Граф коммитов

3103 Коммитов

Автор SHA1 Сообщение Дата
James Zern 4950dbceaf Merge changes from topic 'rm-loopfilter-count-param'
* changes:
  lpf_8_test: remove unneeded function wrapper
  remove loopfilter 'count' param TODOs
  split vpx_highbd_lpf_horizontal_16 in two
  split vpx_lpf_horizontal_16 in two
  vpx_highbd_lpf_horizontal_4: remove unused count param
  vpx_highbd_lpf_horizontal_8: remove unused count param
  vpx_highbd_lpf_vertical_4: remove unused count param
  vpx_highbd_lpf_vertical_8: remove unused count param
  vpx_lpf_horizontal_4: remove unused count param
  vpx_lpf_horizontal_8: remove unused count param
  vpx_lpf_vertical_4: remove unused count param
  vpx_lpf_vertical_8: remove unused count param
  lpf_8_test: add missing dspr2 tests
  lpf_8_test: add missing vpx_lpf_horizontal_4 tests
  lpf_8_test: add missing vpx_lpf_vertical_4 tests
  lpf_8_test: simplify function wrapper generation
2016-02-18 18:47:48 +00:00
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
James Zern 110d377899 remove loopfilter 'count' param TODOs
Change-Id: I25ce7314372ce2f521526ea7864ffc4ab62e4519
2016-02-16 23:14:03 -08:00
James Zern 9b44d9d00f split vpx_highbd_lpf_horizontal_16 in two
replace with vpx_highbd_lpf_horizontal_edge_16 and
vpx_highbd_lpf_horizontal_edge_8 to avoid passing a count parameter

Change-Id: I551f8cec0fce57032cb2652584bb802e2248644d
2016-02-16 23:13:58 -08:00
James Zern 1b519fb666 split vpx_lpf_horizontal_16 in two
replace with vpx_lpf_horizontal_edge_16 and vpx_lpf_horizontal_edge_8 to
avoid passing a count parameter

Change-Id: I848c95c02a3c6ebaa6c2bdf0983dce05cd645271
2016-02-16 22:57:45 -08:00
James Zern e7a23d703b vpx_highbd_lpf_horizontal_4: remove unused count param
Change-Id: I655a771e1b1a8753be5669ef9348a312ba6cfdbc
2016-02-16 22:57:45 -08:00
James Zern 5171857329 vpx_highbd_lpf_horizontal_8: remove unused count param
Change-Id: Iaca71ea3796115d4c2d43563b4e6f3914e21f1bf
2016-02-16 22:57:44 -08:00
James Zern 3c1019e49d vpx_highbd_lpf_vertical_4: remove unused count param
Change-Id: Ic6da723c5cf3cd8127db1f476c3e46ea134cb774
2016-02-16 22:57:44 -08:00
James Zern 72a9f06ac2 vpx_highbd_lpf_vertical_8: remove unused count param
Change-Id: Id16f7259897654831d31642c2d5e0bbe5e13416c
2016-02-16 22:57:44 -08:00
James Zern b1e97c6a25 vpx_lpf_horizontal_4: remove unused count param
Change-Id: Iec7d8eda343991f7d7d46931dca17af23c821d11
2016-02-16 22:57:27 -08:00
James Zern bd5a5bb561 vpx_lpf_horizontal_8: remove unused count param
Change-Id: I48741e167a7b09b7c9ad3bfc1c4b88ef1029ae46
2016-02-16 22:54:40 -08:00
James Zern 109a47b342 vpx_lpf_vertical_4: remove unused count param
Change-Id: I43a191cb3d42e51e7bca266adfa11c6239a8064c
2016-02-16 14:59:00 -08:00
James Zern 37225744db vpx_lpf_vertical_8: remove unused count param
Change-Id: Ic69406da00afb0f06588e8c0deb2b043952b078c
2016-02-16 14:59:00 -08:00
Marco 3cbc26f31b vp9-resize: Fix an issue with external dynamic resize.
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
2016-02-12 15:06:48 -08:00
James Zern ecd32d6faa Merge "Vidyo patch: Optimization for 1-to-2 downsampling and upsampling." 2016-02-05 02:36:03 +00:00
Scott LaVarnway 989c69303d Vidyo patch: Optimization for 1-to-2 downsampling and upsampling.
Change-Id: I9cc9780f506e025aea57485a9e21f0835faf173c
2016-02-04 14:50:26 -08:00
Paul Wilkins e062eb16fb Merge "Loop filter search resets on overlay frame." 2016-02-02 14:44:47 +00:00
hui su 5afc4e4c77 Fix some typos.
Change-Id: I32aacd014df6c927cf2893dc096cbe6ec7604b9b
2016-01-27 16:12:49 -08:00
Scott LaVarnway 5232326716 VP9: Eliminate MB_MODE_INFO
Change-Id: Ifa607dd2bb366ce09fa16dfcad3cc45a2440c185
2016-01-19 16:40:20 -08:00
paulwilkins 733bbab53a Loop filter search resets on overlay frame.
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
2016-01-19 13:05:15 +00:00
Scott LaVarnway d4bc17d696 Merge "VP9: inline vp9_use_mv_hp()" 2016-01-14 13:36:40 +00:00
Scott LaVarnway a85e552d95 VP9: Remove decoder args from find_mv_refs_idx()
The decoder does not use this function.

Change-Id: Ie67f909c0f4108ef286789c70df867d4b960a780
2016-01-13 13:30:40 -08:00
Scott LaVarnway de993a847f VP9: inline vp9_use_mv_hp()
Change-Id: Ib275bfc4c29c572d6c70e5ec6dbfc241590d3e3e
2016-01-13 08:02:05 -08:00
Scott LaVarnway 15939cb2d7 Merge "VP9: Eliminate unnecessary nearest/near searches" 2016-01-12 20:00:59 +00:00
Scott LaVarnway d8aa40634a VP9: Eliminate unnecessary nearest/near searches
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
2016-01-12 05:09:06 -08:00
Yaowu Xu 2bd4f44409 Assert no mv clamping for scaled references
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
2016-01-05 14:55:05 -08:00
Yaowu Xu ce6d3f1de4 Merge "Assert no 8x4/4x8 partition for scaled references" 2016-01-05 20:35:46 +00:00
Yaowu Xu 03a021a6fc Assert no 8x4/4x8 partition for scaled references
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
2016-01-04 18:33:37 -08:00
James Zern d36659cec7 move vp9_avg to vpx_dsp
Change-Id: I7bc991abea383db1f86c1bb0f2e849837b54d90f
2015-12-14 14:42:12 -08:00
Jacky Chen d9bba21306 Merge "Add vp9_avg_4x4_neon and the unit test." 2015-12-09 06:09:33 +00:00
jackychen 303f144eef Add vp9_avg_4x4_neon and the unit test.
Change-Id: I3ef9a9648841374ed3cc865a02053c14ad821a20
2015-12-08 17:23:36 -08:00
Scott LaVarnway f0b0b1fe62 VP9: Add ssse3 version of vpx_idct32x32_135_add()
Change-Id: I9a780131efaad28cf1ad233ae64c5c319a329727
2015-12-02 04:50:46 -08:00
James Zern fd51d90159 Merge changes Iaf8cbe95,I6748183d,I2a49811d
* changes:
  add vp9_satd_neon
  fix vp9_satd_sse2
  vp9_satd: return an int
2015-11-25 01:48:53 +00:00
James Zern eb1d0f8d60 add vp9_satd_neon
~60-65% faster at the function level across block sizes

Change-Id: Iaf8cbe95731c43fdcbf68256e44284ba51a93893
2015-11-24 16:09:10 -08:00
Alex Converse 4b038ad2ef Merge "Deduplicate some high bit depth tables" 2015-11-24 18:24:32 +00:00
James Zern 60760f710f fix vp9_satd_sse2
accumulate satd in 32-bits
+ add unit test

Change-Id: I6748183df3662ddb9d635f9641f9586f2fd38ad5
2015-11-20 14:35:46 -08:00
James Zern 3e0138edb7 vp9_satd: return an int
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
2015-11-20 14:35:38 -08:00
paulwilkins 0149fb3d6b Changes to exhaustive motion search.
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
2015-11-13 10:16:31 +00:00
Geza Lore 5eefd3ebfd Add AVX vectorized vp9_diamond_search_sad
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
2015-11-11 14:03:47 +00:00
James Zern 30466f26b4 Revert "Add AVX vectorized vp9_diamond_search_sad"
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
2015-11-06 13:15:01 -08:00
Yunqing Wang 57cae22c1e Merge "Add AVX vectorized vp9_diamond_search_sad" 2015-11-05 20:17:13 +00:00
Geza Lore f1342a7b07 Add AVX vectorized vp9_diamond_search_sad
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
2015-11-05 10:02:17 +00:00
Alex Converse 246e0eaa71 Deduplicate some high bit depth tables
Change-Id: I6977f7d155cc1e81ae2393933893caac6770821f
2015-11-03 15:40:44 -08:00
hui su e085fb643f Generate intra prediction reference values only when necessary
This can help increase encoding speed substantially.

Change-Id: Id0c009146e6e74d9365add71c7b10b9a57a84676
2015-11-02 10:26:50 -08:00
Alex Converse 989193c797 Make the zero handling in extend_to_full_distribution more explicit.
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
2015-10-29 14:46:55 -07:00
Alex Converse 663960e757 Revert "Replace the zero handling in extend_to_full_distribution."
This reverts commit 7f56cb2978.

It causes uninitialized reads in the first pass setting up later cost tables.

Change-Id: I2df498df3f5c03eff359f79edf045aed0c618dc9
2015-10-28 11:51:40 -07:00
Alex Converse 7f56cb2978 Replace the zero handling in extend_to_full_distribution.
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
2015-10-26 11:29:46 -07:00
Geza Lore aa8f85223b Optimize vp9_highbd_block_error_8bit assembly.
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
2015-10-21 12:30:40 +01:00
Yaowu Xu 568429512e Add a new enum type vpx_color_range_t
to make meaning of color_range obvious.

Change-Id: I303582e448b82b3203b497e27b22601cc718dfff
2015-10-16 16:27:18 -07:00
Geza Lore 0134764fa6 Optimization of 8bit block error for high bitdepth
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
2015-10-08 14:05:25 -07:00