1. 15 Feb, 2018 1 commit
    • Yaowu Xu's avatar
      Remove CONFIG_TX64X64 · d3d4159f
      Yaowu Xu authored
      The experiment is fully adopted.
      
      Change-Id: I6cc80a2acf0c93c13b0e36e6f4a2378fe5ce33c3
      d3d4159f
  2. 12 Feb, 2018 1 commit
    • Peng Bin's avatar
      Add inv txfm2d sse2 for sizes with 4 · 18976fa5
      Peng Bin authored
      Implement av1_lowbd_inv_txfm2d_add_4x4_sse2
      Implement av1_lowbd_inv_txfm2d_add_4x8_sse2
      Implement av1_lowbd_inv_txfm2d_add_8x4_sse2
      Implement av1_lowbd_inv_txfm2d_add_4x16_sse2
      Implement av1_lowbd_inv_txfm2d_add_16x4_sse2
      
      A brief speed test shows that using the included SSE2 functions
      completed by this CL, for speed1 lowbitdepth encoder speeds up >9%
      and lowbitdepth decoder speeds up >25%, comparing to the highbitdepth
      implementation in the baseline.
      
      Change-Id: I0576a2a146c0b1a7b483c9d35c3d21d979e263cd
      18976fa5
  3. 11 Feb, 2018 2 commits
    • Peng Bin's avatar
      Add inv txfm2d 64 sse2 · a7ba23f6
      Peng Bin authored
      Implement av1_lowbd_inv_txfm2d_add_32x64_sse2
      Implement av1_lowbd_inv_txfm2d_add_64x32_sse2
      Implement av1_lowbd_inv_txfm2d_add_16x64_sse2
      Implement av1_lowbd_inv_txfm2d_add_64x16_sse2
      
      Change-Id: I1b27618f153583cc787e7bf6ef1616e7c6932990
      a7ba23f6
    • Peng Bin's avatar
      Add inv txfm2d {8x32,32x8} sse2 · 008c6430
      Peng Bin authored
      Implement av1_lowbd_inv_txfm2d_add_8x32_sse2
      Implement av1_lowbd_inv_txfm2d_add_32x8_sse2
      
      Change-Id: Ibd5de72e1d2c4dabba5af020a06e8cfac329dc3d
      008c6430
  4. 10 Feb, 2018 1 commit
  5. 09 Feb, 2018 4 commits
  6. 08 Feb, 2018 2 commits
    • Peng Bin's avatar
      Implement av1_lowbd_inv_txfm2d_add_8x8_sse2 · 58bd7ced
      Peng Bin authored
      1. Change the output type from uint16_t to uint8_t;
      2. sse2 version comparing to av1_inv_txfm2d_add_8x8_c achieves 4x~9x
         speedup.
      
      Change-Id: I08675a018d11f9904d5b0a40aa896fe8fd437352
      58bd7ced
    • Angie Chiang's avatar
      Bitstream change - avoid overflow of identity txfm · a87dc765
      Angie Chiang authored
      1) Use 12-bit precision for square-root-2 scaling in identity txfms
      2) Use generic method to compute the identity txfm
      3) Test the identity config for tx64
      
      BUG=aomedia:1327
      
      Change-Id: I7350682bae9134b1f70a9ad80d5e64236f7728e2
      a87dc765
  7. 25 Jan, 2018 1 commit
    • Angie Chiang's avatar
      Simplify txfm table · 0c7b8d84
      Angie Chiang authored
      Instead of listing all possible stage_range,
      we use set_fwd_txfm_non_scale_range() to generate 2d stage_range
      from 1d stage_range.
      
      This will reduce the complexity of txfm table significantly.
      
      This is a lossless change.
      The coding performance isn't changed.
      The txfm config is exactly the same as it was before.
      
      Change-Id: Ibd1d9e53772bb928faaeecc98d81cbc8f38b27ed
      0c7b8d84
  8. 24 Jan, 2018 2 commits
    • Angie Chiang's avatar
      Simplify cos_bit setting in txfm · d4327bce
      Angie Chiang authored
      Move cos_bit from txfm 1d cfg to 2d cfg
      Each txfm stage only uses one cos_bit
      
      This is a lossless change and it speeds up encoder by 2%
      
      Change-Id: I45d398761e4729b8c4c37729571fe3765cb0c83f
      d4327bce
    • Johann's avatar
      adopt some clang 5.0.0 formatting · 123e8a60
      Johann authored
      At least the changes that don't conflict with 4.0.1
      
      Change-Id: Iaa2fda027b8ab2b023d608cf5ec7b377a72b851e
      123e8a60
  9. 23 Jan, 2018 3 commits
  10. 11 Jan, 2018 1 commit
    • Angie Chiang's avatar
      Move rect txfm's shift setting into config · a4c80485
      Angie Chiang authored
      For 2:1 txfm, move the sqrt2 scaling to the end of the fwd txfm
      function
      
      This only affects the txfm accuracy slightly but simplifies the
      range management a lot.
      
      Change-Id: I2eef8c15b7e8958a1fb0f2ae632604fdbab86046
      a4c80485
  11. 05 Jan, 2018 1 commit
  12. 04 Jan, 2018 1 commit
  13. 03 Jan, 2018 1 commit
    • Debargha Mukherjee's avatar
      Various fixes to scale managed txfms and tests · 1158bff7
      Debargha Mukherjee authored
      This patch clears all test failures with coeff range checking
      enabled for forward and inverse transforms. Also this
      patch ensures that there are no transposes for any of the
      rectangular transforms.
      Some fine-tunnig and refactoring are still pending.
      Some of the tests still need to be rewritten.
      
      Change-Id: Ib0e3a4ceccef665ba007d121f536fad7135f38d5
      1158bff7
  14. 29 Dec, 2017 1 commit
  15. 27 Dec, 2017 1 commit
  16. 21 Dec, 2017 1 commit
  17. 16 Dec, 2017 1 commit
  18. 15 Dec, 2017 3 commits
    • Urvang Joshi's avatar
      txfm2d_test: Relax some error thresholds. · 46e8fda0
      Urvang Joshi authored
      Thresholds needed to be relaxed when CONFIG_TXMG is off.
      
      All the transform tests pass with that config now, except ones related
      to TX_16X64, which is being looked at separately.
      
      BUG=aomedia:1114
      
      Change-Id: I974a742c66b5bb70c7050c4d3a317e159840d8e4
      46e8fda0
    • Yaowu Xu's avatar
      Add a missed array alignment · e3fb5689
      Yaowu Xu authored
      Change-Id: I4196cc5fa03353bcb760bfbf5d1182f0609f2475
      e3fb5689
    • Yaowu Xu's avatar
      Align arrays · b971c866
      Yaowu Xu authored
      Change-Id: I6f2a312bf34c87845688503907120dc915459e21
      b971c866
  19. 14 Dec, 2017 2 commits
    • Urvang Joshi's avatar
      AV1InvTxfm2d: Better test for txfms with approx. · 5dbb9b4c
      Urvang Joshi authored
      Transforms with one of the dimensions being 64 use an approximate 2D DCT
      transform, which makes round trips errors large for randomized input.
      
      So, for such transform sizes, we use a modified test that compares (a)
      reference forward transform + inverse transform to (b) forward transform
      + inverse transform.
      
      BUG=aomedia:1114
      
      Change-Id: Ic47b0037a0883add4c9f84ac16e03d75b258d5ae
      5dbb9b4c
    • Urvang Joshi's avatar
      AV1InvTxfm2d.RunRoundtripCheck: Add 1:4 transforms · e035b5d9
      Urvang Joshi authored
      Also added some missing 1:2 transforms to the test.
      
      BUG=aomedia:1114
      
      Change-Id: If5ee878c56ba32450931e00b70c00cd95b56359c
      e035b5d9
  20. 13 Dec, 2017 1 commit
  21. 12 Dec, 2017 1 commit
  22. 30 Nov, 2017 1 commit
    • Urvang Joshi's avatar
      Pass TXFM_2D_FLIP_CFG by pointer; don't return. · c502216a
      Urvang Joshi authored
      This avoids copying a whole struct on return. This is important, because
      this function is called frequently: once for every block that is transformed.
      
      Overall, aomenc speed seems to be improved slightly: by 0.15% - 0.30%.
      
      Change-Id: If44c351e1141c0fe0458daa2262aeb4afe44de46
      c502216a
  23. 28 Nov, 2017 1 commit
  24. 15 Nov, 2017 1 commit
  25. 02 Nov, 2017 1 commit
    • Sebastien Alaiwan's avatar
      Remove experimental flag of EXT_TX · 3bac9928
      Sebastien Alaiwan authored
      This experiment has been adopted, we can simplify the code
      by dropping the associated preprocessor conditionals.
      
      Change-Id: I02ed47186bbc32400ee9bfadda17659d859c0ef7
      3bac9928
  26. 30 Oct, 2017 1 commit
  27. 05 Oct, 2017 1 commit
  28. 09 Aug, 2017 1 commit
    • Angie Chiang's avatar
      Add txfm config test · 9c7089a9
      Angie Chiang authored
      This test makes sure two things:
      1) txfm stage range is within desired limit
          (lbd:16 bits hbd:32 bits)
      2) txfm stage range + cos bits is within desired limit
          (lbd:32 bits hbd:32 bits)
      
      Change-Id: Ie2cc3c8265810e034c1461def4717fa9d4c29945
      9c7089a9
  29. 08 Aug, 2017 1 commit