1. 10 Oct, 2017 8 commits
    • Yushin Cho's avatar
      Fix that sse is added twice during early skip in var-tx · 16efec40
      Yushin Cho authored
      The rd_stats->sse is already updated by
      "rd_stats->sse += tmp << 4;",
      which is measured by pixel_diff_dist(), i.e. in pixel domain and
      w/o quantization().
      
      Change-Id: I4dc20a7e80af9dd846aa5de4298cb56e7f0d8f7e
      16efec40
    • Debargha Mukherjee's avatar
      Turn on 32x64 and 64x32 transforms for real · cce6692a
      Debargha Mukherjee authored
      Change-Id: Ie4382b8a1c0f87ce50e9afefd1cef8ca55435c61
      cce6692a
    • Hui Su's avatar
      Remove unused parameter in intra mode reader · aa2965e6
      Hui Su authored
      Change-Id: Ibea4c2c732b16851ad16b475ea40f021d5b5d5b3
      aa2965e6
    • Sarah Parker's avatar
      Compute global refmv candidate at center of current block · 0a5cc5fd
      Sarah Parker authored
      When a neighboring block uses global motion, use the mv
      computed at the center of the current block as the candidate vector
      rather than the mv computed at the center of the neighboring block.
      
      0.15% improvement on cam_lowres
      
      Change-Id: I79eff8bf27a7aa84ae4a6d56e4a10c41a4438fb9
      0a5cc5fd
    • Yaowu Xu's avatar
      Revert "soft enable CDEF-singlepass" · 1542157b
      Yaowu Xu authored
      Temporarily reverting this to allow investigation of a couple of BUGS
      
      BUG=aomedia:881
      BUG=aomedia:887 (merged into #881)
      
      This reverts commit b1d3eda9.
      
      Change-Id: I2605deb7b8fefa4236d78c8695025dc42316edd2
      1542157b
    • Rupert Swarbrick's avatar
      Don't trash memory in select_tx_type_yrd · de2ea94e
      Rupert Swarbrick authored
      This patch fixes a bug in select_tx_type_yrd. The function works by
      looping over possible transform types to find the best option (calling
      select_tx_size_fix_type for each). Whenever there's a new best
      candidate, the code copies information about the transform from the
      mbmi structure into stack-allocated "best candidate" structures. At
      the end, it copies the "best candidate" data back to mbmi.
      
      Before the patch, if ref_best_rd was small, each call to
      select_tx_size_fix_type might return INT64_MAX (because they don't
      find anything better than ref_best_rd) and so we'd never actually copy
      anything to the "best candidate" structures. Then, at the end of the
      function, we'd merrily overwrite mbmi with whatever happened to be on
      the stack, causing general mayhem when something tried to read the
      data from mbmi later.
      
      This patch exits early if no candidates were found. It also adds an
      assertion saying that if no candidates were found, ref_best_rd must
      have been less than INT64_MAX. This should hopefully catch any bugs
      where the continue keywords in the loop stop us ever actually calling
      select_tx_size_fix_type.
      
      Change-Id: I54b998148281dd80f98d1570f736964593dc753f
      de2ea94e
    • Rupert Swarbrick's avatar
      Add an SSE4.1 implementation of av1_highbd_convolve_2d_scale · 724d31eb
      Rupert Swarbrick authored
      For large blocks this is about 8x the speed of the C version. The code
      needs SSE 4.1 for the PMULLD instruction that we use to do SIMD 32-bit
      multiplies.
      
      The patch uses av1_convolve_scale_test (written already to test the
      low bit depth path) to make sure the optimised code matches the C
      version.
      
      Change-Id: I9304d6bb3d2cb31390de93ed08ff1a852e3ace86
      724d31eb
    • Rupert Swarbrick's avatar
      Add an SSE4.1 implementation of av1_convolve_2d_scale · 98dc22b8
      Rupert Swarbrick authored
      For large blocks this is almost 8x the speed of the C version. The
      code needs SSE 4.1 for the PMULLD instruction that we use to do SIMD
      32-bit multiplies.
      
      This patch also makes av1_convolve_scale_test actually test something,
      making sure the optimised code matches the C version. The slightly
      excessive generality in the test (all the templating) is because of a
      following patch, which is for the high bit depth path and can then use
      most of the same test code.
      
      Change-Id: I6732bc6b2378ffaadae5aa6441100cf660f7ee11
      98dc22b8
  2. 09 Oct, 2017 11 commits
    • Angie Chiang's avatar
      Avoid updating non-used transform · ca8016ef
      Angie Chiang authored
      Since 32x32 transform use DCT only, we can avoid update other
      types of transform
      
      Change-Id: I51dd8ec71975187d249d7e25130e994a48cac5c1
      ca8016ef
    • Sarah Parker's avatar
      Change rectangular vartx recursion depth to 2 · d25ef8c6
      Sarah Parker authored
      0.15% improvement on lowres set
      
      Change-Id: If16a8e07797c64508f9e2d9b26ae874ac53c57a4
      d25ef8c6
    • Rupert Swarbrick's avatar
      Catch invalid block sizes in bitstream · 415c8f1f
      Rupert Swarbrick authored
      There's a bitstream conformance requirement that says that any block
      must subsample to a valid block size with the current subsampling
      mode. For example, this means that BLOCK_4X8 is illegal if there is
      subsampling in only the horizontal direction (since there is no
      BLOCK_2X8).
      
      This patch checks the bitstream is conformant as it reads partition
      information in decodeframe.c
      
      BUG=aomedia:875
      
      Change-Id: I18139aa76d6f965282402edbb0b68959478a46c3
      415c8f1f
    • Urvang Joshi's avatar
      Revert wrong uses of TX_SIZE enum. · ab8840eb
      Urvang Joshi authored
      Introduced by: https://aomedia-review.googlesource.com/c/aom/+/25181
      
      Change-Id: I1f25178d6b273fbeade4c33f153b5f2bac4a8b99
      ab8840eb
    • Rupert Swarbrick's avatar
      Add av1_convolve_scale_test · 1ea7ab4e
      Rupert Swarbrick authored
      This unit test doesn't actually provide any test coverage and merely
      exists to benchmark the C function, av1_convolve_2d_scale_c. The
      following patch will add an SSE version of that function and extend
      this test to check that the SSE code matches the C code.
      
      Change-Id: Ic942ad8f9fd57d2659fc60f92c5a0b6c9a9f8cac
      1ea7ab4e
    • Debargha Mukherjee's avatar
      Enable 32x64 and 64x32 transforms · 1a86b013
      Debargha Mukherjee authored
      Change-Id: I73e9d2d327b062828a75bc99fb348441dd32174a
      1a86b013
    • Debargha Mukherjee's avatar
      Resolve some static analysis warnings · e36a08c4
      Debargha Mukherjee authored
      Change-Id: Iaff923f34100ecdce76d2319fab67cde59d485ae
      e36a08c4
    • Cheng Chen's avatar
      Match braces in VIM for rdopt.c · 1483a714
      Cheng Chen authored
      Change-Id: I23344af711d9a31b819fca35ae3ad3b7edf4852e
      1483a714
    • Rupert Swarbrick's avatar
      Define block_signals_txsize function · fcff0b25
      Rupert Swarbrick authored
      This returns true if a block signals tx_size in the stream and uses it
      in the bitstream writing code and the decoder.
      
      Note that we can't quite use it in pack_inter_mode_mvs when
      CONFIG_VAR_TX && !CONFIG_RECT_TX but I've switched the code to using
      it the rest of the time since rect-tx is adopted and eventually the
      other code path should be deleted.
      
      Also use the helper function in tx_size_cost in rdopt.c, where the
      test was wrong and caused underestimates of block
      costs. (Specifically, the code that subtracts tx_size_cost from
      this_rate_tokenonly in rd_pick_intra_sby_mode ended up subtracting
      zero for a 4x8 block).
      
      The behaviour of the decoder should be unchanged. The only change in
      the encoder's behaviour should be in tx_size_cost where it should now
      match the rest of the code.
      
      Change-Id: I97236c9ce444993afe01ac5c6f4a0bb9e5049217
      fcff0b25
    • Zoe Liu's avatar
      Add experiment ext_skip · a3c5b9da
      Zoe Liu authored
      This coding tool is to introduce a new prediction mode for the
      bi-predictive frames that have a forward referernce within 2 frames
      away (distance denoted as 'fwd_delta'), and a backward reference,
      within (3-fwd_delta) frames away.
      
      If this prediction mode, namely 'ext_skip' is set, it will be coded
      using compound prediction with the most recent forward and backward
      reference frames as its reference pair, NEARESTMV as its motion mode,
      and the skip flag is set for the residue.
      
      Change-Id: I826034ccf1a956f4b350f0bc2e2dca8ea71b5197
      a3c5b9da
    • Zoe Liu's avatar
      Add encoder/decoder support to frame_sign_bias · 17af2748
      Zoe Liu authored
      Frame sign bias value will not be signaled in frame header. Instead,
      the sign bias of reference frames are derived from their corresponding
      frame offsets at both encoder and decoder.
      
      The tool of 'frame_sign_bias' is dependent of 'frame_marker'. Compared
      against baseline, the enabling of both tools obtains a small coding gain
      of -0.08 ~ -0.11% in BDRate over Google lowres/midres tests.
      
      Change-Id: I8d85dc427ced0b2152712ccf61be4be6068075b9
      17af2748
  3. 08 Oct, 2017 10 commits
  4. 07 Oct, 2017 11 commits
    • Luc Trudeau's avatar
      [CFL] Support for 4:4:4 High Bit Depth · 69d9e878
      Luc Trudeau authored
      Change-Id: I13ba0dbe57297b540b78512d21a119f05a86a849
      69d9e878
    • Luc Trudeau's avatar
      [CFL] Support for 4:2:0 High Bit Depth · 056d1f40
      Luc Trudeau authored
      high bit depth (_hbd) and low bit depth (_lbd) versions
      of the cfl functions: sum_above_row, sum_left_col,
      cfl_build_prediction, cfl_luma_subsampling_420 (4:4:4 will
      be added in subsequent commit) and cfl_alpha_dist. For
      cfl_alpha_dist, special care is given to scale the SSE
      according to the bit depth.
      
      BUG=aomedia:835
      
      Change-Id: I5b72845100d88fb8a438efe665bcae7fe1ba50b8
      056d1f40
    • Urvang Joshi's avatar
      Add a macro SCALE_HORIZONTAL_ONLY. · e58b564d
      Urvang Joshi authored
      When enabled, scaling through resize and superres will occur only in the
      frame's width; the height will not be scaled.
      
      Macro is off by default.
      
      Change-Id: I501b2b0b2766aa4a86da5937b57c4d5aee4e34c4
      e58b564d
    • Urvang Joshi's avatar
      inspect.c: Update to include new tx size enums. · b2752174
      Urvang Joshi authored
      Change-Id: I27292b7cdb27cec23754a6f017c5c7c55eb38bb5
      b2752174
    • Debargha Mukherjee's avatar
      Add cmake dependency constraint · c5600655
      Debargha Mukherjee authored
      ext-partition-types and supertx are incompatible
      
      Change-Id: I6c4cce16453cff13b0acbaad93dde7d089891038
      c5600655
    • Urvang Joshi's avatar
      FRAME_SUPERRES: Rework to use scale factor of 8/D · de71d142
      Urvang Joshi authored
      Earlier, the superres scale was in the form of:
      N/16, where N ranged from 8 to 16.
      
      We change this to the form:
      8/D, where D ranges from 8 to 16.
      
      This helps on the decoder side, by making it possible to work on 8x8
      blocks at a time.
      
      Change-Id: I6c72d4b3e8d1c830e61d4bb8d7f6337a100c3064
      de71d142
    • Urvang Joshi's avatar
      Superres: Remove unused variable. · e61923fa
      Urvang Joshi authored
      cm->superres_scale_numerator is used for both keyframes and
      non-keyframes, and is initialized from either
      oxcf->superres_scale_numerator or oxcf->superres_kf_scale_numerator as
      appropriate.
      
      Change-Id: Ie46df576ef3830e181643ae591d836449a4bd38f
      e61923fa
    • Rupert Swarbrick's avatar
      Upscale frame correctly in foreach_rtile_in_tile · b66894ac
      Rupert Swarbrick authored
      The restoration tiles (rtiles) divide the upscaled frame, not the
      encoded one.
      
      Change-Id: I2d08fe926d694fee7064461685289d3fd1c1de0c
      b66894ac
    • Debargha Mukherjee's avatar
      Remove the speed optimization for rd_stats_stack · 9245d89d
      Debargha Mukherjee authored
      This optimization for speed was useful only when max tx-size
      was 32x32. However with tx64x64 this was breaking certain assumptions
      causing huge drops in coding efficiency. So I am removing this
      optimization for now. This can be brought back latger as a speed feature.
      The removal of this optimzation brings back the loss when 32x64
      and 64x32 transforms are used.
      
      Change-Id: I15987ea9ff53fa36a2962fe5f156c30a11e809ed
      9245d89d
    • Joe Young's avatar
      [intra-edge] Pad intra edge samples to avoid valgrind warning · 7cfd5343
      Joe Young authored
      The SSE4 function filter_intra_edge_sse4_1() reads data slightly
      past the initialized part of the array. Those data are discarded
      later, but causes a valgrind warning. This change avoids the warning
      by initializing the array an extra +16 positions.
      
      BUG=aomedia:868
      
      Change-Id: Ib610492cff91492ae379c5d62895773f8747c4bc
      7cfd5343
    • Luc Trudeau's avatar
      [CFL] Extract sum top row and left col from custom DC_PRED · 13281d4f
      Luc Trudeau authored
      To simplify high bit depth commit, the summing the top row and the left
      column are extracted out of cfl_dc_pred. This does not change the
      bitstream.
      
      Change-Id: I5c9fe91df4942f736c5af29c1d93abb3a6c8501f
      13281d4f