Commit Graph

17098 Commits

Author SHA1 Message Date
Luca Barbato
d68d37872c ppc: dc_128 predictor 32x32
6x faster.

Change-Id: I1da8f51b4262871cb98f0aa03ccda41b0ac2b08b
2017-04-19 19:49:31 -07:00
Luca Barbato
f9d20e6df2 ppc: dc_128 predictor 16x16
20x faster.

Change-Id: I05f0deb2d38ae7966eae6b71fbc0aa51880e5709
2017-04-19 19:49:31 -07:00
Luca Barbato
0d9417de4a ppc: tm predictor 32x32
About 8x faster.

Change-Id: I9bad827ccbdf47ec95406e961c74ac2ff45f80cf
2017-04-19 19:49:26 -07:00
James Zern
a81f037f15 Merge changes I1f5a3752,I95123051,I3bb724e0,Ie81077fa,Ic80f3c05, ...
* changes:
  ppc: tm predictor 16x16
  ppc: tm predictor 8x8
  ppc: horizontal predictor 32x32
  ppc: horizontal predictor 16x16
  ppc: vertical intrapred 16x16 and 32x32
  configure: Workaround clang not enabling altivec on -mvsx
  configure: Match power*64* as ppc64
2017-04-20 02:45:45 +00:00
Linfeng Zhang
fbbdba3b04 Merge changes I9e18a73b,Ie47c8cd4
* changes:
  Clean CONVERT_TO_BYTEPTR/SHORTPTR in convolve
  Create CAST_TO_BYTEPTR/SHORTPTR
2017-04-19 23:55:58 +00:00
Linfeng Zhang
bf8a49abbd Clean CONVERT_TO_BYTEPTR/SHORTPTR in convolve
Replace by CAST_TO_BYTEPTR/SHORTPTR.
The rule is: if a short ptr is casted to a byte ptr, any offset
operation on the byte ptr must be doubled. We do this by casting to
short ptr first, adding offset, then casting back to byte ptr.

BUG=webm:1388

Change-Id: I9e18a73ba45ddae58fc9dae470c0ff34951fe248
2017-04-19 12:13:49 -07:00
Marco Paniconi
977356a72b Merge "vp9: Add phase to get averaging filter for 1:2 downsampling." 2017-04-19 15:27:55 +00:00
Marco
f34be01190 vp9: Fix the disabling of a SVC 3TL datarate test.
Change-Id: Ib42d23ab5ee39ab3c85e1d9a84e36249e59fe74e
2017-04-19 08:01:44 -07:00
Marco
348bdc0195 vp9: Add phase to get averaging filter for 1:2 downsampling.
The scaling filter with zero shift will give sub-sampling for
2x downsampling. Allow for a phase shift to get an averaging filter.

Usage is for source scaling in 1 pass SVC mode for 1:2 downscale.
Reduces aliasing in downsampled image.

Keep the phase to 0/off for now.

Change-Id: Ic547ea0748d151b675f877527e656407fcf4d51e
2017-04-18 16:56:15 -07:00
Luca Barbato
479443a570 ppc: tm predictor 16x16
About 10x faster.

Change-Id: I1f5a3752d346459df3b45f92963208bf3e520f06
2017-04-19 01:48:10 +02:00
Luca Barbato
c8f5a55df4 ppc: tm predictor 8x8
About 5x faster.

Change-Id: I951230517f49c0dca9ac9eac2efa8916a303b85a
2017-04-19 01:48:09 +02:00
Luca Barbato
7b0e12934e ppc: horizontal predictor 32x32
About 5x faster.

Change-Id: I3bb724e07baffd901aa2d0f65060ba48882cc9b8
2017-04-19 01:48:09 +02:00
Luca Barbato
a7a2d1653b ppc: horizontal predictor 16x16
About 10x faster.

Change-Id: Ie81077fa32ad214cdb46bdcb0be4e9e2c7df47c2
2017-04-19 01:48:09 +02:00
Luca Barbato
7ad1faa6f8 ppc: vertical intrapred 16x16 and 32x32
Change-Id: Ic80f3c050cfbe7697e81a311b4edaaa597b85cab
2017-04-19 01:48:09 +02:00
Luca Barbato
a39b723eb3 configure: Workaround clang not enabling altivec on -mvsx
The flag `-mvsx` implies `-maltivec`.

Change-Id: I7544553eba131a533467b387f8bf329d57f5af5c
2017-04-19 01:48:04 +02:00
Luca Barbato
3252e6b63d configure: Match power*64* as ppc64
Change-Id: Ie640dff50a5db935bb57c5a2570b423ce8946f2c
2017-04-19 01:47:56 +02:00
Linfeng Zhang
a02f391cbe Create CAST_TO_BYTEPTR/SHORTPTR
They will replace CONVERT_TO_BYTEPTR/SHORTPTR module by module.

BUG=webm:1388

Change-Id: Ie47c8cd4897696481b9cbbf9e2d439dc22dc85ec
2017-04-18 14:48:11 -07:00
Marco
15afee1938 vp9: Disable some SVC tests for now.
Disable the 1 pass CBR SVC tests with temporal_layers > 1.
Issue with the commit 863f860, which will cause encoder/decoder
mismatch due to skipping encoder loopfilter for non-reference frames.

Will re-enable the tests when fixed.

Change-Id: I74918a0045a17976b069c4be63fbeb921974df0d
2017-04-18 09:51:42 -07:00
Marco
ad2e3598d2 vp9: Add key_frame condition to is_reference check for loopfilter.
This condiiton is not needed as key_frame should set the refresh
of the reference frames, but good to have for clarity in condition.

Change-Id: Icf9838e7e4f0ff5cf0a9562ae3b5d6c7e6f78702
2017-04-17 15:18:46 -07:00
Johann Koenig
a6095333a7 Merge "re-enable vpx_comp_avg_pred_sse2" 2017-04-17 22:07:34 +00:00
Marco Paniconi
9aa429a66d Revert "Revert "vp9: Avoid encoder loopfilter for non-reference frames.""
This reverts commit e9b7f98c56.

Reason for revert:
Commit d578bdad fixes the issue (encoder/decoder mismatch
in 3TL datarate test) that causes the original revert.

Original change's description:
> Revert "vp9: Avoid encoder loopfilter for non-reference frames."
>
> This reverts commit 863f860bfc.
>
> This causes encoder / decoder mismatches in various
> VP9/DatarateTestVP9Large.BasicRateTargeting3TemporalLayers tests
>
> BUG=webm:1408
>
> Change-Id: Ic200c39d7ed9c0b0247ef562f5d6f7b2625f7e14
>

TBR=jzern@google.com,marpan@google.com,builds@webmproject.org,jianj@google.com
BUG=webm:1408

Change-Id: Ifeb81460856d1d56482d4e0477a70ee98f8bfaa6
2017-04-17 11:02:02 -07:00
Marco
d578bdad02 vp9: Datarate test: modify frame flags for 3 TL.
Modify the frame flags to update the ARF on top layer,
for the tests:
VP9/DatarateTestVP9Large.BasicRateTargeting3TemporalLayers
VP9/DatarateTestVP9Large.BasicRateTargeting3TemporalLayersFrameDropping

This is needed to fix the encode/decoder mismatches caused by 863f860,
and removed in the revert e9b7f98.

Change-Id: I6b9fecfdd17315fc0179e29949338c77636026c0
2017-04-17 09:33:20 -07:00
Johann
9fa24f03b5 re-enable vpx_comp_avg_pred_sse2
Buffers on 32 bit x86 builds only guaranteed 8 byte alignment. Fixed
with "AvgPred test: use aligned buffers" and "sad avg: align
intermediate buffer"

Also re-enable asserts on the C version.

BUG=webm:1390

Change-Id: I93081f1b0002a352bb0a3371ac35452417fa8514
2017-04-17 08:40:43 -07:00
Johann Koenig
9e19102972 Merge "AvgPred test: use aligned buffers" 2017-04-17 15:36:41 +00:00
Johann
069b772915 sad avg: align intermediate buffer
comp_avg_pred has started declaring a requirement for aligned buffers.

BUG=webm:1390

Change-Id: Idaf6667498ea343e8d49b32bc9d8b9d0aa43ef5c
2017-04-17 14:26:33 +00:00
James Zern
4ba20da8b1 Merge "Add AVX2 optimization to copy/avg functions" 2017-04-15 00:26:08 +00:00
Yi Luo
aa5a941992 Add AVX2 optimization to copy/avg functions
Change-Id: Ibcef70e4fead74e2c2909330a7044a29381a8074
2017-04-14 16:50:10 -07:00
Johann Koenig
7178e68bbe Merge "Disable vpx_comp_avg_pred_sse2" 2017-04-14 22:01:39 +00:00
Johann
e3b2710b04 AvgPred test: use aligned buffers
BUG=webm:1390

Change-Id: Idb6d1ce119a09c5e7c9f3c58bbbae3de63463d1d
2017-04-14 12:49:56 -07:00
James Zern
e9b7f98c56 Revert "vp9: Avoid encoder loopfilter for non-reference frames."
This reverts commit 863f860bfc.

This causes encoder / decoder mismatches in various
VP9/DatarateTestVP9Large.BasicRateTargeting3TemporalLayers tests

BUG=webm:1408

Change-Id: Ic200c39d7ed9c0b0247ef562f5d6f7b2625f7e14
2017-04-14 11:50:06 -07:00
Marco Paniconi
b937f1c839 Merge "vp9: SVC: fix to allow use_base_mv to be used for 3 layers." 2017-04-14 17:12:58 +00:00
Johann
eaa7cdf05d Disable vpx_comp_avg_pred_sse2
Failures on windows:
unknown file: error: SEH exception with code 0xc0000005 thrown in the
test body.

Alignment check errors on linux:
test_libvpx: ../libvpx/vpx_dsp/variance.c:230: void
vpx_comp_avg_pred_c(uint8_t *, const uint8_t *, int, int, const uint8_t
*, int): Assertion `((intptr_t)comp_pred & 0xf) == 0' failed.

BUG=webm:1390

Change-Id: I5eed5381c0f1a8fe594a128eb415e77232f544ea
2017-04-14 08:43:06 -07:00
Johann Koenig
bdb593ab20 Merge "vpx_comp_avg_pred: sse2 optimization" 2017-04-14 04:10:56 +00:00
Marco
adb9b4eddf vp9: SVC: fix to allow use_base_mv to be used for 3 layers.
Allow use_base_mv to be used for 3 spatial layers where
base is 4x4 scale from the top layer.

Change-Id: If6641baf8b8e4d0fd5dc67619d873c6d75065f43
2017-04-13 20:43:43 -07:00
Marco Paniconi
f0ccaff553 Merge "vp9: Avoid encoder loopfilter for non-reference frames." 2017-04-14 00:45:42 +00:00
Marco
6bff6cb5a9 vp9: 1 pass VBR: Fix to rate control at low min-q.
Fix to avoid getting stuck at very low Q even
though content is changing, which can happen for --min-q=0.

Fix is to more aggressively increase active_worst_quality
when detecting significant rate_deviation at very low Q.

Change will only affect 1 pass VBR for --min-q < 4, so no
change in ytlive metrics for --min-q >= 4.

Change-Id: I4dd77dd7c08a30a4390da0ff2c8bda6fccfa76d7
2017-04-13 11:44:35 -07:00
Marco
863f860bfc vp9: Avoid encoder loopfilter for non-reference frames.
Useful for SVC, where the top layer enhancement frames may
not update any reference buffers, as is the case for the
patterns in the 1 pass CBR SVC when #temporal_layers > 1.

~3% encoder speedup for SVC patterns with temporal layers
in 1 pass CBR mode.

Updated the SVC datarate tests for the mismatch frames.
Set the frame-dropper off in some tests with #temporal_layers > 1
so we can correctly set #mismatch frames. Adjusted rate target
threshold for tests where frame-dropper was turned off.

Change-Id: Ia0c142f02100be0fed61cd2049691be9c59d6793
2017-04-13 09:51:55 -07:00
Johann
28a8622143 vpx_comp_avg_pred: sse2 optimization
Provides over 15x speedup for width > 8.

Due to smaller loads and shifting for width == 8 it gets about 8x
speedup.

For width == 4 it's only about 4x speedup because there is a lot of
shuffling and shifting to get the data properly situated.

BUG=webm:1390

Change-Id: Ice0b3dbbf007be3d9509786a61e7f35e94bdffa8
2017-04-13 08:44:52 -07:00
Yunqing Wang
f22b828d68 Fix an integer overflow in vp9_mcomp.c
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
2017-04-10 18:09:50 -07:00
Jerome Jiang
2420f44342 Merge "vp9: speed >= 8: Adjust speed settings on ARM." 2017-04-11 00:45:21 +00:00
Jerome Jiang
f16f08e55b vp9: speed >= 8: Adjust speed settings on ARM.
Set adaptive_rd_thresh to 2 when simple block yrd is not used.

Fix regression caused by computing y sad without
int_pro_motion_estimation on low res motion clips.

Overall 0.07% quality loss on rtc_derf.

Change only affects low res on speed 8.

Change-Id: Ic6a188a56529f1034d6431005fb4b0e24e8a7e27
2017-04-11 00:26:56 +00:00
Marco
6557baf336 vp9: 1 pass CBR: avoid nonrd_pick_partition on segment.
For speed 5, 1 pass CBR: Don't use the nonrd_pick_partition
on the segment, rather use choose_partitioning followed by
nonrd_select_partition (as is done on base segment).

Little/no quality loss on RTC and RTC_derf (< 0.3%),
speedup of at least 5%.

Change-Id: I5273d5f950e60adf5e437b4ca8c4f63964641e83
2017-04-10 15:02:49 -07:00
Marco Paniconi
ff1fef9607 Merge "vp9: Fix to noise estimation for temporal denoising." 2017-04-07 17:13:22 +00:00
Yunqing Wang
f496032686 Merge "VP9 motion vector unit test" 2017-04-07 16:46:22 +00:00
Marco
349c3118bd vp9: Fix to noise estimation for temporal denoising.
If the noise estimation is avoided due to large motion,
the last_source for denoising should still be updated.

Change-Id: I67155ea7dbe9ac2785978e64a27bdafd7d57aac0
2017-04-07 09:23:30 -07:00
Marco
18b54ef468 vp9: Adjust consec_zeromv threshold for aq-mode=3.
To reduce refresh on partial super-blocks on boundary,
for noisy input. Reduces some artifacts on noisy input.

Change-Id: I10b5808a296874e08c7f378b3df58466591d8dbe
Edit
2017-04-07 08:54:09 -07:00
James Zern
04e9456567 Merge changes from topic 'Wshorten'
* changes:
  configure: enable -Wshorten-64-to-32 for hbd
  vp9_encodeframe: resolve -Wshorten-64-to-32 in hbd
  Resolve -Wshorten-64-to-32 in highbd variance.
2017-04-07 07:32:14 +00:00
Jerome Jiang
6af42f5102 Merge "Fix compile warnings with enable-internal-stats flag." 2017-04-07 03:34:55 +00:00
Jerome Jiang
b82b574e76 Fix compile warnings with enable-internal-stats flag.
BUG=webm:1402

Change-Id: Ibe9ecb1b559a4b989f6ccedbd097e369f6edde1e
2017-04-06 14:00:01 -07:00
Marco
3227a9be5f vp9; Move the denoising condition for speed 5.
Move the condition for effectively disabling the denoising
for speed 5 into the vp9_denoiser_denoise().

This is cleaner, and also moving the condition into vp9_denoiser_denoise
will keep the denoiser buffer updated with the current source.
This allows for more consistent behavior if speed is changed midstream.

Change-Id: Ia001f591c56e454bf724c3ae73c024badb183ef8
2017-04-06 11:03:04 -07:00