GitLab

This patch simplifies the setting and use of the do_post_rounding
flag consistently across all convolve/warp operations. It is always
set now to 1 for compound modes, but the masked blending operation
can subsequently turn it off.
To accomplish this, for the warped case, the use of the conv_params->dst
buffer is made consistent with the other non-warp convolves,
in the sense that it is used only when is_compound is 1.

Change-Id: If3a37ffa65c3ca75f1df66ca427e6b5df86ac72f

BUG=aomedia:1363

Change-Id: I8ca90f1b0d7dfe92078c611f66dcb19e70a6c1e0

For FRAME_REFS_SIGNALING, when error resilience mode is on for inter
frames, no frame refs signaling options will be provided. Hence no
additional bit is added and the reference frame signaling stays the
same as baseline.

BUG=aomedia:1392

Change-Id: I8ff9688c278373acaae3ee4fb74e9a392132f54d

Replace HORSHEAR_REDUCE_PRECISION_BITS with ROUND0_BITS for
consistency.
Also includes other clean-ups.

Change-Id: If0794cb16fdc8844859a614ed524c761fb46eddf

cm->allow_high_precision_mv and cm->cur_frame_force_integer_mv are
frame-level decisions. In reference MV search, no need to adjust
MV precision for spatial neighbours. This patch doesn't change
bitstreams.

Change-Id: I8a91a4c68fa24b4cddf665ee770f397c3610ba4b

BUG=aomedia:1363

Change-Id: If94c6404836fefcf3352f1a4f380066965ca68ce

Turn on the error resilient mode in the frame dropping unit test.
The temporal mv reference will be disabled in error resilient mode
by design.

BUG=aomedia:1411

Change-Id: Id777061bbdc376f8d1634c1286fce8094fdcc04e

When the block is coded in the bi-directional compound mode,
store up to one mv that points towards last_frame side, instead
of storing up to two mvs each corresponds to one side.

BUG=aomedia:1412

Change-Id: Id9ec295de28c0831457c8fe9ce5fd2085e3b5459

change the simd function pointer name from 16 tap to 14 tap to reflect
teh actual number of filter taps that are used in the deblocking filter.

Change-Id: I6f3683878465028f12c17f7daf8330ea8fc9e640

This experiment has been abandonned for AV1.

Change-Id: I924d172f7ce85b2422bace44684329fcbf71cb5e

One bit is added to the frame header, to signal whether the references
for inter frames will be short-signaled.

When the bit is set, only 6 bits are used at the frame header level to
signal the choices of LAST_FRAME and GOLDEN_FRAME out of the eight
buffered references. All the other 5 references will be specified
using the frame offset info. Hence, compared to the baseline, the use
of frame_refs_signaling provides 2 scenarios:

(1) Reference short-signaling:
    3*7 - (1+3*2) = 14 bits are saved in the frame header;
(2) Reference regular-signaling:
    1 bit is added to the frame header.

BUG=aomedia:1392

Change-Id: I2d4ff0e367a4df107235c1e7066dd28cb2c60431

BUG=aomedia:1348
BUG=aomedia:1347
BUG=aomedia:1345

Change-Id: Ibe94d826b06e44e4ffad5a6ddc24f962b889ae17

- Fixed SSE4.1 implementation of the highbd warp filter.
- Removed the SSE2 and SSSE3 implementations of both the lowbd and highbd
  versions of the warp filter, as these will not work with CONFIG_JNT_COMP=1.
  (There are now only SSE4.1 implementations)
- The warp filter tests have been correspondingly updated.

Change-Id: I2bc1ebb69f2ed876803902908586c3848c03c11e

The compiler was complaining about the 'bd' parameter being
unused in two functions.

Change-Id: Iac2aa86234bb9554bb2dfa6ee6628bd1d09bfb1e

Change _mm_srl_epi32 to _mm_sra_epi32. The difference is keeping
sign bits in shifting. Otherwise there are mismatchings between simd and c
functions, caught by unit tests.

The other change is fix a typo.

Change-Id: I18a9e9e12a661d181e9e904ffc64d802d624a36d

BUG=aomedia:1404

Change-Id: Ie52ef049a33a47915869a424f2ca73d706d8e5d0

The experiment is now fully adopted.

Change-Id: I09960ed25defae98d1c3e5ee8bc93691f79677ec

This fixes a build warning.

Change-Id: I1a7ca22e2a793c873df17d2335e964f4a82f7ea1

Change-Id: Ia5ab13d7219ae79e685c999692a9f8bc35f0d360

This is to reduce the size of static tables.

BUG=aomedia:1030

Change-Id: Id9a00352560af38eb381f403510ec3f3d7f1c29a

Allow the CDF update to operate in different modes, e.g. update vs no
update.

The update mode is transmitted in the uncompressed frame header of
every keyframe and intra-only frame.

This patch only adds bitstream signaling and API support. The
implementation of the update modes will be in later patches.

Change-Id: Ic9fcd60e8a75f9c01f414253823d78cf9b3113dd

In av1_iht8x4_32_add_sse2, iidtx8 was using uninitialized values for
rows 4,5,6,7.

Change-Id: I12b41cedde6cdcc4c153a375571853a10d26c2df

This is correct in the spec, but the code wasn't doing the check.

BUG=aomedia:1405

Change-Id: I6e22f78d724583ad4cfdbbffc15e61b636810f76

Change-Id: I19c49e5a5f0df0d767f156dcf006e5bf105bf6a0

Introduced in Change-Id: I773f7ec9202d17c57bba44a698350e05101fa871

Change-Id: I147cce7c02db7144d8687855885910dbf8d9ab29

Fixes the mismatches reported in the issue tracker.

BUG=aomedia:1328

Change-Id: If60ef1c6cb5c28e21a6d6e71379fe1cb46ab6f0a

covers horizontal and vertical variations and
including low and high bitdepth types.

Appropriate tests are enabled

Performance changes, SSE2 over C:
Horizontal methods: up to  3x
Vertical   methods: up to  2x

Change-Id: If430a916394c7befa743e4fbaa9913fd37c535ed

Make the av1_convolve_x_sr_sse2/avx2 support various bit
shift options.

Addition of asserts in the convolve functions.

Change-Id: Ib6d1ada6c00a20e6e498af2672bd0bb76040d7d0

Fixes several failing 12-bit tests.

Change-Id: If6da50bbd595adb270cb11820df97b34a22d7b82

BUG=aomedia:1400

Change-Id: I9cf4f0de4fb15295e4acbd8b511d93abd7897165

This aligns with the intention of data size, reduces stack size.

Change-Id: Ib772143c20c860c871a1e4399f4a3d1024499680

The shift is as large as 49, which does not fit in 32 bit.

Change-Id: Ic84e4eacfde7be386cbff411391366e0559a2824

Change-Id: I1b74c2efcef553f7d9a5a4a701ff29fa905a61f5

Tested in the low latency setting. Comparing the three options:

(1). Cut off temporal mv altogether.
(2). Use collocated mv referncing.
(3). Use mfmv referencing.

The current baseline is using (1).

The results are:
(2) improves the compression performance over (1) by -0.59% for
lowres and -0.65% for midres.

(3) improves over (1) by -1.05% for lowres and -1.43% for midres

BUG=aomedia:1403

Change-Id: Ie683110ecbe16131f8aea123bca490f8897ebdd8

This avoid warnings of double->int conversion.

Change-Id: I61ed7675354dfe4292511409dd56c534c2b48914

The restoration buffers are freed and allocated at every frame, which
is not necessary. This patch makes the allocation to happen only when
it is necessary.

Change-Id: I773f7ec9202d17c57bba44a698350e05101fa871

The HIGHPRECISION_INTBUF=1 enables round_0 to be 3.
This is now merged into the LOWPRECISION_BLEND flags
for ease of management.

Change-Id: Ia336f24ea87d824d7034012cfed9c4c3129bf6eb

By default turn:
CONFIG_LOWPRECISION_BLEND=2
CONFIG_HIGHPRECISION_INTBUF=1

This is the adopted configuration from the codec WG.

Change-Id: Ic8a324a4bf6316753bea07a17160ab93afacca3b

The compound tests are not needed for the Sr convolve
functions, and are broken with LOWPRECISION_BLEND.

Change-Id: I426bf872e70fc013cbd38f2bf0585aa920f8debd

The CL makes the context design for single reference frame coding the
same as that for the compound reference frame coding. There are 3
contexts designed for each of the binary symbols for the single
reference frame scenario, and the designed contexts simply rely on the
counts of the references used in the neighboring two blocks.

Once this CL is merged, the coding of the reference frames, regardless
of single prediction or compound prediction, will all follow the same
context design pattern for all the binary symbols. The design logic is
much simpler and the lines of code for each binary symbol context
identification are reduced by 80%.

Further, this CL has obtained a small coding gain for 30 frames with
the default coding tools:

lowres: avg_psnr -0.015%; ovr_psnr -0.021%; ssim -0.002%
midres: avg_psnr -0.108%; ovr_psnr -0.139%; ssim -0.135%

BUG=aomedia:1402
BUG=aomedia:973

Change-Id: Ia72a1d18e85ac3a05308675b60b95f80f2219c46