)]}'
{
  "commit": "0b436b0d6554f27ebdcabe0fc4a85f97a1cfd2ae",
  "tree": "77da9f1e5b4db9b7de4a813358e2b2cf8c95e73d",
  "parents": [
    "0e1c1a328c813116c1257e5d759e6d9b6a435c80"
  ],
  "author": {
    "name": "Salome Thirot",
    "email": "salome.thirot@arm.com",
    "time": "Thu Jan 16 11:29:38 2025 +0000"
  },
  "committer": {
    "name": "James Zern",
    "email": "jzern@google.com",
    "time": "Mon Jan 27 12:35:56 2025 -0800"
  },
  "message": "Optimize Neon implementation of aom_lpf_14\n\nThe Neon implementations of aom_lpf_vertical_14 and\naom_lpf_horizontal_14 compute every filter (filter4(), filter8() and\nfilter14()) before selecting for each element which filter is actually\nneeded. In practice, however, a lot of cases only need one of the\nfilters, so specialize for these scenarios, computing only the filters\nthat are needed and eliminate bitwise select.  This makes the case where\nall filters are needed slightly slower, but as it is far from the most\ncommon case this is ok.\n\nAlso move the actual filter computation to separate functions to avoid\ncode duplication.\n\nThis is a port from 4d6f560d1f4a35a47e5dfc9b7c2dc4f5483d49dc in SVT-AV1.\n\nChange-Id: Id608edad6fee69fc263eb6ee9e0956ca6311d77f\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "7822a144b51fe31c370c5fa729f7b0e1156fce92",
      "old_mode": 33188,
      "old_path": "aom_dsp/arm/loopfilter_neon.c",
      "new_id": "fe201bc28ea6555cba85d943e1f618701227ae9a",
      "new_mode": 33188,
      "new_path": "aom_dsp/arm/loopfilter_neon.c"
    }
  ]
}