visual studio will warn if a 32-bit shift is implicitly converted to 64.
in this case integer storage is enough for the result.
since:
f3a9ae5ba Fix ubsan failure in vp9_mcomp.c.
Change-Id: I7e0e199ef8d3c64e07b780c8905da8c53c1d09fc
A previous patch turned on allow_exhaustive_searches feature only for
FC_GRAPHICS_ANIMATION content. This patch further modified the feature
by removing the exhaustive search limit, and made it no longer adaptive.
As a result, the 2 counts that recorded the number of motion searches
were removed, which helped achieve the determinism in the row based
multi-threading encoding. Tests showed that this patch didn't cause
the encoder much slower.
Used exhaustive_searches_thresh for this speed feature, and removed
allow_exhaustive_searches. Also, refactored the speed feature code
to follow the general speed feature setting style.
Change-Id: Ib96b182c4c8dfff4c1ab91d2497cc42bb9e5a4aa
The MV unit test revealed an integer overflow issue in vp9_mcomp.c.
This was caused if the MV was very large. In mv_err_cost(), when
mv->row = 8184, mv->col = 8184 and ref_mv is 0, mv_cost = 34363
and error_per_bit = 132412, causing the overflow.
BUG=webm:1406
Change-Id: I35f8299f22f9bee39cd9153d7b00d0993838845e
To prevent the motion vector out of range bug, added a motion vector unit
test in VP9. In the 4k video encoding, always forced to use extreme motion
vectors and also encouraged to use INTER modes. In the decoding, checked if
the motion vector was valid, and also checked the encoder/decoder mismatch.
The tests showed that this unit test could reveal the issue we saw before.
Change-Id: I0a880bd847dad8a13f7fd2012faf6868b02fa3b4
BUG=webm:1397
(yunqingwang)
To verify that this patch wouldn't cause much performance change,
the Borg tests were run. Here was the result:
avg_psnr overall_psnr ssim
hdres: -0.002 0.006 0.013
midres: 0 0 0
lowres: 0 0 0
Change-Id: Iae395ae7b741e0513cf5bab9dcace110b792a67d
(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
This commit resolves the compression performance regression in
real-time encoding setting when high bit-depth mode is enabled.
The current solution temporarily disables the SIMD implementations
of vpx_satd, hadamard8x8, and hadamard16x16 in high bit-depth mode.
The commit makes the coding results bit-wise identical between
regular coding pipeline and high bit-depth at profile 0.
BUG=webm:1365
Change-Id: Icfb900821733749685370460a1a5a7e07f76f4bf
For out-of-range cases, returned UINT_MAX instead of INT_MAX in the
sub-pixel mv search to be consistent with the "uint32_t" return type.
Change-Id: I8e206d771228c13d89bafbbe9f14722c8ecc6a7a
+5.857% BD-RATE on SCREEN_CONTENT
Leaving this off for non-screen content because:
+25.300% on TWITCH120
+37.833% BD-RATE on RTC
Change-Id: Ie0a312182d6cc859fb04298e4cd81d02b39e23fe
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
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
Re-investigated the second-level sub-pixel motion search. Improved the
way of choosing search points. Rewrote the second-level search code.
At speed 0, the borg tests showed:
1. for stdhd set, Avg PSNR gain: 0.216%; Overall PSNR gain: 0.196%;
SSIM gain: 0.206%. Only 1 out of 15 clips showed PSNR loss.
2. for derf set, Avg PSNR gain: 0.171%; Overall PSNR gain: 0.192%;
SSIM gain: 0.207%. Only 3 out of 30 clips showed PSNR losses.
Added the condition for third-point checking, namely, less points
were checked. Speed tests showed no speed loss(Avg 0.3% speedup at
speed 0).
Change-Id: I6284ebb3fa7ba63be8528184c49e06757211a7f1
Choose a different diagonal point to check when the two costs are
the same, making it consistent with the way we choose the best mv.
This slightly changes the encoding result, and the derflr set borg
test at speed 0 shows 0.027% Overall PSNR gain, 0.024% Avg PSNR
gain, and 0.043% SSIM gain.
Change-Id: Ic8ee3a6767394866d159e4f9e1c777604dd73c17
If the current best mv(namely, the search center) is still the best mv
after the first level search, the second level checks is skipped. This
patch doesn't change the bitstream. At speed 0, it speeds up the encoder
by 1% - 2%.
Change-Id: I054c91b884d3f7aef157436c061744562bd6506d
The only difference between the two was that the vp9 function allowed
for every step in the bilinear filter (16 steps) while vp8 only allowed
for half of those. Since all the call sites in vp9 (<< 1) the input, it
only ever used the same steps as vp8.
This will allow moving the subpel variance to vpx_dsp with the rest of
the variance functions.
Change-Id: I6fa2509350a2dc610c46b3e15bde98a15a084b75
