)]}'
{
  "commit": "153c72c6705e6622c1d74679c30c20bcbdebad42",
  "tree": "04688d55be8211534f919497355bb566f1267843",
  "parents": [
    "2d4ae3b7134294311f15c886c944e88f499b37b3"
  ],
  "author": {
    "name": "Rohan Baid",
    "email": "rohan.baid@ittiam.com",
    "time": "Fri Mar 06 13:44:55 2026 +0530"
  },
  "committer": {
    "name": "Rohan Baid",
    "email": "rohan.baid@ittiam.com",
    "time": "Fri Mar 06 13:44:55 2026 +0530"
  },
  "message": "Improve av1_convolve_x_sr_general_avx2()\n\nThis patch introduces the following optimizations:\n- Added a specialized width wise handling for 2/4 tap filtering\n- Improved the existing 6/8-tap filtering for w \u003e\u003d 16\n\nThe scaling w.r.t existing avx2 implementation is provided below:\n\nBlockSize  2tap   4tap   6tap   8tap\n2x2        1      0.94    -      -\n4x4        0.95   1       -      -\n8x8        0.95   0.98   0.97   1\n16x16      0.99   0.98   1.01   1\n32x32      1      0.93   0.99   0.97\n64x64      0.99   1.01   0.99   1\n128x128    1      1      1      1\n\nChange-Id: I7f90448107f8d906a4cafd2a2c00161512c8bf6b\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "727156268abdc68901a5e37fd6a64ce675af364b",
      "old_mode": 33188,
      "old_path": "aom_dsp/x86/convolve_avx2.h",
      "new_id": "76a6ca0c5515d9a0df2cfd61d605e9ca28c5a85f",
      "new_mode": 33188,
      "new_path": "aom_dsp/x86/convolve_avx2.h"
    },
    {
      "type": "modify",
      "old_id": "d250d88427898930bc66c5b16a2434ef38fa9446",
      "old_mode": 33188,
      "old_path": "av1/common/x86/convolve_avx2.c",
      "new_id": "f6ebc15313ba263e2539d01fd507f09ca247499b",
      "new_mode": 33188,
      "new_path": "av1/common/x86/convolve_avx2.c"
    }
  ]
}