GitLab

- Unit tests are added for AVX2 SIMD.
- Encoder speed improvement:
  AV1 baseline and EXT_TX, three 1080p sequences at bitrate:
  800 Kbps, 2 Mbps, 6 Mbps, on i7-6700 CPU, average
  user level time reduction: 3.86%.

Change-Id: Ibbd7837ee3a831c6b1e4e471bf6c8d3fa3a19ff4

This allows the hardware decoder to start decoding ref_mv_idx
syntax prior to the sorting stage and hide the latency of entropy
decoding. The compression performance change is about 0.01% level.

Change-Id: I86b34f31f6c99a36ae2780416175cc0bd90ff492

Change-Id: I8efb64eac3438484e7a77a8a1db198223fc52bfa

Change-Id: If36417c1384646da57453344b208e7653a4d31e5

https://bugs.chromium.org/p/webm/issues/detail?id=1306

Change-Id: Icd11d373ff08954121c097728e4c7791791e223f

This is a similar change to following aom CL
https://aomedia-review.googlesource.com/#/c/1961/

Move SIMD related functions from filter.c/h to following files
av1_convolve_ssse3.c
av1_highbd_convolve_filters_sse4.c

Change following c files to header files.
av1_highbd_convolve_filters_sse4.c
av1_convolve_filters_ssse3.c

Change-Id: I41a3cc6b0789e632451aeda82f5eb97a4d78e370

BUG=webm:1305

Change-Id: I4c56631359e298b99e618c07bcbae9f793c5e2ac

- Move filter look-up functions to corresponding optimization modules.

BUG=webm:1296

Change-Id: I87f399609052db2dbc7e5a590afb08b82e3fa89f

Change-Id: I9a08bc5da1a84c3d4b8fe2d457bb80406c0bc028

Change-Id: I4175070840a6561c1cec5f5a50b64e425f3e2926

Change-Id: I22dacbc2e4933a60ce7151204af9ee253990ca1f

This test checks if there's any basic change to the bitstream or default
encoder by running an encode and checking that the md5 from the decode
doesn't change.

Any change to the default encoder or bitstream should be accompanied by
a change to the md5 in this file.

Change-Id: Ibdd5a1442296fd3e946823ec1f43e8ac4e66dd34

Change-Id: I0dcb0643cf83ee99b63336df851cbca749c11b68

Change-Id: I134d276234b3b8aa7df1ab647892b5d739647f4c

Refactor to streamline the number of profiles needed, in
preparation for the next steps.

NO change in performance.

Change-Id: I753b89299897857f3c250c316b4cdc4fedcb90e8

cherry-picked from aom/master:
2b407394

Change-Id: Ia7b3bfd68e2c31b21f49a429fecc4d0b701b045f

Cherry-picked from aom/master:
e2721a65

Change-Id: I4ade58be91e7bca0cc4f2bed98a43177d7f590a5

Change-Id: Ib5226d4fe3dcf916fe8954c7240966e3a32eed31

Change-Id: I4fea280d72d7e428f2ab0820fd728997d5a903c9

Change-Id: Ia33ce4918d3d40eba331f81909f3f1f0f3ab7a58

When the block has width/height above or equal to 64, use 16x16
block search step for reference motion vector search in the non-
immediate rows and columns.

Change-Id: If11ce97a9328b879f30ef87115086aa0cd985a2f

Use 16x16 block resolution for collocated reference motion vector
search.

Change-Id: I1091b5b178e255eb6cc0b994de360994f7661b79

* changes:
  Remove multi-entropy coder hacks from the treewriter
  Rename rans_dec_lut to rans_lut

Change-Id: I9152f898dfacdf3877ed719f193bb1e0dbee0a1a

Fix the logical OR computation in .mk file. Otherwise, when both
experiments are on, the output of $(filter... will be two 'yes',
which will cause missing library issue.

Change-Id: I53c44e925dc9ea77c7467217c20e4f1bc7e20fc3