James Zern [Tue, 5 Apr 2022 02:52:06 +0000 (02:52 +0000)]
Merge changes I0b6520be,I1f006daa,I7ee8e367 into main
* changes:
vp9[loongarch]: Optimize vpx_convolve8_avg_horiz_c
vp8[loongarch]: Optimize dequant_idct_add_y/uv_block
loongarch: Fix bugs
Johann Koenig [Mon, 4 Apr 2022 00:59:23 +0000 (00:59 +0000)]
Merge "remove unused vp8_encode_intra parameter" into main
James Zern [Thu, 31 Mar 2022 21:54:24 +0000 (21:54 +0000)]
Merge changes I4c63beeb,I28f3b98f into main
* changes:
Revert "quantize: replace highbd versions"
Revert "quantize: remove highbd version"
Johann Koenig [Thu, 31 Mar 2022 21:43:47 +0000 (21:43 +0000)]
Merge "subpel variance: add speed test" into main
James Zern [Thu, 31 Mar 2022 19:11:01 +0000 (12:11 -0700)]
Revert "quantize: replace highbd versions"
This reverts commit
2200039d33c49a9f7a5c438656df143755b022c4.
This causes failures with VP9/EndToEndTestLarge.EndtoEndPSNRTest/*; it
seems the assembly does not match the C code.
Bug: webm:1586
Change-Id: I4c63beebf88d4c12789d681b0d38014510b147fe
James Zern [Thu, 31 Mar 2022 19:10:03 +0000 (12:10 -0700)]
Revert "quantize: remove highbd version"
This reverts commit
89cfe3835c47dabf77d38edb3af190155984fa9a.
This is a prerequisite for reverting
2200039d33c49a9f7a5c438656df143755b022c4 which causes high bitdepth test
failures
Bug: webm:1586
Change-Id: I28f3b98f3339f3573b1492b88bf733dade133fc0
yuanhecai [Wed, 30 Mar 2022 06:04:10 +0000 (14:04 +0800)]
vp9[loongarch]: Optimize vpx_convolve8_avg_horiz_c
1. vpx_convolve8_avg_horiz_lsx
Bug: webm:1755
Change-Id: I0b6520be0afa1689da329f56ec6cd95c1730250c
yuanhecai [Mon, 28 Mar 2022 09:12:57 +0000 (17:12 +0800)]
vp8[loongarch]: Optimize dequant_idct_add_y/uv_block
1. vp8_dequant_idct_add_uv_block_lsx
2. vp8_dequant_idct_add_y_block_lsx
Bug: webm:1755
Change-Id: I1f006daaefb2075b422bc72a3f69c5abee776e2e
yuanhecai [Tue, 22 Mar 2022 05:58:50 +0000 (13:58 +0800)]
loongarch: Fix bugs
Fix bugs from loopfilter_filters_lsx.c, vpx_convolve8_avg_lsx.c
Bug: webm:1755
Change-Id: I7ee8e367d66a49f3be10d7e417837d3b6ef50bdb
Johann Koenig [Thu, 31 Mar 2022 05:45:23 +0000 (05:45 +0000)]
Merge "quantize: remove highbd version" into main
Johann [Wed, 30 Mar 2022 06:33:40 +0000 (15:33 +0900)]
quantize: remove highbd version
The only difference between the code is the clamp. For
8 bit it is purely an optimization. The values outside
this range will still saturate.
Change-Id: I2a770b140690d99e151b00957789bd72f7a11e13
Johann [Mon, 4 Nov 2019 21:58:07 +0000 (15:58 -0600)]
remove unused vp8_encode_intra parameter
Follow it up and also remove it from other functions.
BUG=webm:1612
Change-Id: I9d3cb785ab0d68c6fcae185043c896d8a135e284
James Zern [Thu, 31 Mar 2022 01:53:51 +0000 (01:53 +0000)]
Merge "Optimize FHT functions for NEON" into main
Johann [Thu, 31 Mar 2022 01:43:29 +0000 (10:43 +0900)]
subpel variance: add speed test
Was used to verify assembly speed versus an attempt to rewrite
in intrinsics.
Change-Id: I011fe5494334b8fcda04b9d54c6093dbcfc55710
Johann Koenig [Thu, 31 Mar 2022 00:45:31 +0000 (00:45 +0000)]
Merge "remove sad x3,x8 specializations" into main
Johann [Wed, 30 Mar 2022 05:57:46 +0000 (14:57 +0900)]
quantize: replace highbd versions
The optimized quantize functions were already built to handle
highbd values. The only difference is the clamping. All highbd
functions expand to 32bits when running in highbd mode.
Removes vpx_highbd_quantize_32x32_sse2 as it is slower than the
C version in the worst case.
Bug: webm:1586
Change-Id: I49bf8a6a2041f78450bf43a4f655c67656b0f8d9
Konstantinos Margaritis [Wed, 16 Mar 2022 14:27:27 +0000 (16:27 +0200)]
Optimize FHT functions for NEON
[NEON]
Optimize vp9_fht4x4, vp9_fht8x8, vp9_fht16x16 for NEON
Following change #3516278, the improvement for these functions is:
Before:
4.10% 0.75% vpxenc vpxenc [.] vp9_fht16x16_c
2.93% 0.65% vpxenc vpxenc [.] vp9_fht8x8_c
0.93% 0.77% vpxenc vpxenc [.] vp9_fht4x4_c
And after the patch:
0.69% 0.16% vpxenc vpxenc [.] vp9_fht16x16_neon
0.28% 0.28% vpxenc vpxenc [.] vp9_fht8x8_neon
0.54% 0.53% vpxenc vpxenc [.] vp9_fht4x4_neon
Bug: webm:1634
Change-Id: I6748a0c4e0cfaafa3eefdd4848d0ac3aab6900e4
James Zern [Wed, 30 Mar 2022 03:00:00 +0000 (20:00 -0700)]
sad4d_avx2: fix VS 2014 build error
after:
d60b671a7 gcc 11 warning: mismatched bound
error C2719: 'sums': formal parameter with requested alignment of 32
won't be aligned
Change-Id: Iaba46d00ef2334a5e2d9ee69b5d03478fdc73a60
Johann [Tue, 29 Mar 2022 03:40:12 +0000 (12:40 +0900)]
remove skip_block from quantize
Whether a block is skipped is handled by mi->skip. x->skip_block
is kept exclusively to verify that the quantize functions are not
called for skip blocks.
Finishes the cleanup in
13eed991f
Bug: libvpx:1612
Change-Id: I1598c3b682d3c5e6c57a15fa4cb5df2c65b3a58a
Johann [Mon, 28 Mar 2022 06:27:46 +0000 (15:27 +0900)]
remove sad x3,x8 specializations
These would compute the sum of absolute differences (sad) for a
group of 3 or 8 references. This was used as part of an exhaustive
search.
vp8 only uses these functions in speed 0 and best quality.
For vp9 this is only used with the --enable-non-greedy-mv
experiment.
This removes the 3- and 8-at-a-time optimized functions and uses
the fall back code which will process 1 or 4 (vpx_sadMxNx4d) at
a time.
For configure --target=x86_64-linux-gcc --enable-realtime-only:
libvpx.a
before: 3002424 after: 2937622 delta: 64802
after 'strip libvpx.a'
before: 2116998 after: 2073090 delta: 43908
Change-Id: I566d06e027c327b3bede68649dd551bba81a848e
Johann Koenig [Tue, 29 Mar 2022 03:18:02 +0000 (03:18 +0000)]
Merge "gcc 11 warning: mismatched bound" into main
Johann [Wed, 23 Mar 2022 05:28:29 +0000 (14:28 +0900)]
gcc 11 warning: mismatched bound
Clean up a new build warning with gcc11:
argument 3 of type ‘const uint8_t * const[]’ with
mismatched bound [-Warray-parameter=]
Standardize sad functions with array sizes.
Change-Id: Iea4144e61368f6a8279e2f3ae96c78aff06c8b41
James Zern [Sat, 26 Mar 2022 17:25:18 +0000 (10:25 -0700)]
ads2armasm_ms.pl: fix thumb::FixThumbInstructions call
broken since:
642529248 ads2gas[_apple].pl: remove unused stanzas
Change-Id: I1eac77e2fe23cc3f162251e9e0102a4909f7b997
James Zern [Thu, 24 Mar 2022 19:39:10 +0000 (19:39 +0000)]
Merge "Make sure only NEON FDCT functions are called." into main
Johann [Wed, 23 Mar 2022 05:18:58 +0000 (14:18 +0900)]
ads2gas: maintain whitespace
Don't use tabs during conversion. Save and restore
existing spacing.
Change-Id: Ib8f443db542c091d36e9ab9836e3e3e292d711f7
Johann [Wed, 23 Mar 2022 04:58:46 +0000 (13:58 +0900)]
ads2gas: fix .size measurement
The distance between PROC and END is used to generate .size
information for debugging. When the leading underscore was
removed the pattern used to match the function name broke.
Change-Id: I90bf67d95ecdc2d214606e663773f88d2a2d6b9c
Johann Koenig [Wed, 23 Mar 2022 04:24:28 +0000 (04:24 +0000)]
Merge "ads2gas*.pl: strip trailing whitespace after transforms" into main
James Zern [Tue, 22 Mar 2022 20:51:27 +0000 (13:51 -0700)]
ads2gas*.pl: strip trailing whitespace after transforms
Change-Id: I0bea977b256e464231706c72cc14a5c8b6e90775
Jerome Jiang [Tue, 22 Mar 2022 20:07:31 +0000 (13:07 -0700)]
Fix ClangTidy style warning
Change-Id: I6c4711e488cda6b97af96d5e1b6b249786e709de
Konstantinos Margaritis [Fri, 11 Mar 2022 18:19:25 +0000 (20:19 +0200)]
Make sure only NEON FDCT functions are called.
[NEON]
Added vpx_fdct4x4_pass1_neon(),
Added vpx_fdct8x8_pass1_notranspose_neon(),
Added vpx_fdct8x8_pass1_neon() to avoid code duplication
Refactored vpx_fdct4x4_neon() and vpx_dct8x8_neon() to use the above
Rename dct_body to vpx_fdct16x16_body to reuse later
Add transpose_s16_16x16()
I have run make test and all tests/configurations seem to pass.
Profiled using this command on an Ampere Altra VM:
sudo perf record -g ./vpxenc --codec=vp9 --height=1080 --width=1920 \
--fps=25/1 --limit=20 -o output.mkv \
../original_videos_Sports_1080P_Sports_1080P-0063.mkv --debug –rt
Before this optimization:
1.32% 1.32% vpxenc vpxenc [.] vpx_fdct4x4_neon
0.16% 0.16% vpxenc vpxenc [.] vpx_fdct4x4_c
0.79% 0.79% vpxenc vpxenc [.] vpx_fdct8x8_c
0.52% 0.52% vpxenc vpxenc [.] vpx_fdct8x8_neon
1.23% 1.23% vpxenc vpxenc [.] vpx_fdct16x16_c
0.54% 0.54% vpxenc vpxenc [.] vpx_fdct16x16_neon
So, even though a _neon() version exists, the C version was called \
as well. After this patch:
1.42% 1.36% vpxenc vpxenc [.] vpx_fdct4x4_neon
0.87% 0.82% vpxenc vpxenc [.] vpx_fdct8x8_neon
0.74% 0.74% vpxenc vpxenc [.] vpx_fdct16x16_neon
Change-Id: Id4e1dd315c67b4355fe4e5a1b59e181a349f16d0
yuanhecai [Fri, 11 Mar 2022 02:56:07 +0000 (10:56 +0800)]
vp8[loongarch]: Optimize idct_add, filter_bv/bh
1. vp8_dc_only_idct_add_lsx
2. vp8_loop_filter_bh_lsx
3. vp8_loop_filter_bv_lsx
Bug: webm:1755
Change-Id: I9b629767e2a4e9db8cbb3ee2369186502dc6eb00
yuanhecai [Thu, 10 Mar 2022 06:56:42 +0000 (14:56 +0800)]
vp9[loongarch]: Optimize convolve/convolve8_avg_c
1. vpx_convolve8_avg_lsx
2. vpx_convolve_avg_lsx
Bug: webm:1755
Change-Id: I4af5c362a94f11d0b5d1760e18326660bdbc0559
yuanhecai [Tue, 1 Mar 2022 08:33:47 +0000 (16:33 +0800)]
vp9[loongarch]: Optimize convolve8_horiz/vert/c
1. vpx_convolve8_lsx
2. vpx_convolve8_vert_lsx
3. vpx_convolve8_horiz_lsx
Bug: webm:1755
Change-Id: I9897e1ed6a904ac74d1078bd22b275af44db142d
Johann [Sat, 12 Mar 2022 22:02:03 +0000 (07:02 +0900)]
ads2gas_apple.pl: remove gcc-isms
The gcc assembler was incompatible for a long
time. It is now based on clang and accepts
more modern syntax, although not enough to
remove the script entirely.
Change-Id: I667d29dca005ea02a995c1025c45eb844081f64b
Johann [Sat, 12 Mar 2022 21:28:16 +0000 (06:28 +0900)]
ads2gas[_apple].pl: remove unused stanzas
Many of the features in ads2gas are no longer used.
Remove all patterns which are no longer used in
libvpx.
Simplify between the two to minimize differences.
Change-Id: Ia1151eb8b694cbe51845a1374a876cc7b798899c
James Zern [Thu, 3 Mar 2022 18:35:53 +0000 (18:35 +0000)]
Merge "vp9[loongarch]: Optimize horizontal/vertical_8_c" into main
yuanhecai [Tue, 1 Mar 2022 01:48:13 +0000 (09:48 +0800)]
vp9[loongarch]: Optimize horizontal/vertical_8_c
1. vpx_lpf_vertical_8_lsx
2. vpx_lpf_horizontal_8_lsx
Bug: webm:1755
Change-Id: I6b05d6b1b2ac4d2a75beb9c9ca9700976fc3af55
Marco Paniconi [Wed, 2 Mar 2022 07:03:27 +0000 (23:03 -0800)]
vp9-svc: Remove VP9E_SET_TEMPORAL_LAYERING_MODE
The control was never implemented, no need to keep this.
temporal_layering_mode is set in the config.
Bug: webm:1753
Change-Id: I9a6eb50e82344605ab62775911783af82ac2d401
yuanhecai [Wed, 23 Feb 2022 03:23:52 +0000 (11:23 +0800)]
vp9[loongarch]: Optimize lpf_horizontal/vertical_16_dual with LSX
Change-Id: I82c6bc16ea57c3f7ac5f4d212a12a5f70cb55ffc
James Zern [Fri, 11 Feb 2022 20:43:29 +0000 (12:43 -0800)]
svc_datarate_test.cc: remove stale TODO
Bug: webm:1554
Change-Id: I547223763b86c6a24fa32851f7b30ebab4b7472a
Gregor Jasny [Thu, 10 Feb 2022 08:01:49 +0000 (09:01 +0100)]
support visual studio 2022 (vs17)
Change-Id: I8380283d09b0c90183f224399f953dcc527181c5
Marco Paniconi [Wed, 9 Feb 2022 02:41:23 +0000 (02:41 +0000)]
Merge "rtc-vp9: Fix intra-only for bypass mode" into main
Marco Paniconi [Tue, 8 Feb 2022 20:36:39 +0000 (12:36 -0800)]
rtc-vp9: Fix intra-only for bypass mode
Allow intra-only frame in svc to also work
in bypass (flexible-svc) mode.
Added unittest for the flexible svc case.
And fix the gld_fb_idx for (SL0, TL1) in bypass/flexible
mode pattern in the sample encoder: force it to be 0
(same as lst_fb_idx), since the slot is unused on SL0.
Change-Id: Iada9d1b052e470a0d5d25220809ad0c87cd46268
Lu Wang [Wed, 10 Nov 2021 07:21:17 +0000 (15:21 +0800)]
vp8[loongarch]: Optimize vp8_loop/sixtap, vpx_dc with LSX.
1. vp8_loop_filter_mbh, vp8_loop_filter_mbv
2. vp8_sixtap_predict16x16, vp8_sixtap_predict8x8
3. vpx_dc_predictor_16x16, vpx_dc_predictor_8x8
./vpxdec --progress -o YUV_1920X1080.yuv original_1200f/VP8_1920X1080.webm
before: 37.77fps
after : 220.90fps
Bug: webm:1755
Change-Id: I1a3ce16f0c872261d813b6531cfdf25bd59bb774
Lu Wang [Wed, 10 Nov 2021 07:05:42 +0000 (15:05 +0800)]
vpx_util[loongarch]: Add loongson_intrinsics.h v1.0.5.
Bug: webm:1755
Change-Id: Id2fa999bdb8788bd4285114c748c547fa262a95e
Wan-Teh Chang [Tue, 8 Feb 2022 02:20:35 +0000 (02:20 +0000)]
Merge "Handle NV12 in vpx_img_chroma_subsampling()" into main
Wan-Teh Chang [Sat, 5 Feb 2022 22:32:49 +0000 (14:32 -0800)]
Handle NV12 in vpx_img_chroma_subsampling()
Change-Id: Ibac9f6f8dcdcae0d0c10ae1a118d13baf2407270
Wan-Teh Chang [Sat, 5 Feb 2022 20:14:37 +0000 (12:14 -0800)]
Update error messages in validate_img()
Change-Id: I4aa6d2e16e077d29e4e9eabfc7056fcfed6786d6
Marco Paniconi [Thu, 3 Feb 2022 17:04:46 +0000 (17:04 +0000)]
Merge "rtc-vp9: Fix to tests for intra-only frame." into main
Marco Paniconi [Thu, 3 Feb 2022 00:17:58 +0000 (16:17 -0800)]
rtc-vp9: Fix to tests for intra-only frame.
Fix some issues with the test, and add new
test that verifies that we can decode base stream
startinig at middle of sequence where intra-only
frame is inserted.
Change-Id: I398d23927113eb58ef64694feca25e60ce60a5f7
James Zern [Wed, 2 Feb 2022 00:24:07 +0000 (16:24 -0800)]
vp9_roi_test: apply iwyu
Change-Id: I715c27e329495940d989f95df65ac10e021261d2
James Zern [Tue, 1 Feb 2022 19:57:05 +0000 (11:57 -0800)]
vp9_thread_test: parameterize VP9DecodeMultiThreadedTest
on a per-file basis; this will make sharding more effective
Change-Id: Ib797681a7cc3bd7ec835bb0c1c7a8d9f23512a0d
Jerome Jiang [Mon, 31 Jan 2022 15:17:22 +0000 (15:17 +0000)]
Merge "Use background segmentation mask with ROI" into main
James Zern [Sat, 29 Jan 2022 03:20:41 +0000 (03:20 +0000)]
Merge "vpx/vp8[cd]x.h,cosmetics: normalize ctrls to enum order" into main
James Zern [Fri, 28 Jan 2022 19:47:08 +0000 (11:47 -0800)]
vpx/vp8[cd]x.h,cosmetics: normalize ctrls to enum order
Change-Id: I49bbd956b3a64008d1abe54de87d7831bc3eede6
Jin Bo [Tue, 6 Jul 2021 09:18:48 +0000 (17:18 +0800)]
libvpx[loongarch]: Add loongarch support.
LSX and LASX are enabled by default if compiler supports them.
Bug: webm:1754
Change-Id: Ic36b113bc4313c50e9d2bbab91199b3aa46d00dc
James Zern [Thu, 27 Jan 2022 21:16:55 +0000 (21:16 +0000)]
Merge changes I2db20130,I4e643c83 into main
* changes:
vp8dx.h,cosmetics: normalize #define/type order
vp8dx.h: add missing define for VP9_SET_BYTE_ALIGNMENT
James Zern [Thu, 27 Jan 2022 21:16:08 +0000 (21:16 +0000)]
Merge "fix some include guards" into main
Jerome Jiang [Tue, 28 Sep 2021 23:59:21 +0000 (16:59 -0700)]
Use background segmentation mask with ROI
RTC sample encoder vpx_temporal_svc_encoder can take mask files as input
when ROI_MAP is set to 1.
Uses ROI and segmentation of vp9 to skip background encoding when
source_sad is low and the correspond block in previous frame is also
skipped.
Change-Id: I8590e6f9a88cecfa1d7f375d4cc480f0f2af87b6
James Zern [Thu, 27 Jan 2022 03:44:33 +0000 (19:44 -0800)]
vp8dx.h,cosmetics: normalize #define/type order
Change-Id: I2db20130cc366bead5e576b375479917f9aee024
James Zern [Thu, 27 Jan 2022 03:41:41 +0000 (19:41 -0800)]
vp8dx.h: add missing define for VP9_SET_BYTE_ALIGNMENT
Change-Id: I4e643c837bb010bd58f4fc8179045f8df18f8ae1
James Zern [Wed, 26 Jan 2022 23:05:22 +0000 (15:05 -0800)]
fix some include guards
Change-Id: I0233d352c134bdda3ca160d41b4671d1c45ab01c
James Zern [Wed, 26 Jan 2022 18:40:27 +0000 (18:40 +0000)]
Merge "libwebm: update to libwebm-1.0.0.28-28-gee0bab5" into main
James Zern [Wed, 26 Jan 2022 04:06:59 +0000 (20:06 -0800)]
libwebm: update to libwebm-1.0.0.28-28-gee0bab5
https://chromium.googlesource.com/webm/libwebm/+log/
206d268d4d8066e5a37c49025325b80c95c771dd..
ee0bab576c338c9807249b99588e352b7268cb62
only one commit affects this snapshot:
ee0bab5 Revert "mkvmuxer,Cluster::Size: make uint64 conversion explicit"
Change-Id: Ib1f21fc5589098af346d110ff88c94bb1ba0a027
Jianhui Dai [Sat, 1 Jan 2022 00:01:48 +0000 (08:01 +0800)]
Reland "Add vp9 ref frame to flag map function"
Original change's description:
> Add vp9 ref frame to flag map function
>
> Change-Id: I371c2346b9e0153c0f8053cab399ce14cd286c56
Change-Id: I04a407ee0ef66c01a0d224b4468e043213f8791f
Jerome Jiang [Tue, 11 Jan 2022 16:47:52 +0000 (08:47 -0800)]
Revert "Set unused reference frames to first ref"
This reverts commit
e7f33a53cf404bbb3688af9b13375b5c090daae4.
Change-Id: I54e807220885cb78af6f3c6e48b3eb2c9f1e70b4
Jerome Jiang [Tue, 11 Jan 2022 16:46:59 +0000 (08:46 -0800)]
Revert "Add vp9 ref frame to flag map function"
This reverts commit
44e611482e13fdffa0acde780a20dd68ee153498.
Change-Id: Ic900cc01be4de7983fab42178a488277efab77b3
Jianhui Dai [Sat, 1 Jan 2022 00:01:48 +0000 (08:01 +0800)]
Add vp9 ref frame to flag map function
Change-Id: I371c2346b9e0153c0f8053cab399ce14cd286c56
James Zern [Wed, 22 Dec 2021 03:03:11 +0000 (03:03 +0000)]
Merge "vp9_prob_diff_update_savings_search_model: quiet conv warnings" into main
James Zern [Tue, 21 Dec 2021 19:53:51 +0000 (11:53 -0800)]
vp9_prob_diff_update_savings_search_model: quiet conv warnings
under Visual Studio:
Warning C4244 '=': conversion from 'int64_t' to 'vpx_prob', possible loss of
data
after:
ea042a676 vp9 encoder: fix integer overflows
'newp' has already been range checked earlier in the loop so the cast won't
have any unexpected results
Change-Id: Ic10877db2c0633d53fffdf8852d5095403c23a02
James Zern [Tue, 21 Dec 2021 03:56:01 +0000 (19:56 -0800)]
vpx_int_pro_row: normalize declaration w/aom
this is a followup to:
7fbcee49d quiet -Warray-parameter warnings
and conforms to aom in:
06e13e817 quiet -Warray-parameter warnings
the sad functions are more varied in libvpx and will require a separate
pass
Change-Id: I765fd6704df615e836ba0b184ff8266ce926c394
Fyodor Kyslov [Thu, 16 Dec 2021 22:29:13 +0000 (22:29 +0000)]
Merge "vp9 encoder: fix test failure on 32 bit arch" into main
Fyodor Kyslov [Thu, 16 Dec 2021 07:11:15 +0000 (23:11 -0800)]
vp9 encoder: fix test failure on 32 bit arch
test fails with memory error. Reducing testing resolution
bug: webm:1750
Change-Id: I75664088022aa660bdf6e69de2d11121db44716f
Marco Paniconi [Wed, 15 Dec 2021 19:27:09 +0000 (19:27 +0000)]
Merge "Set unused reference frames to first ref" into main
Fyodor Kyslov [Wed, 15 Dec 2021 02:51:31 +0000 (02:51 +0000)]
Merge "vp9 encoder: fix integer overflows" into main
Fyodor Kyslov [Tue, 14 Dec 2021 17:59:17 +0000 (09:59 -0800)]
vp9 encoder: fix integer overflows
fixing integer overflow with 16K content and enabling the test
Bug: webm:1750
Fixed: webm:1750
Change-Id: I76eebd915bcae55bc755613251a98e1716dea4c0
Jianhui Dai [Thu, 9 Dec 2021 05:38:22 +0000 (13:38 +0800)]
Set unused reference frames to first ref
If a reference frame is not referenced, then set the index for that
reference to the first one used/referenced instead of unused slot.
Unused slot means key frame, as key frame resets all slots with itself.
This CL extracts `get_first_ref_frame()` from `reset_fb_idx_unused()`
with a typo fixing, and sets all unused reference frames to first ref in
vp9 uncompressed header.
Bug: webrtc:13442
Change-Id: I99523bc2ceedf27efe376d1113851ff342982181
James Zern [Fri, 10 Dec 2021 19:42:06 +0000 (19:42 +0000)]
Merge "vp9_diamond_search_sad_avx: quiet -Wmaybe-uninitialized warning" into main
James Zern [Fri, 10 Dec 2021 18:32:44 +0000 (18:32 +0000)]
Merge "vp[89]_initalize_enc(): protect against multiple invocations" into main
James Zern [Fri, 10 Dec 2021 02:34:18 +0000 (18:34 -0800)]
vp[89]_initalize_enc(): protect against multiple invocations
this removes the burden from callers; the rtcd functions are left with a
mostly redundant (outside of tests) once() as top-level functions should
ensure their constraints are met
Change-Id: I5bdbcfa4671c6a1492cfe9c7d886c361c26caaa9
James Zern [Fri, 10 Dec 2021 02:02:30 +0000 (18:02 -0800)]
vp9_diamond_search_sad_avx: quiet -Wmaybe-uninitialized warning
w/gcc-11
v_these_mv_w is always initialized in this block with _mm_add_epi16();
converting this to a _mm_storeu_si32(tmp) call also works, but
introduces more stack usage
|| ../vp9/encoder/x86/vp9_diamond_search_sad_avx.c: In function
‘vp9_diamond_search_sad_avx’:
vp9/encoder/x86/vp9_diamond_search_sad_avx.c|285 col 19| warning:
‘v_these_mv_w’ may be used uninitialized [-Wmaybe-uninitialized]
|| 285 | new_bmv = ((const int_mv *)&v_these_mv_w)[local_best_idx];
|| | ~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
vp9/encoder/x86/vp9_diamond_search_sad_avx.c|149 col 21| note:
‘v_these_mv_w’ declared here
|| 149 | const __m128i v_these_mv_w = _mm_add_epi16(v_bmv_w, v_ss_mv_w);
|| | ^~~~~~~~~~~~
Change-Id: I1cd2fcb41030db16f51c94f3a70eb8eb2a526401
James Zern [Fri, 10 Dec 2021 01:32:30 +0000 (17:32 -0800)]
vp9_bitstream.c: quiet -Wstringop-overflow warning
w/gcc-11
as noted in
the size of interp_filter_selected[][]'s first dimension varies between
VP9_COMP and VP9BitstreamWorkerData as noted in the latter's definition:
// The size of interp_filter_selected in VP9_COMP is actually
// MAX_REFERENCE_FRAMES x SWITCHABLE. But when encoding tiles, all we ever do
// is increment the very first index (index 0) for the first dimension. Hence
// this is sufficient.
int interp_filter_selected[1][SWITCHABLE];
normalize the function signatures of write_modes*(), etc. to take this
into account.
vp9/encoder/vp9_bitstream.c|948 col 3| warning: ‘write_modes’ accessing
64 bytes in a region of size 16 [-Wstringop-overflow=]
|| 948 | write_modes(cpi, xd, &cpi->tile_data[data->tile_idx].tile_info,
|| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|| 949 | &data->bit_writer, tile_row, data->tile_idx,
|| | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
|| 950 | &data->max_mv_magnitude, data->interp_filter_selected);
|| | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
vp9/encoder/vp9_bitstream.c|948 col 3| note: referencing argument 8 of
type ‘int (*)[4]’
vp9/encoder/vp9_bitstream.c|488 col 13| note: in a call to function
‘write_modes’
Change-Id: I0898cd7c3431633c382a0c3a1be2f0a0bea8d0f9
James Zern [Thu, 9 Dec 2021 05:42:28 +0000 (21:42 -0800)]
quiet -Warray-parameter warnings
w/gcc-11
this matches the definition of the function with the declaration
Change-Id: I757b731b9560cb0b0ceec4ec258ec5af5a183b3d
James Zern [Thu, 9 Dec 2021 18:03:55 +0000 (18:03 +0000)]
Merge "test_intra_pred_speed: match above ext w/reconintra" into main
James Zern [Thu, 9 Dec 2021 05:35:26 +0000 (21:35 -0800)]
test_intra_pred_speed: match above ext w/reconintra
only 2 x block_size is needed
+ remove a related TODO; C & assembly rely on this extension
Change-Id: Iea430267624251cccbbdaec8045eb81d01ae1db1
James Zern [Thu, 9 Dec 2021 03:34:47 +0000 (19:34 -0800)]
vp9_thread_test.cc: remove incorrect TODO
the row-based loop filter is ok (and being used) in this case; since
it's serialized the previous row will always be done
Change-Id: I024a0c78e7488178956cc22a4c4680a00dc6eade
James Zern [Tue, 7 Dec 2021 21:11:46 +0000 (13:11 -0800)]
clear -Wextra-semi/-Wextra-semi-stmt warnings x2
some additional neon file updates after:
31b954deb clear -Wextra-semi/-Wextra-semi-stmt warnings
Bug: chromium:1257449
Change-Id: I3e2664f2bd8f6f7328ec91bf6595ba5fc09862bd
James Zern [Thu, 2 Dec 2021 21:11:56 +0000 (13:11 -0800)]
clear -Wextra-semi/-Wextra-semi-stmt warnings
Bug: chromium:1257449
Change-Id: Ia9aafccc09b611521d4a7aedfe3723393a840c62
James Zern [Fri, 19 Nov 2021 03:06:17 +0000 (03:06 +0000)]
Merge "vp9 encoder: fix row-mt crash w/thread config change" into main
James Zern [Wed, 17 Nov 2021 02:21:34 +0000 (18:21 -0800)]
vp9 encoder: fix row-mt crash w/thread config change
previously row-mt would allocate thread data once, so increasing the
number of threads with a config change would cause a heap overflow.
Bug: chromium:1261415
Bug: chromium:1270689
Change-Id: I3c5ec8444ae91964fa34a19dd780bd2cbb0368bf
James Zern [Thu, 18 Nov 2021 22:35:20 +0000 (22:35 +0000)]
Merge "vp9 encoder: fix some integer overflows" into main
Johann Koenig [Thu, 18 Nov 2021 07:13:09 +0000 (07:13 +0000)]
Merge "replaced bsr() with get_msb() from bitops.h" into main
Fyodor Kyslov [Wed, 17 Nov 2021 21:15:00 +0000 (13:15 -0800)]
vp9 encoder: fix some integer overflows
cap bitrate to 1000Mbps, change bitsaving budget to int64_t
this make test coverage for 2048x2048 - same as for vp8
Bug: webm:1749
Fixed: webm:1749
Change-Id: Ic58d73cb7529b0826d1f501ad09af8e80f706a6e
Johann Koenig [Thu, 18 Nov 2021 04:41:58 +0000 (04:41 +0000)]
Merge "faster vp8_regular_quantize_b_sse4_1" into main
Ilya Kurdyukov [Wed, 17 Nov 2021 07:16:02 +0000 (14:16 +0700)]
replaced bsr() with get_msb() from bitops.h
The modified line should now compile into two instructions instead of four.
Change-Id: Ie2eb6b13ff1e29b3107cb9e76f37ff9065504316
Ilya Kurdyukov [Sat, 13 Nov 2021 11:22:14 +0000 (18:22 +0700)]
faster vp8_regular_quantize_b_sse4_1
Gives 10% faster VP8 encoding in simple tests.
This patch requires testing on wider datasets and encoder
settings to see if this speedup is achieved on most data.
Change-Id: If8e04819623e78fff126c413db66c964c0b4c11a
James Zern [Tue, 16 Nov 2021 03:20:36 +0000 (19:20 -0800)]
encode_api_test.cc: unify kCodecs[] definitions
and rename the table to kCodecIfaces[] to be a little more specific and
avoid shadowing kCodecs[] in SetRoi()
Change-Id: I64905f48d8bf76e812bdba8374b82e3f7654686f
Johann Koenig [Wed, 17 Nov 2021 03:35:50 +0000 (03:35 +0000)]
Merge "MacOS 12 is darwin21" into main
Yunqing Wang [Tue, 16 Nov 2021 21:47:09 +0000 (21:47 +0000)]
Merge "test/DummyVideoSource::ReallocImage: check img_ alloc" into main
Johann [Tue, 16 Nov 2021 21:02:15 +0000 (06:02 +0900)]
MacOS 12 is darwin21
Remove -mmacosx-version-min. The library does not use
any calls which are affected by the platform version.
There is also no version 10.16 as it went from 10.15
to 11 and now to 12.
At some point it may be good to clarify that the bare
-darwin- target is for iOS and the -darwinN- targets
are for macOS.
Change-Id: I2fd5f7cae2637905acf3ab77bfddfbe367abbb68