer@intel.com)
Change-Id: I6173cf53956d52918a047d1c53d9a673f952ec46
This commit moves the primitive multi-threading files from vp9
folder to vpx_thread, which will be accessible by all vpx codec.
Change-Id: Ib51e66e9c69801c10631fab56d35a0c0aaed5883
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
