Commit Graph

2657 Commits

Author SHA1 Message Date
James Zern
2d58761993 Revert "Improved 8t filters"
This is incompatible with most toolchains other than gcc.

Revert "Deleted #include <inttypes.h>"

This reverts commit 4d018be950.

This reverts commit d22a504d11.

Change-Id: I1751dc6831f4395ee064e6748281418e967e1dcf
2013-09-13 15:13:06 -07:00
Jingning Han
e8a967d960 Merge "Adaptive motion search control" 2013-09-13 14:43:23 -07:00
Jingning Han
c4826c5941 Adaptive motion search control
This commit enables adaptive constraint on motion search range for
smaller partitions, given the motion vectors of collocated larger
partition as a candidate initial search point.

It makes speed 0 runtime of bus at CIF and 2000 kbps goes from
167s down to 162s (3% speed-up), at 0.01dB performance gains. In
the settings of speed 1, this makes the runtime goes from 33687 ms
to 32142 ms (4.5% speed-up), at 0.03dB performance gains.

Compression performance wise, it gains at speed 1:
derf  0.118%
yt    0.237%
hd    0.203%
stdhd 0.438%

Change-Id: Ic8b34c67810d9504a9579bef2825d3fa54b69454
2013-09-13 13:58:10 -07:00
Deb Mukherjee
0c3038234d Merge "Clean up of the search best filter speed feature" 2013-09-13 11:03:59 -07:00
Paul Wilkins
5d8642354e Merge "Fix VP9_mode_order[]" 2013-09-13 09:19:31 -07:00
Scott LaVarnway
8fc95a1b11 Merge "New mode_info_context storage -- undo revert" 2013-09-13 08:56:20 -07:00
Paul Wilkins
1407cf8588 Fix VP9_mode_order[]
Mis-merge of the following change managed to break mode order
and delete two mode options (new alt ref and near alt ref)
It also created a situation where we could test two undefined
modes off the end of the VP9_mode_order[] data structure.
  "clang warnings : remove split and i4x4_pred fake modes"
  "Change Id: I8ef3c*"

Initial testing on Akiyo at speed 2.
101.35	 44.567	 44.447 improves to
96.82	 44.915	 44.815

Approx 0.3-0.4db gain and 2.5% size reduction

Change-Id: Icff813e7c0778d140ad4f0eea18cf1ed203c4e34
2013-09-13 13:33:26 +01:00
Paul Wilkins
9c9a3b2775 Merge "Deleted #include <inttypes.h>" 2013-09-13 01:05:31 -07:00
Jim Bankoski
324ebb704a Merge "fix clang warning in rdopt" 2013-09-12 16:39:05 -07:00
hkuang
86fb12b600 Merge "Add neon optimize iht8x8 which is 282% faster than C." 2013-09-12 15:42:44 -07:00
Christian Duvivier
25655e5794 Merge "First draft of vp9_short_idct32x32_add_neon." 2013-09-12 14:23:00 -07:00
hkuang
182366c736 Add neon optimize iht8x8 which is 282% faster than C.
Change-Id: I963dd4a6e8671957403ccbb9a16ea7de703e3530
2013-09-12 11:49:05 -07:00
Jim Bankoski
9ee9918dad fix clang warning in rdopt
either missed this or it crept back in

Change-Id: I6cc1519d09e558be7250254c25bde2ae720555ea
2013-09-12 06:39:42 -07:00
Jim Bankoski
e7f2aa0fb8 clang warnings : ref frame enum mismatch
Convert from refframe_type_t to VP9_REFFRAME

Change-Id: Iff4043c3fdb3e1c9c2b412bdffd5da8ed913ec13
2013-09-12 06:29:07 -07:00
Jim Bankoski
cddde51ec5 Merge "clang warnings : remove split and i4x4_pred fake modes" 2013-09-12 06:20:45 -07:00
Paul Wilkins
4d018be950 Deleted #include <inttypes.h>
This seems not to be needed and is not supported
in the Windows build.

Change-Id: Iaca3bbf8cca283aee6bc336cb31ba9dd4610322b
2013-09-12 13:43:07 +01:00
Paul Wilkins
66755abff4 Merge "Changes in speed 2 settings" 2013-09-12 02:22:45 -07:00
Jim Bankoski
7fb42d909e clang warnings : remove split and i4x4_pred fake modes
Change-Id: I8ef3c7c0f08f0f1f4ccb8ea4deca4cd8143526ee
2013-09-11 16:34:55 -07:00
Christian Duvivier
6a501462f8 First draft of vp9_short_idct32x32_add_neon.
Lots of TODO which will be taken care in upcoming changes. As is,
about 6x faster than C version.

Change-Id: Ie2557b72fd2d8edca376dbf400a4d173aa5e63e0
2013-09-11 15:19:38 -07:00
Deb Mukherjee
b964646756 Clean up of the search best filter speed feature
Removes this speed feature since it is very slow and unlikely
to be used in practice. This cleanup removes a bunch of unnecessary
complications in the outer encode loop.

Change-Id: I3c66ef1ca924fbfad7dadff297c9e7f652d308a1
2013-09-11 15:16:36 -07:00
Scott LaVarnway
23845947c4 Merge "Improved 8t filters" 2013-09-11 14:34:54 -07:00
Jim Bankoski
d09abfa9f7 Merge "resolve clang issue : implicit convert tx_mode -> tx_size" 2013-09-11 13:40:11 -07:00
Scott LaVarnway
d22a504d11 Improved 8t filters
Reformatted version of a patch submitted by Erik/Tamar
from Intel.  For the test clips used, the decoder
performance improved by ~2%.

Change-Id: Ifbc37ac6311bca9ff1cfefe3f2e9b7f13a4a511b
2013-09-11 13:56:32 -04:00
Deb Mukherjee
69fe840ec4 Changes in speed 2 settings
Propose some changes to the speed 2 settings to improve quality.
In particular, turns off the adjust_thresholds_by_speed feature
which improves results by 6%. Also removes the code for
adjust_thresholds_by_speed since it conflicts with the adaptive
rd thresh feature.

Overall, with this change speed 2 is -15.2% from speed 0 settings,
on derf, which is significantly better than -21.6% down before.

Change-Id: I6e90a563470979eb0c258ec32d6183ed7ce9a505
2013-09-11 10:54:07 -07:00
Scott LaVarnway
ac6093d179 New mode_info_context storage -- undo revert
mode_info_context was stored as a grid of MODE_INFO structs.
The grid now constists of pointers to MODE_INFO structs.  The
MODE_INFO structs are now stored as a stream (decoder only),
eliminating unnecessary copies and is a little more cache
friendly.

Change-Id: I031d376284c6eb98a38ad5595b797f048a6cfc0d
2013-09-11 13:45:44 -04:00
Yunqing Wang
079183c1a8 code cleanup
Removed unused function.

Change-Id: Icb12a09e4d303968be6aec9fae1ef05935913a4f
2013-09-11 09:32:00 -07:00
Jingning Han
65fe7d7605 Merge "Remove redundant condition check in 32x32 quant" 2013-09-10 16:39:18 -07:00
Jingning Han
cb24406da5 Merge "Remove the use of uninitialized_safe in encode_sb_" 2013-09-10 12:05:22 -07:00
Jingning Han
5d93feb6ad Remove redundant condition check in 32x32 quant
The c code implementation of 32x32 quantization does the zbin check
of all coefficients prior to the quant/dequant loop, hence removing
the redundant zbin check inside the loop. This only affects the
c code version. SSSE3 version does not separate the zbin check out.

Change-Id: Ic197a7d61d0b25fcac3cc092987651378cb56e4e
2013-09-10 12:04:33 -07:00
Deb Mukherjee
3d22d3ae0c Merge "Small tweaks on the constant quality mode" 2013-09-10 11:16:47 -07:00
Deb Mukherjee
09830aa0ea Small tweaks on the constant quality mode
Improves results a little.
derf is now +1.078% over bitrate control.

Change-Id: I4812136f3e67be21d14ec089419976a32a841785
2013-09-10 10:16:19 -07:00
Yunqing Wang
0607abc3dd Stop partition checking when distortion is small
If the current obtained distortion is very small, which happens
for static image case, we pick the current partition type without
further split checking.

This won't affect regular videos. For static videos, we got 10%~12%
encoding speed gain. PSNR was better for some clips, and worse for
others. Overall it was even.

Change-Id: If787a57bedf46fc595ca4f5ded2b0c0a69e9fdef
2013-09-10 10:13:24 -07:00
Yunqing Wang
939791a129 Modify encode breakout for static frames
Thank Paul for the suggestions. While turning on static-thresh
for static-image videos, a big jump on bitrate was seen. In this
patch, we detected static frames in the video using first-pass
stats. For different cases, disable encode breakout or reduce
encode breakout threshold to limit the skipping.

More modification need be done to break incorrect partition
picking pattern for static frames while skipping happens.

Change-Id: Ia25f47041af0f04e229c70a0185e12b0ffa6047f
2013-09-10 09:06:03 -07:00
hkuang
f4a6f936b5 Merge "Speed up idct16x16 by rearrange instructions." 2013-09-10 08:23:57 -07:00
Paul Wilkins
4f660cc018 Modified mode skip functionality.
A previous speed feature skipped modes not used in earlier
partitions but this not longer worked as intended following
changes to the partition coding order and in conjunction
with some other speed features (Especially speed 2 and above).

This modified mode skip feature sets a mask after the first X
modes have been tested in each partition depending on the
reference frame of the current best case.

This patch also makes some changes to the order modes are
tested to fit better with this skip functionality.

Initial testing suggests speed and rd hit count improvements
of up to 20% at speed 1. Quality results. (derf -1.9%, std hd  +0.23%).

Change-Id: Idd8efa656cbc0c28f06d09690984c1f18b1115e1
2013-09-10 13:30:10 +01:00
Paul Wilkins
901c495482 Added extra check to rd_auto_partition_range()
Added check that the returned max and minimum are
valid in bottom and right border cases.

Change-Id: I2d6cdc9b5f04c7d0ff512ddcf3228331e028bf9b
2013-09-10 13:29:23 +01:00
hkuang
fc5ec206a7 Speed up idct16x16 by rearrange instructions.
Speed improve from 376% to 400% faster base on assembly-perf.

Change-Id: If0b2eccc39d5793dc101ce9feb7fcadf88396ea2
2013-09-09 18:00:13 -07:00
Ivan Maltz
20abe595ec Merge "API extensions and sample app for spacial scalable encoder" 2013-09-09 16:57:01 -07:00
Ivan Maltz
01b35c3c16 API extensions and sample app for spacial scalable encoder
Sample app: vp9_spatial_scalable_encoder
vpx_codec_control extensions:
  VP9E_SET_SVC
  VP9E_SET_WIDTH, VP9E_SET_HEIGHT, VP9E_SET_LAYER
  VP9E_SET_MIN_Q, VP9E_SET_MAX_Q
expanded buffer size for vp9_convolve

modified setting of initial width in vp9_onyx_if.c so that layer size
can be set prior to initial encode

Default number of layers set to 3 (VPX_SS_DEFAULT_LAYERS)
Number of layers set explicitly in vpx_codec_enc_cfg.ss_number_layers

Change-Id: I2c7a6fe6d665113671337032f7ad032430ac4197
2013-09-09 15:57:56 -07:00
Jingning Han
18c780a0ff Remove the use of uninitialized_safe in encode_sb_
Initialize the probability model context with default value in
encode_sb.

Change-Id: Id826114024dfc21c7ef41aea9f4a0316d4a5cb95
2013-09-09 15:41:16 -07:00
James Zern
c1913c9cf4 Merge "Revert "New mode_info_context storage"" 2013-09-09 14:38:01 -07:00
James Zern
54a03e20dd Revert "New mode_info_context storage"
This reverts commit dae17734ec

Encode crashes, leaks and increases integer overflow errors.

Change-Id: I595aa2649bb8d0b6552ff91652837a74c103fda2
2013-09-09 13:37:01 -07:00
Yaowu Xu
b19126b291 Merge "Reduce the amount of extension in src frames" 2013-09-09 08:09:56 -07:00
Paul Wilkins
740acd6891 Merge "Enable kf restrictions at speed 4" 2013-09-09 05:39:13 -07:00
Yaowu Xu
65c2444e15 Reduce the amount of extension in src frames
The commit changes the border pixel extension from 160 pixel each side
to what is necessary in arnr filter or motion estimation portion, i.e.
16 pixel on top and left side. For right or bottom side, the extension
is changed to either round up image size to multiple of 64 or at least
16 pixels.

Change-Id: Ic05e19b94368c1ab4df568723aae5734e6c3d2c5
2013-09-08 15:51:54 -07:00
Jim Bankoski
9faa7e8186 resolve clang issue : implicit convert tx_mode -> tx_size
Change-Id: Ifc9da470358f58e800e3d0d70a565b61e5f7834a
2013-09-08 07:17:12 -07:00
Jim Bankoski
e378566060 Merge "New mode_info_context storage" 2013-09-08 07:16:25 -07:00
Jingning Han
09bc942b47 Fix overflow issue in 16x16 quantization SSSE3
The 16x16 transform unit test suggested that the peak coefficient
value can reach 32639. This could cause potential overflow issue
in the SSSE3 implmentation of 16x16 block quantization. This commit
fixes this issue by replacing addition with saturated addition.

Change-Id: I6d5bb7c5faad4a927be53292324bd2728690717e
2013-09-06 21:06:10 -07:00
Paul Wilkins
f15cdc7451 Enable kf restrictions at speed 4
Change-Id: I453409d3be3f5fe118b15affde45cb52184aef20
2013-09-06 11:16:04 -07:00
Deb Mukherjee
e378a89bd6 Support a constant quality mode in VP9
Adds a new end-usage option for constant quality encoding in vpx. This
first version implemented for VP9, encodes all regular inter frames
using the quality specified in the --cq-level= option, while encoding
all key frames and golden/altref frames at a quality better than that.

The current performance on derfraw300 is +0.910% up from bitrate control,
but achieved without multiple recode loops per frame.

The decision for qp for each altref/golden/key frame will be improved
in subsequent patches based on better use of stats from the first pass.
Further, the qp for regular inter frames may also be varied around the
provided cq-level.

Change-Id: I6c4a2a68563679d60e0616ebcb11698578615fb3
2013-09-06 10:30:53 -07:00