1. 25 Mar, 2016 1 commit
  2. 24 Mar, 2016 2 commits
  3. 23 Mar, 2016 2 commits
    • Yi Luo's avatar
      Misc. updates for highbd changes · 659c2c98
      Yi Luo authored
      - Use Makefile to control the build for highbd_fwd_txfm_sse4.c.
      - Fixed hybrid transform (HT) types due to recent update.
      - Added new unit test cases for highbd HT.
      Change-Id: Ifd768a9b429a8c21ed40c1de8152fb5ac71e2f90
    • Debargha Mukherjee's avatar
      Make the tile independence test shorter · a61e5062
      Debargha Mukherjee authored
      Uses 15 frames instead of 30. Also only test speed 0 for VP10.
      Change-Id: Icace050edd974622d83bdf843058f63bd8d3a84b
  4. 21 Mar, 2016 2 commits
    • Debargha Mukherjee's avatar
      Adds 1D transforms for ADST/FlipADST to make 16 · 1b175593
      Debargha Mukherjee authored
      Makes a set of 16 transforms total, adding all 1D
      combinations of ADST and FlipADST, and removng all DST
      lowres, midres both improve by about 0.1% and hdres by
      -0.378% in BDRATE but with fewer transforms that are also
      Further experiments to continue later.
      Change-Id: I7348a4c0e12078fdea5ae3a2d36a89a319ffcc6e
    • Angie Chiang's avatar
      mv vp10_fwd_txfm2d_#x# into vp10_rtcd.h · abd447e3
      Angie Chiang authored
      Change-Id: Iad7352698786791b0fd7c005a7edfd1724b71599
  5. 17 Mar, 2016 1 commit
  6. 14 Mar, 2016 1 commit
  7. 08 Mar, 2016 1 commit
    • Yi Luo's avatar
      Implemented DST 16x16 SSE2 intrinsics optimization · 50a164a1
      Yi Luo authored
      - Implemented fdst16_sse2(), fdst16_8col() against C version: fdst16().
      - Turned on 7 DST related hybrid txfm types in vp10_fht16x16_sse2().
      - Replaced vp10_fht10x10_c() with vp10_fht16x16_sse2() in
      - Added vp10_fht16x16_sse2() unit test against C version:
        vp10_fht16x16_c() (--gtest_filter=*VP10Trans16x16*).
      - Unit test passed.
      - Speed improvement: 2.4%, 3.2%, 3.2%, for city_cif.y4m, garden_sif.y4m,
        and mobile_cif.y4m.
      Change-Id: Ib30a67ce5d5964bef143d588d0f8fa438be8901f
  8. 07 Mar, 2016 2 commits
    • Yi Luo's avatar
      Added vp10_fht8x8_sse2() unit test · 6ab06212
      Yi Luo authored
      - Inherited base class TransformTestBase to derived class VP10Trans8x8HT.
      - Employed RunCoeffCheck() to test vp10_fht8x8_sse2() against C reference
        function vp10_fht8x8_c().
      - fdst8_sse2() related seven hybrid transform cases are covered in this
      - Test passed (4 test cases w/o EXT_TX; 16 test cases with EXT_TX).
      Change-Id: Id9a9b308c707164a120d9ceb2c30e572026fb1d0
    • Geza Lore's avatar
      Extend convolution functions to 128x128 for ext-partition. · 938b8dfc
      Geza Lore authored
      Change-Id: I7f7e26cd1d58eb38417200550c6fbf4108c9f942
  9. 04 Mar, 2016 1 commit
    • Yi Luo's avatar
      Added vp10_fht4x4_sse2() unit test · 267f73a1
      Yi Luo authored
      Inherited class TransformTestBase to derived class VP10Trans4x4HT.
      Employed RunCoeffCheck() to test vp10_fht4x4_sse2() against
      C reference vp10_fht4x4_c().
      fdst4_sse2() related seven hybrid transform cases are covered
       in this test.
      Wrote a header file for test base class. Some modification to
      make sure the base class can be used for 8x8, 16x16, 32x32 cases.
      All related tests passed.
      Change-Id: I6b19a39d3ea30b657847781e78e73b829998a57a
  10. 03 Mar, 2016 2 commits
    • Geza Lore's avatar
      Add 128 pixel variance and SAD functions · 697bf5be
      Geza Lore authored
      Change-Id: I8fde245b32c9e586683a28aa6925da0b83850b39
    • Alex Converse's avatar
      ANS: Switch from PDFs to CDFs. · 6bbbe316
      Alex Converse authored
      Make the RANS implementation operate on cumulative distribution
      functions rather than individual probability distribution functions.
      CDFs have shown themselves more flexible to work with.
      Reduces decoding memory usage from scaling O(num_distributions *
      symbol_resolution) to O(num_distributions).
      No bitstream change. This is an purely implementation change.
      Change-Id: I4e18d3a0a3d37a36a61487c3d778f9d088b0b374
  11. 02 Mar, 2016 1 commit
  12. 26 Feb, 2016 1 commit
  13. 25 Feb, 2016 1 commit
    • Angie Chiang's avatar
      convolve8 sse2 test · 8878fa4f
      Angie Chiang authored
      This experiment shows that when frame size is 64x64
      vpx_highbd_convolve8_sse2 and vpx_convolve8_sse2's speed are similar.
      However when frame size becomes 1024x1024
      vpx_highbd_convolve8_sse2 is around 50% slower than vpx_convolve8_sse2
      we think the bottleneck is from memory IO
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_8_64 (17 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_16_64 (42 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_32_64 (139 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_64_64 (499 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_8_64 (16 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_16_64 (40 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_32_64 (130 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_64_64 (485 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_8_1024 (32 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_16_1024 (61 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_32_1024 (196 ms)
      VP10ConvolveTest.vpx_highbd_convolve8_sse2_speed_64_1024 (694 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_8_1024 (21 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_16_1024 (44 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_32_1024 (138 ms)
      VP10ConvolveTest.vpx_convolve8_sse2_speed_l_64_1024 (491 ms)
      Change-Id: I3131a031e0380e8eae748cfcccc6cbb961d05943
  14. 24 Feb, 2016 1 commit
  15. 23 Feb, 2016 1 commit
  16. 22 Feb, 2016 3 commits
    • Yaowu Xu's avatar
      Cleanup psnr.h · 38cfc45e
      Yaowu Xu authored
      Change-Id: Id026e72ee655ee5bd645a89e378da0d462be367d
    • Yaowu Xu's avatar
      Add shift stage in FASTSSIM computation · d1c5cd4a
      Yaowu Xu authored
      This commits adds a shift stage for FASTSSIM computaton when source
      bit depth is different from working bit depth, to make sure metric
      results are calculated in bit_depth consistent with source.
      Change-Id: I997799634076ef7b00fd051710544681ed536185
    • Yaowu Xu's avatar
      Move psnrhvs function declaration to psnr.h · 6e695da2
      Yaowu Xu authored
      From "ssim.h"
      Change-Id: Ie53378794149ef8a844b4eb47ad4f08579de4b60
  17. 21 Feb, 2016 1 commit
    • Yaowu Xu's avatar
      Extend HBDMetricTest · f6a7b17a
      Yaowu Xu authored
      This commit extends the HBDMetricTests to handle testing for metric
      computation where input source depth is different from working bit
      Change-Id: I5d11101cc9603a3fd09e8439816bb982a0f1b654
  18. 20 Feb, 2016 1 commit
    • Angie Chiang's avatar
      Fix 12 TAP convolution bug · 1e403064
      Angie Chiang authored
      Priviously, we do 12-tap interpolation even there is no sub pixel,
      This could cause a bug becuase decoder doesn't extend border when there
      is no sub pixel. In this situation, if we still do interpolation, we
      will access the border extension which doesn't exist and cause a
      memory error
      Change-Id: I55b879722f0a10c5d13261bd9617a75c826a2418
  19. 17 Feb, 2016 10 commits
  20. 16 Feb, 2016 5 commits