James Zern [Tue, 15 Nov 2022 19:19:43 +0000 (19:19 +0000)]
Merge "[NEON] Optimize FHT functions, add highbd FHT 4x4" into main
Johann [Mon, 14 Nov 2022 08:59:45 +0000 (17:59 +0900)]
quantize: remove vp9_regular_quantize_b_4x4
This was just a helper function which called vpx_quantize_b or
vpx_highbd_quantize_b. It also checked for skip_block, which was
necessary when webm:1439 was filed but does not appear to be
necessary now.
Removes a quantize variant and makes subsequent cleanups easier.
Change-Id: Ibe545eccd19370f07ff26c8e151f290c642efd2a
Konstantinos Margaritis [Wed, 9 Nov 2022 09:30:58 +0000 (09:30 +0000)]
[NEON] Optimize FHT functions, add highbd FHT 4x4
Refactor & optimize FHT functions further, use new butterfly functions
4x4 5% faster, 8x8 & 16x16 10% faster than previous versions.
Highbd 4x4 FHT version 2.27x faster than C version for --rt.
Change-Id: I3ebcd26010f6c5c067026aa9353cde46669c5d94
Marco Paniconi [Fri, 11 Nov 2022 02:50:19 +0000 (18:50 -0800)]
vp9-rc: Fix key frame setting in external RC
Bug: b/
257368998
Change-Id: I03e35915ac99b50cb6bdf7bce8b8f9ec5aef75b7
James Zern [Mon, 7 Nov 2022 21:48:50 +0000 (21:48 +0000)]
Merge "Add Neon implementation of vpx_hadamard_32x32" into main
Sam James [Sun, 6 Nov 2022 04:11:59 +0000 (04:11 +0000)]
build: fix -Wimplicit-int (Clang 16)
Clang 16 will make -Wimplicit-int error by default which can, in addition to
other things, lead to some configure tests silently failing/returning the wrong result.
Fixes this error:
```
+/var/tmp/portage/media-libs/libvpx-1.12.0/temp/vpx-conf-1802-30624.c:1:15: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
```
For more information, see LWN.net [0] or LLVM's Discourse [1], gentoo-dev@ [2],
or the (new) c-std-porting mailing list [3].
[0] https://lwn.net/Articles/913505/
[1] https://discourse.llvm.org/t/configure-script-breakage-with-the-new-werror-implicit-function-declaration/65213
[2] https://archives.gentoo.org/gentoo-dev/message/
dd9f2d3082b8b6f8dfbccb0639e6e240
[3] hosted at lists.linux.dev.
Bug: https://bugs.gentoo.org/879705
Change-Id: Id73a98944ab3c99a368b9da7a5e902ddff9d937f
Signed-off-by: Sam James <sam@gentoo.org>
Andrew Salkeld [Thu, 13 Oct 2022 15:28:41 +0000 (16:28 +0100)]
Add Neon implementation of vpx_hadamard_32x32
Add an Arm Neon implementation of vpx_hadamard_32x32 and use it
instead of the scalar C implementation.
Also add test coverage for the new Neon implementation.
Change-Id: Iccc018eec4dbbe629fb0c6f8ad6ea8554e7a0b13
Konstantinos Margaritis [Wed, 26 Oct 2022 22:09:32 +0000 (22:09 +0000)]
[NEON] Optimize highbd 32x32 DCT
For --best quality, resulting function
vpx_highbd_fdct32x32_rd_neon takes 0.27% of cpu time in
profiling, vs 6.27% for the sum of scalar functions:
vpx_fdct32, vpx_fdct32.constprop.0, vpx_fdct32x32_rd_c for rd.
For --rt quality, the function takes 0.19% vs 4.57% for the scalar
version.
Overall, this improves encoding time by ~6% compared for highbd
for --best and ~9% for --rt.
Change-Id: I1ce4bbef6e364bbadc76264056aa3f86b1a8edc5
James Zern [Wed, 2 Nov 2022 02:21:18 +0000 (02:21 +0000)]
Merge "[NEON] Optimize and homogenize Butterfly DCT functions" into main
Konstantinos Margaritis [Wed, 26 Oct 2022 21:37:31 +0000 (21:37 +0000)]
[NEON] Optimize and homogenize Butterfly DCT functions
Provide a set of commonly used Butterfly DCT functions for use in
DCT 4x4, 8x8, 16x16, 32x32 functions. These are provided in various
forms, using vqrdmulh_s16/vqrdmulh_s32 for _fast variants, which
unfortunately are only usable in pass1 of most DCTs, as they do not
provide the necessary precision in pass2.
This gave a performance gain ranging from 5% to 15% in 16x16 case.
Also, for 32x32, the loads were rearranged, along with the butterfly
optimizations, this gave 10% gain in 32x32_rd function.
This refactoring was necessary to allow easier porting of highbd
32x32 functions -follows this patchset.
Change-Id: I6282e640b95a95938faff76c3b2bace3dc298bc3
Johann Koenig [Thu, 27 Oct 2022 08:38:48 +0000 (08:38 +0000)]
Merge "MacOS 13 is darwin22" into main
Johann Koenig [Thu, 27 Oct 2022 08:38:18 +0000 (08:38 +0000)]
Merge "rtcd: allow disabling neon on armv8" into main
Johann [Thu, 27 Oct 2022 02:40:19 +0000 (11:40 +0900)]
MacOS 13 is darwin22
Bug: webm:1783
Change-Id: I97d94ab8c8aebe13aedb58e280dc37474814ad5d
Johann [Wed, 26 Oct 2022 23:49:37 +0000 (08:49 +0900)]
rtcd: allow disabling neon on armv8
Change-Id: Idef943775456eb95b46be5c92c114c1d215f38d7
Johann [Wed, 26 Oct 2022 08:14:21 +0000 (17:14 +0900)]
mailmap: add johann@duck.com
Change-Id: I3b48951e69ba1f4a9fafdbb81fac48f79587a342
James Zern [Tue, 25 Oct 2022 19:16:46 +0000 (19:16 +0000)]
Merge changes I36545ff4,Id1aa29da into main
* changes:
vp9_highbd_quantize_fp*_neon: normalize fn param name
highbd_sad_avx2: normalize function param names
James Zern [Tue, 25 Oct 2022 19:16:08 +0000 (19:16 +0000)]
Merge "SAD*Test: mark virtual Run() as overridden" into main
Johann Koenig [Tue, 25 Oct 2022 13:26:37 +0000 (13:26 +0000)]
Merge "quantize: consolidate sse2 conditionals" into main
Johann Koenig [Tue, 25 Oct 2022 13:26:22 +0000 (13:26 +0000)]
Merge "vp9 quantize: rewrite ssse3 in intrinsics" into main
James Zern [Mon, 24 Oct 2022 22:37:26 +0000 (15:37 -0700)]
SAD*Test: mark virtual Run() as overridden
this comes from AbstractBench
Change-Id: Ie0b5a26a68bfbffd80f132125d15a1bdfc990c22
James Zern [Mon, 24 Oct 2022 22:28:47 +0000 (15:28 -0700)]
vp9_highbd_quantize_fp*_neon: normalize fn param name
count -> n_coeffs. aligns the name with the rtcd header; clears a
clang-tidy warning
Change-Id: I36545ff479df92b117c95e494f16002e6990f433
James Zern [Mon, 24 Oct 2022 22:24:51 +0000 (15:24 -0700)]
highbd_sad_avx2: normalize function param names
(src|ref)8_ptr -> (src|ref)_ptr. aligns the names with the rtcd header;
clears some clang-tidy warnings
Change-Id: Id1aa29da8c0fa5860b46ac902f5b2620c0d3ff54
Marco Paniconi [Tue, 18 Oct 2022 05:36:25 +0000 (22:36 -0700)]
Fix to VP8 external RC for buffer levels
On a dynamic change of temporal layers:
starting/maimum/optimal were being set twice,
causing incorrect large values.
Bug: b/
253927937
Change-Id: I204e885cff92530336a9ed9a4363c486c5bf80ae
Johann [Mon, 17 Oct 2022 07:22:23 +0000 (16:22 +0900)]
quantize: consolidate sse2 conditionals
Change-Id: I43de579e30f2967b97064063e29676e0af1a498f
Johann [Sat, 1 Oct 2022 02:47:05 +0000 (11:47 +0900)]
vp9 quantize: rewrite ssse3 in intrinsics
Change-Id: I3177251a5935453a23a23c39ea5f6fd41254775e
Marco Paniconi [Sat, 15 Oct 2022 01:56:46 +0000 (01:56 +0000)]
Merge "Fix to VP8 external RC for dynamic update of layers" into main
Marco Paniconi [Wed, 12 Oct 2022 07:10:47 +0000 (00:10 -0700)]
Fix to VP8 external RC for dynamic update of layers
On change/update of rc_cfg: when number of temporal
layers change call vp8_reset_temporal_layer_change(),
which in turn will call vp8_init_temporal_layer_context()
only for the new layers.
Bug:b/
249644737
Change-Id: Ib20d746c7eacd10b78806ca6a5362c750d9ca0b3
Konstantinos Margaritis [Thu, 13 Oct 2022 15:19:46 +0000 (15:19 +0000)]
[NEON] fix clang compile warnings
Change-Id: Ib7ce7a774ec89ba51169ea64d24c878109ef07d1
Scott LaVarnway [Thu, 13 Oct 2022 11:31:51 +0000 (11:31 +0000)]
Merge "Add vpx_highbd_sad64x{64,32}_avg_avx2." into main
Konstantinos Margaritis [Fri, 7 Oct 2022 15:13:29 +0000 (15:13 +0000)]
[NEON] Add highbd FDCT 16x16 function
90-95% faster than C version in best/rt profiles
Change-Id: I41d5e9acdc348b57153637ec736498a25ed84c25
James Zern [Wed, 12 Oct 2022 20:07:51 +0000 (20:07 +0000)]
Merge "[NEON] Add highbd FDCT 8x8 function" into main
Scott LaVarnway [Wed, 12 Oct 2022 19:50:55 +0000 (19:50 +0000)]
Merge "Add vpx_highbd_sad32x{64,32,16}_avg_avx2." into main
Scott LaVarnway [Wed, 12 Oct 2022 19:44:44 +0000 (19:44 +0000)]
Merge "Add vpx_highbd_sad16x{32,16,8}_avg_avx2." into main
Konstantinos Margaritis [Thu, 6 Oct 2022 16:00:43 +0000 (16:00 +0000)]
[NEON] Add highbd FDCT 8x8 function
50% faster than C version in best/rt profiles
Change-Id: I0f9504ed52b5d5f7722407e91108ed4056d66bc2
Scott LaVarnway [Wed, 12 Oct 2022 17:26:43 +0000 (10:26 -0700)]
Add vpx_highbd_sad64x{64,32}_avg_avx2.
~2.8x faster than the sse2 version.
Bug: b/
245917257
Change-Id: Ib727ba8a8c8fa4df450bafdde30ed99fd283f06d
Konstantinos Margaritis [Thu, 6 Oct 2022 14:53:56 +0000 (14:53 +0000)]
[NEON] Add highbd FDCT 4x4 function
~80% faster than C version for both best/rt profiles.
Change-Id: Ibb3c8e1862131d2a020922420d53c66b31d5c2c3
Scott LaVarnway [Wed, 12 Oct 2022 13:05:46 +0000 (06:05 -0700)]
Add vpx_highbd_sad32x{64,32,16}_avg_avx2.
2.1x to 2.8x faster than the sse2 version.
Bug: b/
245917257
Change-Id: I1aaffa4a1debbe5559784e854b8fc6fba07e5000
Scott LaVarnway [Mon, 10 Oct 2022 15:38:44 +0000 (08:38 -0700)]
Add vpx_highbd_sad16x{32,16,8}_avg_avx2.
1.6x to 2.1x faster than the sse2 version.
Bug: b/
245917257
Change-Id: I56c467a850297ae3abcca4b4843302bb8d5d0ac1
Konstantinos Margaritis [Thu, 6 Oct 2022 13:05:01 +0000 (13:05 +0000)]
[NEON] Move helper functions for reuse
Move all butterfly functions to fdct_neon.h
Slightly optimize load/scale/cross functions
in fdct 16x16.
These will be reused in highbd variants.
Change-Id: I28b6e0cc240304bab6b94d9c3f33cca77b8cb073
Scott LaVarnway [Mon, 10 Oct 2022 20:34:02 +0000 (20:34 +0000)]
Merge "SADavgTest: Add speed test." into main
Scott LaVarnway [Mon, 10 Oct 2022 19:20:37 +0000 (12:20 -0700)]
SADavgTest: Add speed test.
Change-Id: Ie14c0f6d15f410adf749f7ab74cf9f2bf35f3d5f
Konstantinos Margaritis [Thu, 6 Oct 2022 10:58:27 +0000 (10:58 +0000)]
[NEON] move transpose_8x8 to reuse
Change-Id: I3915b6c9971aedaac9c23f21fdb88bc271216208
James Zern [Mon, 10 Oct 2022 18:37:05 +0000 (18:37 +0000)]
Merge "[NEON] highbd partial DCT functions" into main
Konstantinos Margaritis [Thu, 6 Oct 2022 10:26:05 +0000 (10:26 +0000)]
[NEON] highbd partial DCT functions
Change-Id: I7dd4e698469562f5b1f948cc36f8403b490dcb6a
Scott LaVarnway [Fri, 7 Oct 2022 12:53:50 +0000 (05:53 -0700)]
Add vpx_highbd_sad64x{64,32}_avx2.
~2.8x faster than the sse2 version.
Bug: b/
245917257
Change-Id: Ibc8e5d030ec145c9a9b742fff98fbd9131c9ede4
Johann Koenig [Fri, 7 Oct 2022 08:17:03 +0000 (08:17 +0000)]
Merge "vp9 quantize: change index" into main
Scott LaVarnway [Wed, 5 Oct 2022 21:03:55 +0000 (14:03 -0700)]
Add vpx_highbd_sad32x{64,32,16}_avx2.
2.7x to 3.1x faster than the sse2 version.
Bug: b/
245917257
Change-Id: Idff3284932f7ee89d036f38893205bf622a159a3
Scott LaVarnway [Wed, 5 Oct 2022 14:04:27 +0000 (07:04 -0700)]
Add vpx_highbd_sad16x{32,16,8}_avx2.
1.9x to 2.4x faster than the sse2 version.
Bug: b/
245917257
Change-Id: I686452772f9b72233930de2207af36a0cd72e0bb
Cheng Chen [Tue, 4 Oct 2022 16:15:49 +0000 (16:15 +0000)]
Merge "L2E: Rework recode decisions for external max frame size and q" into main
Johann [Sat, 1 Oct 2022 02:18:09 +0000 (11:18 +0900)]
vp9 quantize: change index
In assembly it made sense to iterate using n_coeffs.
In intrinsics it's just as fast to use index and
easier to read.
Change-Id: I403c959709309dad68123d0a3d0efe183874543d
Scott LaVarnway [Mon, 19 Sep 2022 12:09:23 +0000 (05:09 -0700)]
vpx_subpixel_8t_intrin_avx2.c: quiet -Wuninitialized
warning: ‘s2[3]’ may be used uninitialized
and
warning: ‘s1[3]’ may be used uninitialized
The warnings exposed unused code.
Change-Id: I75cf1f9db75e811cb42e2f143be1ad76f3e4dee9
Scott LaVarnway [Mon, 26 Sep 2022 23:18:04 +0000 (23:18 +0000)]
Merge "vp9_rd.c quiet -Wstringop-overflow" into main
Johann [Sat, 24 Sep 2022 01:53:05 +0000 (10:53 +0900)]
quantize: standardize vp9_quantize_fp_sse2
Match style for vpx_quantize_b_sse2 and prepare to rewrite
ssse3 version in intrinsics.
Need to evaluate the value of threshold breakout before
going further.
Change-Id: I9cfceb1bb0dc237cd6b73fc8d41d78bba444a15b
Scott LaVarnway [Fri, 23 Sep 2022 16:17:18 +0000 (09:17 -0700)]
vp9_rd.c quiet -Wstringop-overflow
../libvpx/vp9/encoder/vp9_rd.c:594:20: warning: writing 1 byte into a region of size 0 [-Wstringop-overflow=]
594 | t_above[i] = !!*(const uint32_t *)&above[i];
| ~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../libvpx/vp9/encoder/vp9_rd.c:572:47: note: at offset [64, 254] into destination object ‘t_above’ of size [0, 16]
572 | ENTROPY_CONTEXT t_above[16],
| ~~~~~~~~~~~~~~~~^~~~~~~~~~~
Change-Id: Ie9ef24e685af417cdd35f6aa7284805e422b6ae2
Johann [Sat, 24 Sep 2022 01:55:52 +0000 (10:55 +0900)]
quantize: add untested function
vp9_quantize_fp_sse2 was only tested in non-hbd
configuration. Missed when fixing this for
vpx_quantize_b_sse2.
Change-Id: Ide346e5727d74281c774f605c90d280050e0bf62
Johann [Fri, 16 Sep 2022 23:47:28 +0000 (08:47 +0900)]
quantize: increase iscan by 1
All of the assembly adds 1 to iscan to convert from
a 0 based array to the EOB value.
Add 1 to all iscan values and remove the extra
instructions from the assembly.
Change-Id: I219dd7f2bd10533ab24b206289565703176dc5e9
Scott LaVarnway [Wed, 21 Sep 2022 23:41:42 +0000 (23:41 +0000)]
Merge "resize_test.cc: quiet -Wmaybe-uninitialized" into main
Scott LaVarnway [Wed, 21 Sep 2022 19:15:16 +0000 (12:15 -0700)]
resize_test.cc: quiet -Wmaybe-uninitialized
warning: ‘expected_w’ may be used uninitialized
Change-Id: I915efd82d3263250cea90391345f7683c1330fc8
Scott LaVarnway [Wed, 21 Sep 2022 20:53:07 +0000 (20:53 +0000)]
Merge "post_proc_sse2.c: quiet -Wuninitialized" into main
Scott LaVarnway [Wed, 21 Sep 2022 18:37:04 +0000 (11:37 -0700)]
post_proc_sse2.c: quiet -Wuninitialized
In file included from ../libvpx/vpx_dsp/x86/post_proc_sse2.c:12:
In function ‘_mm_add_epi16’,
inlined from ‘vpx_mbpost_proc_down_sse2’ at ../libvpx/vpx_dsp/x86/post_proc_sse2.c:88:13:
/usr/lib/gcc/x86_64-linux-gnu/12/include/emmintrin.h:1060:35: warning: ‘below_context’ may be used uninitialized [-Wmaybe-uninitialized]
1060 | return (__m128i) ((__v8hu)__A + (__v8hu)__B);
| ^~~~~~~~~~~
../libvpx/vpx_dsp/x86/post_proc_sse2.c: In function ‘vpx_mbpost_proc_down_sse2’:
../libvpx/vpx_dsp/x86/post_proc_sse2.c:39:13: note: ‘below_context’ was declared here
39 | __m128i below_context;
Change-Id: I2fc592f121c4e85d0aff1640014c3444f5eb09fd
James Zern [Tue, 20 Sep 2022 23:24:44 +0000 (23:24 +0000)]
Merge "CHECK_MEM_ERROR: add an assert for a valid jmp target" into main
Johann Koenig [Tue, 20 Sep 2022 00:12:13 +0000 (00:12 +0000)]
Merge "quantize: test lowbd in highbd builds" into main
Johann [Sun, 18 Sep 2022 01:26:00 +0000 (10:26 +0900)]
quantize: test lowbd in highbd builds
Change-Id: I7af273e979415a8b8cafb7494728d2736862f4a5
Johann [Fri, 16 Sep 2022 22:54:40 +0000 (07:54 +0900)]
fwd_txfm: remove avx2 file from non-hbd
Resolves warning on OS X:
file: libvpx_g.a(fwd_txfm_avx2.c.o) has no symbols
Change-Id: Ie8b290bb3ed329656beb883d552c98353f1ed5e5
Cheng Chen [Wed, 14 Sep 2022 18:40:50 +0000 (11:40 -0700)]
L2E: Rework recode decisions for external max frame size and q
Allow to handle external q and external max frame size separately.
Rely on libvpx's decision to catch overshoot/undershoot and recode frames.
Previously, when external max frame size is set, we didn't handle
undershoot cases, and now we fall back to libvpx's decision to
recode a frame if overshoot/undershoot is seen.
Change-Id: Ic3eee042cfe104b528c5f2c6c82b98dd5d8fa8ca
Scott LaVarnway [Wed, 14 Sep 2022 10:36:46 +0000 (03:36 -0700)]
Add vpx_highbd_sad64x{64,32}x4d_avx2.
~2x faster than the sse2 version.
Bug: b/
245917257
Change-Id: I4742950ab7b90d7f09e8d4687e1e967138acee39
Scott LaVarnway [Mon, 12 Sep 2022 14:40:39 +0000 (07:40 -0700)]
Add vpx_highbd_sad32x{64,32,16}x4d_avx2.
~2.4x faster than the sse2 version.
Bug: b/
245917257
Change-Id: I6df2bd62b46e5e175c8ad80daa6de3a1c313db0f
James Zern [Sat, 28 May 2022 04:53:49 +0000 (21:53 -0700)]
CHECK_MEM_ERROR: add an assert for a valid jmp target
callers of CHECK_MEM_ERROR() expect failures to not return
tested with:
configure --enable-debug --enable-vp9-postproc --enable-postproc \
--enable-multi-res-encoding --enable-vp9-temporal-denoising \
--enable-error-concealment
--enable-internal-stats has unrelated assertion failures currently
Change-Id: Ic12073b1ae80a6f434f14d24f652e64d30f63eea
Scott LaVarnway [Mon, 12 Sep 2022 12:18:19 +0000 (12:18 +0000)]
Merge "Add vpx_highbd_sad16x{32,16,8}x4d_avx2." into main
Wan-Teh Chang [Thu, 8 Sep 2022 22:35:13 +0000 (15:35 -0700)]
Update third_party/googletest to v1.12.1
See https://github.com/google/googletest/releases/tag/release-1.12.1.
Modeled after https://aomedia-review.googlesource.com/c/aom/+/162601.
Change-Id: If0ced3097b4c8490985e3381aaac9b3266d52ae7
Scott LaVarnway [Thu, 8 Sep 2022 20:05:55 +0000 (13:05 -0700)]
Add vpx_highbd_sad16x{32,16,8}x4d_avx2.
1.98x to 2.3x faster than the sse2 version.
Bug: b/
245917257
Change-Id: Ie4f9bb942ffaf4af7d395fb5a5978b41aabfc93c
James Zern [Thu, 8 Sep 2022 01:41:13 +0000 (18:41 -0700)]
vp8_decode: declare 2 variables volatile
fixes -Wclobbered warnings with gcc 12.1.0:
vp8/vp8_dx_iface.c|278 col 16| warning: variable 'w' might be clobbered
by 'longjmp' or 'vfork' [-Wclobbered]
vp8/vp8_dx_iface.c|278 col 19| warning: variable 'h' might be clobbered
by 'longjmp' or 'vfork' [-Wclobbered]
Change-Id: Ib2c606a3450188d7869c066cacaf5615d9746181
James Zern [Tue, 6 Sep 2022 22:23:30 +0000 (22:23 +0000)]
Merge "x86,cosmetics: prefer _mm_setzero_si128/_mm256_setzero_si256" into main
James Zern [Fri, 2 Sep 2022 23:55:43 +0000 (16:55 -0700)]
sad_neon: enable UDOT implementation w/aarch32
Change-Id: Ia28305ec5c61518b732cbacbd102acd2cb7f9d82
James Zern [Fri, 2 Sep 2022 23:44:14 +0000 (16:44 -0700)]
variance_neon.cc: simplify __ARM_FEATURE_DOTPROD check
missed in
447e27588 vpx_dsp,neon: simplify __ARM_FEATURE_DOTPROD check
+ fix #if comments
only check that the macro is defined, the value doesn't have any effect.
from https://arm-software.github.io/acle/main/acle.html:
5.5.7.7. Dot Product extension
__ARM_FEATURE_DOTPROD is defined if the dot product data manipulation
instructions are supported and the vector intrinsics are available.
Note that this implies:
- __ARM_NEON == 1
Change-Id: I098b96421b7de5928bb3b11612ca1f32e7b6cbc4
James Zern [Fri, 2 Sep 2022 23:17:52 +0000 (16:17 -0700)]
x86,cosmetics: prefer _mm_setzero_si128/_mm256_setzero_si256
over *_set1_*(0)
Change-Id: I136e1798a2ce286480ebb9418db67a2f1e92b9a2
James Zern [Fri, 2 Sep 2022 19:17:20 +0000 (12:17 -0700)]
vpx_dsp,neon: simplify __ARM_FEATURE_DOTPROD check
only check that the macro is defined, the value doesn't have any effect.
from https://arm-software.github.io/acle/main/acle.html:
5.5.7.7. Dot Product extension
__ARM_FEATURE_DOTPROD is defined if the dot product data manipulation
instructions are supported and the vector intrinsics are available.
Note that this implies:
- __ARM_NEON == 1
Change-Id: I164fe121ccefda99050a9b6a99738a2b518520f3
James Zern [Fri, 2 Sep 2022 01:47:50 +0000 (18:47 -0700)]
neon,load_unaligned_*: use dup for lane 0
this produces better assembly with gcc (11.3.0-3); no change in assembly
using clang from the r24 android sdk (Android (8075178, based on
r437112b) clang version 14.0.1
(https://android.googlesource.com/toolchain/llvm-project
8671348b81b95fc603505dfc881b45103bee1731)
Change-Id: Ifec252d4f499f23be1cd94aa8516caf6b3fbbc11
James Zern [Wed, 31 Aug 2022 23:35:08 +0000 (16:35 -0700)]
test/*,cosmetics: normalize void parameter lists
replace (void) with (); use of this synonym is more common in C++ code.
Change-Id: I9813e82234dc9caa7115918a0491b0040f6afaf4
Yaowu Xu [Tue, 30 Aug 2022 16:04:58 +0000 (09:04 -0700)]
Remove const for pass-by-value parameters
This also fixes MSVC compiler warnings.
Change-Id: I20dc9ac821275ba95598f3016fc6b23e884e13b7
Cheng Chen [Tue, 30 Aug 2022 04:25:33 +0000 (04:25 +0000)]
Merge "L2E: Add gop size and ARF existence to frame info" into main
James Zern [Sat, 27 Aug 2022 05:12:44 +0000 (22:12 -0700)]
highbd_variance_neon,cosmetics: reorder a few lines
Change-Id: Ia6fa54652d7f94687e64108482bb0f28ca06cf49
Cheng Chen [Fri, 26 Aug 2022 21:29:32 +0000 (14:29 -0700)]
L2E: Add gop size and ARF existence to frame info
Pass the encode frame info to external ml model, with the information
of gop size and whether alt ref is used.
Change-Id: I55be2d3de83d7182c1a1a174e44ead7e19045c9d
James Zern [Fri, 26 Aug 2022 02:07:34 +0000 (02:07 +0000)]
Merge "[NEON] Add highbd *variance* functions" into main
James Zern [Fri, 26 Aug 2022 02:01:55 +0000 (02:01 +0000)]
Merge "vpx_encoder.h: note VPX_ERROR_RESILIENT_PARTITIONS is VP8-only" into main
Konstantinos Margaritis [Wed, 24 Aug 2022 12:28:43 +0000 (12:28 +0000)]
[NEON] Add highbd *variance* functions
Total gain for 12-bit encoding:
* ~7.2% for best profile
* ~5.8% for rt profile
Change-Id: I5b70415fb89d1bbb02a0c139eb317ba6b08adede
James Zern [Thu, 25 Aug 2022 20:44:41 +0000 (20:44 +0000)]
Merge "vp9: fix ubsan sub-overflows" into main
James Zern [Thu, 25 Aug 2022 17:50:16 +0000 (10:50 -0700)]
vpx_encoder.h: note VPX_ERROR_RESILIENT_PARTITIONS is VP8-only
Change-Id: If71b2ec766f9f41253ce5a34987ffd208f9c8381
James Zern [Thu, 25 Aug 2022 16:55:42 +0000 (16:55 +0000)]
Merge "vp8_ratectrl_rtc_test.cc: ensure frame_type is initialized" into main
James Zern [Thu, 25 Aug 2022 01:50:10 +0000 (18:50 -0700)]
libs.doxy_template: remove obsolete CLASS_DIAGRAMS
This was reported with doxygen 1.9.4.
Also update the comment for CLASS_GRAPH by running "doxygen -u" because
the original comment for CLASS_GRAPH mentions the obsolete tag
'CLASS_DIAGRAMS',
Change-Id: I3bca547201f794d363bd814b7c7f7c9d7088797a
James Zern [Wed, 24 Aug 2022 22:48:24 +0000 (15:48 -0700)]
vp8_ratectrl_rtc_test.cc: ensure frame_type is initialized
this fixes a valgrind failure:
==1095597== Conditional jump or move depends on uninitialised value(s)
==1095597== at 0x12E0CC: (anonymous
namespace)::Vp8RcInterfaceTest::PreEncodeFrameHook(libvpx_test::VideoSource*,
libvpx_test:: > Encoder*) (vp8_ratectrl_rtc_test.cc:131)
==1095597== by 0x1255A9:
libvpx_test::EncoderTest::RunLoop(libvpx_test::VideoSource*)
(encode_test_driver.cc:205)
Bug: webm:1776
Change-Id: Id3b40f62573ee513e79c74b6315c71b6ecd22c9a
Fixed: webm:1776
James Zern [Wed, 24 Aug 2022 19:18:25 +0000 (19:18 +0000)]
Merge "[NEON] Improve vpx_quantize_b* functions" into main
clang-format [Sat, 13 Aug 2022 17:33:56 +0000 (10:33 -0700)]
.clang-format: update to clang-format-11
only store the deltas from --style Google in the file and reapply using
Debian clang-format version 11.1.0-6+build1
Bug: b/
229626362
Change-Id: I3e18a2e7c17a90a48405b3cf1b37ebc652aba0db
Konstantinos Margaritis [Sat, 20 Aug 2022 19:02:15 +0000 (19:02 +0000)]
[NEON] Improve vpx_quantize_b* functions
Slight optimization, prefetch gives a 1% improvement in 1st pass
Change-Id: Iba4664964664234666406ab53893e02d481fbe61
James Zern [Tue, 23 Aug 2022 02:33:26 +0000 (19:33 -0700)]
vp9_ratectrl_rtc_test: initialize loopfilter_ctrl[]
this was added in:
7beafefd1 vp9: Allow for disabling loopfilter per spatial layer
but the test doesn't zero initialize its svc_params_ member.
fixes the use of an uninitialized value, reported by valgrind and
integer sanitizer:
[ RUN ] VP9/RcInterfaceSvcTest.Svc/0
==1064682== Conditional jump or move depends on uninitialised value(s)
==1064682== at 0x1C5624: loopfilter_frame (vp9_encoder.c:3285)
==1064682== by 0x1C9B54: encode_frame_to_data_rate (vp9_encoder.c:5595)
==1064682== by 0x1CA2EE: SvcEncode (vp9_encoder.c:5789)
==1064682== by 0x1CEA01: vp9_get_compressed_data (vp9_encoder.c:7891)
==1064682== by 0x185F0E: encoder_encode (vp9_cx_iface.c:1437)
==1064682== by 0x1503BB: vpx_codec_encode (vpx_encoder.c:208)
vp9/encoder/vp9_svc_layercontext.c:362:26: runtime error: implicit
conversion from type 'int' of value -1 (32-bit, signed) to type
'LOOPFILTER_CONTROL' changed the value to
4294967295 (32-bit, unsigned)
#0 0x558925f45377 in vp9_restore_layer_context vp9/encoder/vp9_svc_layercontext.c:362:26
#1 0x558925ef89fd in vp9_get_compressed_data vp9/encoder/vp9_encoder.c:7781:5
#2 0x558925e3ef3e in encoder_encode vp9/vp9_cx_iface.c:1437:20
Bug: b/
229626362
Change-Id: I33d244be7752c68b71efa9c62ca45d6b202ec761
James Zern [Mon, 22 Aug 2022 22:36:09 +0000 (22:36 +0000)]
Merge "vp9.read_inter_block_mode_info: return on corruption" into main
James Zern [Mon, 22 Aug 2022 22:21:46 +0000 (22:21 +0000)]
Merge "highbd_quantize_neon.c: remove unneeded assert.h" into main
James Zern [Mon, 22 Aug 2022 22:21:28 +0000 (22:21 +0000)]
Merge "vp9,search_new_mv: descale rather than scale sse" into main
James Zern [Mon, 22 Aug 2022 22:21:00 +0000 (22:21 +0000)]
Merge changes Iabed118b,I60a384b2 into main
* changes:
use VPX_NO_UNSIGNED_SHIFT_CHECK with entropy functions
compiler_attributes.h: add VPX_NO_UNSIGNED_SHIFT_CHECK
Konstantinos Margaritis [Mon, 22 Aug 2022 19:46:50 +0000 (19:46 +0000)]
[NEON] Add vpx_highbd_subtract_block function
Total gain for 12-bit encoding:
* ~1% for best and rt profile
Change-Id: I4039120dc570baab1ae519a5e38b1acff38d81f0