platform/upstream/libvpx.git
6 years agoAdjust threshold in gf_boost for 1 pass vbr
Marco [Fri, 13 Oct 2017 22:31:02 +0000 (15:31 -0700)]
Adjust threshold in gf_boost for 1 pass vbr

Small inncrease the sad_thresh1, avoids some false
detection of possible scene changes within lag.

Small improvement in few clips on ytlive, otherwise neutral change.

Change-Id: Ia79b53bb657bbce65a7aac7d20666b6373d5af8b

6 years agoMerge "Further Corpus VBR change."
Paul Wilkins [Fri, 13 Oct 2017 15:59:58 +0000 (15:59 +0000)]
Merge "Further Corpus VBR change."

6 years agoMerge "Corpus Wide VBR test implementation."
Paul Wilkins [Fri, 13 Oct 2017 15:59:45 +0000 (15:59 +0000)]
Merge "Corpus Wide VBR test implementation."

6 years agoMerge "vp8: [loongson] optimize dct with mmi"
Shiyou Yin [Fri, 13 Oct 2017 00:37:57 +0000 (00:37 +0000)]
Merge "vp8: [loongson] optimize dct with mmi"

6 years agoMerge "Adjust to scene detection for 1 pass vbr."
Marco Paniconi [Thu, 12 Oct 2017 19:36:33 +0000 (19:36 +0000)]
Merge "Adjust to scene detection for 1 pass vbr."

6 years agoAdjust to scene detection for 1 pass vbr.
Marco [Tue, 10 Oct 2017 22:30:32 +0000 (15:30 -0700)]
Adjust to scene detection for 1 pass vbr.

Expose the threshold for setting key frame on cut,
and increase it for speed 5.
Also small adjustment to min_thresh.

No change in overall metrics or fps.
Small quality improvement and lower encode time on scene cuts.

Change-Id: I36e06ff3b26b6c29aede39c23fce454525fc9026

6 years agoMerge "vp9: use nonrd pick_intra for small blocks on keyframes."
Jerome Jiang [Thu, 12 Oct 2017 17:29:27 +0000 (17:29 +0000)]
Merge "vp9: use nonrd pick_intra for small blocks on keyframes."

6 years agoMerge changes I38783d97,If5160c0c
Kyle Siefring [Thu, 12 Oct 2017 16:12:38 +0000 (16:12 +0000)]
Merge changes I38783d97,If5160c0c

* changes:
  Extend 16 wide AVX2 convolve8 code to support averaging.
  Add AVX2 version of vpx_convolve8_avg.

6 years agovp9: use nonrd pick_intra for small blocks on keyframes.
Jerome Jiang [Thu, 12 Oct 2017 00:13:39 +0000 (17:13 -0700)]
vp9: use nonrd pick_intra for small blocks on keyframes.

Keyframe encoding is more than 2x faster.
Disabled on Speed 8.

Change-Id: I2157318b6ac8253fa5398322c72d98cd7fa9b2b6

6 years agovp8: [loongson] optimize dct with mmi
Shiyou Yin [Wed, 13 Sep 2017 06:03:11 +0000 (14:03 +0800)]
vp8: [loongson] optimize dct with mmi

1. vp8_short_fdct4x4_mmi
2. vp8_short_fdct8x4_mmi
3. vp8_short_walsh4x4_mmi

Change-Id: I89a7df25cfd09fae309fac257ad8b6a3dc1c8acb

6 years agoMerge "vp8: [loongson] optimize quantize with mmi"
Shiyou Yin [Thu, 12 Oct 2017 00:33:17 +0000 (00:33 +0000)]
Merge "vp8: [loongson] optimize quantize with mmi"

6 years agoAdjust threshold in datarate tests for 1 pass VBR
Marco [Wed, 11 Oct 2017 18:06:34 +0000 (11:06 -0700)]
Adjust threshold in datarate tests for 1 pass VBR

Small increase in threshold for the 1 pass VBR datarate tests.
Needed due to commit:
<017257a Adjustment to scene detection and key frame>

Change-Id: I28b3bd7db2192a8cc2bccc3cb0e3b8dbb910ca16

6 years agovp8: [loongson] optimize quantize with mmi
Shiyou Yin [Mon, 11 Sep 2017 10:07:25 +0000 (18:07 +0800)]
vp8: [loongson] optimize quantize with mmi

1. vp8_fast_quantize_b_mmi
2. vp8_regular_quantize_b_mmi

Change-Id: Ic6e21593075f92c1004acd67184602d2aa5d5646

6 years agoAdd 4 to 1 scaling x86 optimization
Linfeng Zhang [Wed, 4 Oct 2017 16:55:56 +0000 (09:55 -0700)]
Add 4 to 1 scaling x86 optimization

Change-Id: I51c190f0a88685867df36912522e67bdae58a673

6 years agoMerge "Fix alignment in vpx_image without external allocation."
Jerome Jiang [Tue, 10 Oct 2017 23:02:05 +0000 (23:02 +0000)]
Merge "Fix alignment in vpx_image without external allocation."

6 years agoFix alignment in vpx_image without external allocation.
Jerome Jiang [Tue, 10 Oct 2017 02:33:03 +0000 (19:33 -0700)]
Fix alignment in vpx_image without external allocation.

This restores behaviors prior to
<40c8fde Fix image width alignment. Enable ImageSizeSetting test.>.

BUG=b/64710201

Change-Id: I559557afe80d5ff5ea6ac24021561715068e7786

6 years agoAdjustment to scene detection and key frame.
Marco [Tue, 10 Oct 2017 00:53:21 +0000 (17:53 -0700)]
Adjustment to scene detection and key frame.

For 1 pass vbr: use higher threshold on avg_sad
and force key frame under scene cut detection if
above the threshold. Allow it for speed >= 6 for now,
since it does not use the full nonrd_pickmode partition
(as in speed 5).

Improves quality somewhat on scene cut frames.
Neutral on overall metrics and fps for speed 6 on
ytlive set.

Change-Id: I12626f7627419ca14f9d0d249df86c7104438162

6 years agoMerge changes I9d4c1af5,I882da3a0
Linfeng Zhang [Tue, 10 Oct 2017 17:29:50 +0000 (17:29 +0000)]
Merge changes I9d4c1af5,I882da3a0

* changes:
  Rename some inline functions in NEON scaling
  Generalize 2:1 vp9_scale_and_extend_frame_ssse3()

6 years agoFurther Corpus VBR change.
paulwilkins [Thu, 17 Aug 2017 13:13:29 +0000 (14:13 +0100)]
Further Corpus VBR change.

Change to the bit allocation within a GF/ARF group.

Normal VBR and CQ mode allocate bits to a GF/ARF group based of the mean
complexity score of the frames in that group but then share bits evenly between
the "normal" frames in that group regardless of the individual frame complexity
scores (with the exception of the middle and last frames).

This patch alters the behavior for the experimental "Corpus VBR" mode such that
the allocation is always based on the individual complexity scores.

Change-Id: I5045a143eadeb452302886cc5ccffd0906b75708

6 years agoCorpus Wide VBR test implementation.
paulwilkins [Fri, 19 May 2017 14:08:15 +0000 (15:08 +0100)]
Corpus Wide VBR test implementation.

This patch makes further changes to support an experimental
corpus wide VBR mode that uses a corpus complexity
number as the midpoint of the distribution used to allocate bits
within a clip, rather than some average error score derived from the
clip itself.

At the moment the midpoint number is hard wired for testing and
the mode is enabled or disabled through a #ifdef.  Ultimately this
would need to be controlled by command line parameters.

Change-Id: I9383b76ac9fc646eb35a5d2c5b7d8bc645bfa873

6 years agoExtend 16 wide AVX2 convolve8 code to support averaging.
Kyle Siefring [Sun, 8 Oct 2017 03:25:03 +0000 (23:25 -0400)]
Extend 16 wide AVX2 convolve8 code to support averaging.

Also adds vpx_convolve8_avg_horiz_avx2.

Change-Id: I38783d972ac26bec77610e9e15a0a058ed498cbf

6 years agoRename some inline functions in NEON scaling
Linfeng Zhang [Wed, 4 Oct 2017 20:04:40 +0000 (13:04 -0700)]
Rename some inline functions in NEON scaling

Change-Id: I9d4c1af53d57f72fc716bacbe3b0965719c045ac

6 years agoMerge "Update vp9_scale_and_extend_frame_ssse3()"
Linfeng Zhang [Mon, 9 Oct 2017 16:20:00 +0000 (16:20 +0000)]
Merge "Update vp9_scale_and_extend_frame_ssse3()"

6 years agoAdd AVX2 version of vpx_convolve8_avg.
Kyle Siefring [Sat, 7 Oct 2017 20:02:02 +0000 (16:02 -0400)]
Add AVX2 version of vpx_convolve8_avg.

vpx_convolve8_avg works by first running a normal horizontal filter then a
vertical filter averages at the end.

The added vpx_convolve8_avg_avx2 calls pre-existing AVX2 code for the
horizontal step.

vpx_convolve8_avg_vert_avx2 is also added, but only uses ssse3 code.

Change-Id: If5160c0c8e778e10de61ee9bf42ee4be5975c983

6 years agoMerge "ppc: Add vpx_idct32x32_1024_add_vsx"
James Zern [Sat, 7 Oct 2017 19:08:26 +0000 (19:08 +0000)]
Merge "ppc: Add  vpx_idct32x32_1024_add_vsx"

6 years agoMerge "Revert "Speed >=5 real-time: add TM intra mode for high_source_sad.""
Marco Paniconi [Fri, 6 Oct 2017 22:41:34 +0000 (22:41 +0000)]
Merge "Revert "Speed >=5 real-time: add TM intra mode for high_source_sad.""

6 years agoRevert "Speed >=5 real-time: add TM intra mode for high_source_sad."
Marco Paniconi [Fri, 6 Oct 2017 22:14:56 +0000 (22:14 +0000)]
Revert "Speed >=5 real-time: add TM intra mode for high_source_sad."

This reverts commit 9311ef18b4b4eff0da3adf9d702a34f489a270ff.

Reason for revert:
Notice small regression in some clips.
Will revisit in another change.

Original change's description:
> Speed >=5 real-time: add TM intra mode for high_source_sad.
>
> Small/neutral change in metrics or speed for ytlive.
> Some improvement in quality on frames with big content change.
>
> Change-Id: Ib3b0703a5f28ea6710e90324436e27598ab7384d

TBR=marpan@google.com,builds@webmproject.org,jianj@google.com

Change-Id: I9d8ec5195bb05ddf329d325699355185affb9b13
No-Presubmit: true
No-Tree-Checks: true
No-Try: true

6 years agoAdjust threshold in scene detection
Marco [Fri, 6 Oct 2017 17:53:40 +0000 (10:53 -0700)]
Adjust threshold in scene detection

For 1 pass vbr: increase min_thresh slightly, and also add
condition on golden/arf update for using full nonrd_pick_partition.

Reduces possible false detection for scene cut detection.

Neutral/small change in metrics or speed for speed 5.

Change-Id: I388f4d9a56e3cc763e0148338c1bc0381e58ad76

6 years agoMerge "Speed >=5 real-time: add TM intra mode for high_source_sad."
Marco Paniconi [Fri, 6 Oct 2017 06:29:46 +0000 (06:29 +0000)]
Merge "Speed >=5 real-time: add TM intra mode for high_source_sad."

6 years agoSpeed >=5 real-time: add TM intra mode for high_source_sad.
Marco [Thu, 5 Oct 2017 19:58:51 +0000 (12:58 -0700)]
Speed >=5 real-time: add TM intra mode for high_source_sad.

Small/neutral change in metrics or speed for ytlive.
Some improvement in quality on frames with big content change.

Change-Id: Ib3b0703a5f28ea6710e90324436e27598ab7384d

6 years agoMerge "vpx_codec.h: namespace local defines"
James Zern [Fri, 6 Oct 2017 05:30:16 +0000 (05:30 +0000)]
Merge "vpx_codec.h: namespace local defines"

6 years agovpx_codec.h: namespace local defines
James Zern [Thu, 5 Oct 2017 22:09:33 +0000 (15:09 -0700)]
vpx_codec.h: namespace local defines

add VPX_ to UNUSED/*DEPRECATED to avoid conflicts with other headers.

Change-Id: Ie16bdac3575bc1af57a05d37e65b994370585377

6 years agovp9_ethread_test: abort early/add more detailed output
James Zern [Thu, 5 Oct 2017 22:02:51 +0000 (15:02 -0700)]
vp9_ethread_test: abort early/add more detailed output

in the case compare_fp_stats fails report the 2 values and their index

Change-Id: I927a832b7a1e24c392961093b7caee1134223def

6 years agoMerge "Adjust threshold for adapt_partition for speed 6."
Marco Paniconi [Thu, 5 Oct 2017 03:28:06 +0000 (03:28 +0000)]
Merge "Adjust threshold for adapt_partition for speed 6."

6 years agoAdjust threshold for adapt_partition for speed 6.
Marco [Thu, 5 Oct 2017 01:01:37 +0000 (18:01 -0700)]
Adjust threshold for adapt_partition for speed 6.

Lower SAD threshold to select non_rd pickmode partition
at superblock level more often.
Small gain in metrics, small/negligible decrease in speed.

Change-Id: I0f728236b91a604e4ca7e02039adc54d5985c4dc

6 years agoMerge "Avoid nonrd_pick_partition for speed >= 6."
Marco Paniconi [Wed, 4 Oct 2017 23:36:27 +0000 (23:36 +0000)]
Merge "Avoid nonrd_pick_partition for speed >= 6."

6 years agoAvoid nonrd_pick_partition for speed >= 6.
Marco [Wed, 4 Oct 2017 22:27:45 +0000 (15:27 -0700)]
Avoid nonrd_pick_partition for speed >= 6.

For 1 pass vbr speed >= 6: when REFERENCE_PARTITION is selected,
avoid doing the full nonrd_pickmode based partition.
No change in overall metrics or speed.
Reduces encode times on scene cuts by 10-20%.

Change-Id: I0310b1610cc1c83793a509e0a9059840e8f18308

6 years agoMerge "Modify early exit for alt_ref in nonrd_pickmode."
Marco Paniconi [Wed, 4 Oct 2017 19:38:49 +0000 (19:38 +0000)]
Merge "Modify early exit for alt_ref in nonrd_pickmode."

6 years agoGeneralize 2:1 vp9_scale_and_extend_frame_ssse3()
Linfeng Zhang [Tue, 3 Oct 2017 23:09:19 +0000 (16:09 -0700)]
Generalize 2:1 vp9_scale_and_extend_frame_ssse3()

Change-Id: I882da3a04884d5fabd4cd591c28682cbb2d76aa5

6 years agoUpdate vp9_scale_and_extend_frame_ssse3()
Linfeng Zhang [Tue, 3 Oct 2017 16:59:11 +0000 (09:59 -0700)]
Update vp9_scale_and_extend_frame_ssse3()

Change-Id: I22622faebfcc36f7a4d1f37e3800ae8ab87c8cd4

6 years agoModify early exit for alt_ref in nonrd_pickmode.
Marco [Wed, 4 Oct 2017 18:41:52 +0000 (11:41 -0700)]
Modify early exit for alt_ref in nonrd_pickmode.

For 1 pass vbr mode:
On no-show_frame/ARF: instead of skipping alt_ref_frame
completely in mode testing, allow for checking (0, 0) on alt_ref.

Small gain in metrics, ~0.18%, no change in speed.

Change-Id: I32a3c24faca64ab70dd5091071a0dc301db7dd1e

6 years agoMerge changes Id6a8c549,Ib1e0650b,Ic369dd86
Linfeng Zhang [Wed, 4 Oct 2017 16:15:14 +0000 (16:15 +0000)]
Merge changes Id6a8c549,Ib1e0650b,Ic369dd86

* changes:
  Refactor x86/vpx_subpixel_8t_intrin_ssse3.c
  Add vpx_dsp/x86/mem_sse2.h
  Add transpose_8bit_{4x4,8x8}() x86 optimization

6 years agoMerge "Fix image width alignment. Enable ImageSizeSetting test."
Jerome Jiang [Wed, 4 Oct 2017 14:48:03 +0000 (14:48 +0000)]
Merge "Fix image width alignment. Enable ImageSizeSetting test."

6 years agoEnable arf usage for speed >= 6, 1 pass vbr.
Marco [Wed, 4 Oct 2017 00:14:24 +0000 (17:14 -0700)]
Enable arf usage for speed >= 6, 1 pass vbr.

For speed 6 on ytlive set:
On average, speed slowdown ~5%, quality gain ~2%.

Change-Id: Ia18237cc1d52c54d7e2cb3c71f571cf37ef61b44

6 years agovp9: 1 pass vbr: Limit qpdelta on high_source_sad.
Marco [Fri, 29 Sep 2017 18:34:00 +0000 (11:34 -0700)]
vp9: 1 pass vbr: Limit qpdelta on high_source_sad.

For 1 pass vbr: when significant content/scene change is detected
(high_source_sad = 1) reduce/turnoff the additional qdelta on the
active_worst_quality. This helps somewhat to reduce the occurrence
of large frame sizes and large encode times.
Allow it only when use_altef_onepass is enabled.

Neutral/no change on metrics.

Change-Id: I1dd97dd2ab892d65f707b841b27a5de300b714ea

6 years agoMerge "vpx: fix nasm build errors"
James Zern [Tue, 3 Oct 2017 21:47:49 +0000 (21:47 +0000)]
Merge "vpx: fix nasm build errors"

6 years agovpx: fix nasm build errors
Scott LaVarnway [Sat, 30 Sep 2017 12:51:24 +0000 (05:51 -0700)]
vpx: fix nasm build errors

BUG=webm:1462,766721

Change-Id: Icfa536a8e38623636b96c396e3c94889bfde7a98

6 years agoRefactor x86/vpx_subpixel_8t_intrin_ssse3.c
Linfeng Zhang [Mon, 2 Oct 2017 21:29:06 +0000 (14:29 -0700)]
Refactor x86/vpx_subpixel_8t_intrin_ssse3.c

Change-Id: Id6a8c549709a3c516ed5d7b719b05117c5ef8bac

6 years agoAdd vpx_dsp/x86/mem_sse2.h
Linfeng Zhang [Mon, 2 Oct 2017 20:46:15 +0000 (13:46 -0700)]
Add vpx_dsp/x86/mem_sse2.h

Add some load and store sse2 inline functions.

Change-Id: Ib1e0650b5a3d8e2b3736ab7c7642d6e384354222

6 years agoUse adapt_partition for ARF in 1 pass.
Marco [Tue, 3 Oct 2017 17:55:55 +0000 (10:55 -0700)]
Use adapt_partition for ARF in 1 pass.

For speed 6 real-time mode: use adapt_partition
on ARF frame instead of REFERENCE_PARTITION (which is slower).
This requires enabling compute_source_sad_onepass for no-show_frames.

Speedup of ~3-5% on some clips that heavily use ARF,
small loss (~0.2%) in quality on ytlive set.

Change-Id: Ib50acc97df06458244a6ac55d2bd882c30012536

6 years agoAdd transpose_8bit_{4x4,8x8}() x86 optimization
Linfeng Zhang [Mon, 2 Oct 2017 20:01:56 +0000 (13:01 -0700)]
Add transpose_8bit_{4x4,8x8}() x86 optimization

Change-Id: Ic369dd86b3b81686f68fbc13ad34ab8ea8846878

6 years agoMerge "ARF in 1 pass vbr: modify skip ref_frame in nonrd_pickmode."
Marco Paniconi [Tue, 3 Oct 2017 03:01:14 +0000 (03:01 +0000)]
Merge "ARF in 1 pass vbr: modify skip ref_frame in nonrd_pickmode."

6 years agoARF in 1 pass vbr: modify skip ref_frame in nonrd_pickmode.
Marco [Mon, 2 Oct 2017 21:00:18 +0000 (14:00 -0700)]
ARF in 1 pass vbr: modify skip ref_frame in nonrd_pickmode.

Speedup of ~2-3% on 1080p clips speed 6.
Neutral/negligible loss in metrics on ytlive.

Change-Id: I7ac47a4d8b58c566920bae29a94a0e8d59c36dee

6 years agoAdd 4 to 3 scaling NEON optimization
Linfeng Zhang [Tue, 12 Sep 2017 18:49:58 +0000 (11:49 -0700)]
Add 4 to 3 scaling NEON optimization

Speed comparing with the one calling vpx_scaled_2d_neon()
  ~1.7 x in general
  ~2.8x for BILINEAR filter

BUG=webm:1419

Change-Id: I8f0a54c2013e61ea086033010f97c19ecf47c7c6

6 years agoSpecialize 4 to 3 frame scaling in C
Linfeng Zhang [Wed, 20 Sep 2017 17:58:39 +0000 (10:58 -0700)]
Specialize 4 to 3 frame scaling in C

Scale 3x3 block instead of 16x16 block in each loop. Disabled by
default.

Benefits:
1. Reduced number of different phase_scaler from 16 to 3.
   Optimization code will be smaller and faster.
2. Maximum phase_scaler drifting will be reduced from 5/16 to 1/24.
   (The drifting is 1/(3*16) in each step.)

BUG=webm:1419

Change-Id: I59a1f7496d89a1b090498c935d30cfcf1d0c282b

6 years agoMerge "vpxdsp: [x86] add highbd_d135_predictor functions"
Scott LaVarnway [Mon, 2 Oct 2017 15:00:19 +0000 (15:00 +0000)]
Merge "vpxdsp: [x86] add highbd_d135_predictor functions"

6 years agoppc: Add vpx_idct32x32_1024_add_vsx
Alexandra Hájková [Mon, 31 Jul 2017 19:07:22 +0000 (19:07 +0000)]
ppc: Add  vpx_idct32x32_1024_add_vsx

Change-Id: I55cd0a1569ccc47a53d0ecf751aac259d510e10d

6 years agoFix partition selection in speed features for arf overlay frame.
Marco [Fri, 29 Sep 2017 21:54:56 +0000 (14:54 -0700)]
Fix partition selection in speed features for arf overlay frame.

For real-time mode. Move the switch to fixed partition
for is_src_frame_alt_ref so all speeds may use it
if use_altref_onepass is set.

Improves metrics by ~2% for ytlive set at speed 4
(where use_altref_onepass is currently used).

Change-Id: I033240386598c9dbd0364da89ccbcca64bc663ee

6 years agoEnable use_altref_onepass for speed 4 real-time mode.
Marco [Fri, 29 Sep 2017 17:53:59 +0000 (10:53 -0700)]
Enable use_altref_onepass for speed 4 real-time mode.

Used for VBR mode with lag-in-frames > 0.
On ytlive set at speed 4: ~3% average gain.

Change-Id: I45dad1700bf8be9d8f177815dc062774f6f2f0de

6 years agovpxdsp: [x86] add highbd_d135_predictor functions
Scott LaVarnway [Fri, 29 Sep 2017 13:34:16 +0000 (06:34 -0700)]
vpxdsp: [x86] add highbd_d135_predictor functions

C vs SSE2 speed gains:
_4x4 : ~1.81x

C vs SSSE3 speed gains:
_8x8 : ~1.96x
_16x16 : ~1.88x
_32x32 : ~2.02x

BUG=webm:1411

Change-Id: Iefaf8b39afbbfe34c1ad1d21e3a003b20f1f61e0

6 years agovpxdsp: [x86] add highbd_d117_predictor functions
Scott LaVarnway [Wed, 20 Sep 2017 12:21:23 +0000 (05:21 -0700)]
vpxdsp: [x86] add highbd_d117_predictor functions

C vs SSE2 speed gains:
_4x4 : ~2.04x

C vs SSSE3 speed gains:
_8x8 : ~2.82x
_16x16 : ~5.93x
_32x32 : ~2.79x

BUG=webm:1411

Change-Id: I31d949695991c067dac89d91e0bed3e666c94993

6 years agoFix image width alignment. Enable ImageSizeSetting test.
Jerome Jiang [Wed, 27 Sep 2017 18:08:37 +0000 (11:08 -0700)]
Fix image width alignment. Enable ImageSizeSetting test.

BUG=b/64710201

Change-Id: I5465f6c6481d3c9a5e00fcab024cf4ae562b6b01

6 years agoSet rc->high_source_sad = 0 before scene detection.
Marco [Thu, 28 Sep 2017 17:47:34 +0000 (10:47 -0700)]
Set rc->high_source_sad = 0  before scene detection.

Only has effect when sf->use_altref_onepass is enabled,
as in that case scene detection is skipped for non-show frame
and so high_source_sad does not get reset to 0.

No change in metrics or speed.

Change-Id: I421f066d239341449c18826089e1810b9fc5967f

6 years agoMerge "vp9: Modification to adapt the ARF usage for 1 pass vbr"
Marco Paniconi [Thu, 28 Sep 2017 16:52:28 +0000 (16:52 +0000)]
Merge "vp9: Modification to adapt the ARF usage for 1 pass vbr"

6 years agovp9: Modification to adapt the ARF usage for 1 pass vbr
Marco [Thu, 21 Sep 2017 17:59:33 +0000 (10:59 -0700)]
vp9: Modification to adapt the ARF usage for 1 pass vbr

Add stats for past ARF usage, and use it to disable
ARF usage based on some conditions.

Overall improvement on ytlive set, reduces the regression
on the problem clips for this feature.

Only affects when sf->use_altref_onepass is enabled
(currently off by default).

Change-Id: I66267f227ea132dc86acb730e9882f85bead2cdb

6 years agoAdd use_svc condition to the scene detection in 1 pass.
Marco [Wed, 27 Sep 2017 21:49:58 +0000 (14:49 -0700)]
Add use_svc condition to the scene detection in 1 pass.

Scene detection is not currently used in SVC 1 pass code.
Speedup of ~0.4%.

Change-Id: I0ab769300919de710cd2da1402014fa3f22a1f86

6 years agoMerge "Revert "Remove the speed condition on scene detection in 1 pass code.""
Marco Paniconi [Wed, 27 Sep 2017 20:42:48 +0000 (20:42 +0000)]
Merge "Revert "Remove the speed condition on scene detection in 1 pass code.""

6 years agoMerge "vpxdsp: [x86] add highbd_d153_predictor functions"
Scott LaVarnway [Wed, 27 Sep 2017 20:40:21 +0000 (20:40 +0000)]
Merge "vpxdsp: [x86] add highbd_d153_predictor functions"

6 years agoRevert "Remove the speed condition on scene detection in 1 pass code."
Marco Paniconi [Wed, 27 Sep 2017 19:42:48 +0000 (19:42 +0000)]
Revert "Remove the speed condition on scene detection in 1 pass code."

This reverts commit 535b7b915ae5574db2f95632243cc5bee865f02e.

This is actually used in CBR to reset the rate control if high source sad is detected.

Original change's description:
> Remove the speed condition on scene detection in 1 pass code.
>
> Scene detection is used for VBR mode and for screen_content mode.
>
> It was also enabled for CBR mode via the speed condition,
> but currently the analysis in the scene detection is not used
> in CRB mode (similar computations are done locally at superblock level
> when the source_sad feature is enabled).
>
> For 1 pass code.
> No change in behavior. Small speed gain, ~0.5%.
>
> Change-Id: I59991d7ef2af320bea7af4b907596e057affa42f

TBR=marpan@google.com,builds@webmproject.org,jianj@google.com

Change-Id: Ib4e6b02047f75632503e7b0fc870af97fa9291c3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true

6 years agoMerge "fix signed integer overflow of idct"
James Zern [Wed, 27 Sep 2017 19:39:11 +0000 (19:39 +0000)]
Merge "fix signed integer overflow of idct"

6 years agoMerge "vp9_dx_iface: Stop using iter parameter incorrectly"
James Zern [Wed, 27 Sep 2017 18:37:20 +0000 (18:37 +0000)]
Merge "vp9_dx_iface: Stop using iter parameter incorrectly"

6 years agofix signed integer overflow of idct
Linfeng Zhang [Tue, 26 Sep 2017 19:33:40 +0000 (12:33 -0700)]
fix signed integer overflow of idct

Exposed by fuzz test in high bitdepth.
The bug is introduced in commit 64653fa.

BUG=webm:1466

Change-Id: Idd77d5c6a60efb9241471611ce1aba0646cb6ff5

6 years agovpxdsp: [x86] add highbd_d153_predictor functions
Scott LaVarnway [Wed, 27 Sep 2017 17:06:14 +0000 (10:06 -0700)]
vpxdsp: [x86] add highbd_d153_predictor functions

C vs SSE2 speed gains:
_4x4 : ~1.95x

C vs SSSE3 speed gains:
_8x8 : ~3.30x
_16x16 : ~5.67x
_32x32 : ~3.87x

BUG=webm:1411

Change-Id: Ib483989b25614aa89b635e8c087d0879a5d71904

6 years agoRemove the speed condition on scene detection in 1 pass code.
Marco [Wed, 27 Sep 2017 17:11:24 +0000 (10:11 -0700)]
Remove the speed condition on scene detection in 1 pass code.

Scene detection is used for VBR mode and for screen_content mode.

It was also enabled for CBR mode via the speed condition,
but currently the analysis in the scene detection is not used
in CRB mode (similar computations are done locally at superblock level
when the source_sad feature is enabled).

For 1 pass code.
No change in behavior. Small speed gain, ~0.5%.

Change-Id: I59991d7ef2af320bea7af4b907596e057affa42f

6 years agovp9_dx_iface: Stop using iter parameter incorrectly
Vignesh Venkatasubramanian [Mon, 25 Sep 2017 23:56:04 +0000 (16:56 -0700)]
vp9_dx_iface: Stop using iter parameter incorrectly

'iter' parameter is being checked for NULL in every call to
decoder_get_frame which is quite pointless because it is always
going to be NULL unless the application changed it. The code works
as described only because vp9_get_raw_frame returns -1 on all
subsequent calls after the first.

Change-Id: Ic736b9e8fe36fc1430fc11d6a9b292be02497248

6 years agoMerge "Add vpx_scaled_2d_neon()"
Linfeng Zhang [Wed, 27 Sep 2017 16:12:48 +0000 (16:12 +0000)]
Merge "Add vpx_scaled_2d_neon()"

6 years agoMerge "Add unit test to expose vp8 bug when width is set odd."
Jerome Jiang [Wed, 27 Sep 2017 01:26:59 +0000 (01:26 +0000)]
Merge "Add unit test to expose vp8 bug when width is set odd."

6 years agoMerge "vp8: [loongson] optimize copymen with mmi"
Shiyou Yin [Wed, 27 Sep 2017 00:49:28 +0000 (00:49 +0000)]
Merge "vp8: [loongson] optimize copymen with mmi"

6 years agoAdd unit test to expose vp8 bug when width is set odd.
Jerome Jiang [Thu, 21 Sep 2017 17:52:20 +0000 (10:52 -0700)]
Add unit test to expose vp8 bug when width is set odd.

BUG=b/64710201

Change-Id: Ia518af5494a42e80949cf1165244fbed59606cf7

6 years agoRemove the speed condition in setting compute_source_sad.
Marco [Tue, 26 Sep 2017 22:47:14 +0000 (15:47 -0700)]
Remove the speed condition in setting compute_source_sad.

The speed condition is not needed, feature can used for any
speed in 1 pass code.

Change-Id: I878ef3f63a075302eda48c0343fa243c80aab9ba

6 years agoReplace flag USE_ALTREF_FOR_ONE_PASS with speed feature.
Marco [Tue, 26 Sep 2017 17:18:43 +0000 (10:18 -0700)]
Replace flag USE_ALTREF_FOR_ONE_PASS with speed feature.

To be used for 1 pass VBR.
Off by default in speed features.

Change-Id: I5d6110d6d191990db526fe68ec9715379a4d1754

6 years agoMerge "SVC: Add setting for max_intra_rate_pct in sample encoder."
Marco Paniconi [Tue, 26 Sep 2017 16:28:30 +0000 (16:28 +0000)]
Merge "SVC: Add setting for max_intra_rate_pct in sample encoder."

6 years agoAdd vpx_scaled_2d_neon()
Linfeng Zhang [Tue, 19 Sep 2017 23:55:35 +0000 (16:55 -0700)]
Add vpx_scaled_2d_neon()

BUG=webm:1419

Change-Id: I39c8033734562efc0ac0e28e7f06fa05130f9b96

6 years agoMerge changes Ib9105462,Idfac00ed,If8d8a0e2
Linfeng Zhang [Tue, 26 Sep 2017 16:10:46 +0000 (16:10 +0000)]
Merge changes Ib9105462,Idfac00ed,If8d8a0e2

* changes:
  cosmetics: NEON scaling code
  Refactor convolve NEON code
  Refactor convolve code

6 years agovp8: [loongson] optimize copymen with mmi
Shiyou Yin [Wed, 6 Sep 2017 09:57:16 +0000 (17:57 +0800)]
vp8: [loongson] optimize copymen with mmi

1. vp8_copy_mem16x16_mmi
2. vp8_copy_mem8x8_mmi
3. vp8_copy_mem8x4_mmi

Change-Id: I3de29a11fa7402df0e48bbb944440b1e66498a65

6 years agoSVC: Add setting for max_intra_rate_pct in sample encoder.
Marco [Mon, 25 Sep 2017 20:36:25 +0000 (13:36 -0700)]
SVC: Add setting for max_intra_rate_pct in sample encoder.

Set it as default to 900.

Change-Id: Id2d990925dccff1f6762411c66ea95973440c92f

6 years agoMerge "vpxdsp: [x86] add highbd_d45_predictor functions"
Scott LaVarnway [Mon, 25 Sep 2017 11:34:14 +0000 (11:34 +0000)]
Merge "vpxdsp: [x86] add highbd_d45_predictor functions"

6 years agovpxdsp: [x86] add highbd_d45_predictor functions
Scott LaVarnway [Wed, 30 Aug 2017 16:27:44 +0000 (09:27 -0700)]
vpxdsp: [x86] add highbd_d45_predictor functions

C vs SSSE3 speed gains:
_4x4 : ~2.45x
_8x8 : ~10.61x
_16x16 : ~11.34x
_32x32 : ~6.36x

BUG=webm:1411

Change-Id: Ic91389a4f1a8ad093f498afe53765b897fb9be09

6 years agoMerge changes If59743aa,Ib046fe28,Ia2345752
James Zern [Fri, 22 Sep 2017 07:35:55 +0000 (07:35 +0000)]
Merge changes If59743aa,Ib046fe28,Ia2345752

* changes:
  Remove the unnecessary cast of (int16_t)cospi_{1...31}_64
  Remove the unnecessary upcasts of (int)cospi_{1...31}_64
  Change cospi_{1...31}_64 from tran_high_t to tran_coef_t

6 years agoMerge "Comma-separate VP9 encoder tmp.stt output"
Andrew Lewis [Thu, 21 Sep 2017 08:50:53 +0000 (08:50 +0000)]
Merge "Comma-separate VP9 encoder tmp.stt output"

6 years agoMerge "vp9: Modify pickmode early exit for ARF in 1pass."
Marco Paniconi [Thu, 21 Sep 2017 01:33:12 +0000 (01:33 +0000)]
Merge "vp9: Modify pickmode early exit for ARF in 1pass."

6 years agovp9: Modify pickmode early exit for ARF in 1pass.
Marco [Wed, 20 Sep 2017 21:55:31 +0000 (14:55 -0700)]
vp9: Modify pickmode early exit for ARF in 1pass.

Add the condition frames_since_golden > 0 to the
early exit check for ARF usage in nonrd_pickmode.
This improves quality of first frame following ARF, where
frame_since_golden = 0.

Small/neutral gain in metrics for speed 6, neutral change in speed.

Only affects when USE_ALTREF_FOR_ONE_PASS is enabled.

Change-Id: I82e73e6ff6fc849e5ca5448563cb8a0515fe0cdc

6 years agoRemove the unnecessary cast of (int16_t)cospi_{1...31}_64
Linfeng Zhang [Mon, 18 Sep 2017 16:33:31 +0000 (09:33 -0700)]
Remove the unnecessary cast of (int16_t)cospi_{1...31}_64

BUG=webm:1450

Change-Id: If59743aafe99226e0ec67ab5d20678ce25f53ab8

6 years agoRemove the unnecessary upcasts of (int)cospi_{1...31}_64
Linfeng Zhang [Wed, 13 Sep 2017 20:05:47 +0000 (13:05 -0700)]
Remove the unnecessary upcasts of (int)cospi_{1...31}_64

BUG=webm:1450

Change-Id: Ib046fe28caec5b9ebdc9d0152df7c54ff4266858

6 years agoChange cospi_{1...31}_64 from tran_high_t to tran_coef_t
Linfeng Zhang [Wed, 13 Sep 2017 00:13:17 +0000 (17:13 -0700)]
Change cospi_{1...31}_64 from tran_high_t to tran_coef_t

The unnecessary upcast to (int) will be cleaned later.

BUG=webm:1450

Change-Id: Ia234575206d5a74540526924b06ed3939322d063

6 years agoMerge "Bug fix: fadst4() in vp9/encoder/vp9_dct.c"
James Zern [Wed, 20 Sep 2017 21:12:45 +0000 (21:12 +0000)]
Merge "Bug fix: fadst4() in vp9/encoder/vp9_dct.c"

6 years agoBug fix: fadst4() in vp9/encoder/vp9_dct.c
Linfeng Zhang [Wed, 20 Sep 2017 16:18:04 +0000 (09:18 -0700)]
Bug fix: fadst4() in vp9/encoder/vp9_dct.c

A new bug was introduced in a80bdfd "Change sinpi_{1,2,3,4}_9 from
tran_high_t to int16_t". Reverted the change in this file.

BUG=webm:1450

Failed test C/TransHT.AccuracyCheck/26.

Change-Id: Id001f57aad811803ef7d367d2b2bc008d8499991

6 years agoMerge "vp9: Modify simple_block_yrd condition for SVC"
Marco Paniconi [Wed, 20 Sep 2017 16:42:31 +0000 (16:42 +0000)]
Merge "vp9: Modify simple_block_yrd condition for SVC"

6 years agoMerge "vpxdsp: [x86] add highbd_d63_predictor functions"
Scott LaVarnway [Wed, 20 Sep 2017 11:39:28 +0000 (11:39 +0000)]
Merge "vpxdsp: [x86] add highbd_d63_predictor functions"

6 years agotemporal_filter_apply_sse2.asm: add ':' to label
James Zern [Wed, 20 Sep 2017 01:59:11 +0000 (18:59 -0700)]
temporal_filter_apply_sse2.asm: add ':' to label

quiets nasm warning:
label alone on a line without a colon might be in error

BUG=webm:1462

Change-Id: I660407ca60e8c9a810dba9d76afb65852029a29c