)]}'
{
  "commit": "7802470342aa3336a45a000e116bcda92a037a36",
  "tree": "1afceddf4e63b5c0ef348c17aeb554263d2d13a1",
  "parents": [
    "fae21034041de1802fe696b3755a7af743b5a3af"
  ],
  "author": {
    "name": "George Steed",
    "email": "george.steed@arm.com",
    "time": "Fri Sep 08 11:42:20 2023 +0100"
  },
  "committer": {
    "name": "James Zern",
    "email": "jzern@google.com",
    "time": "Mon Sep 18 19:55:56 2023 +0000"
  },
  "message": "Rework round_shift_sqrt2 Neon helpers, use immediate shifts\n\nThe existing store_rect_16bit_to_32bit functions use a vector of shift\namounts which are always constant, so just use the immediate form\ninstead.\n\nAlso inline these functions into the store functions since that is their\nonly use and pull out a common helper function to handle the shifting\nitself instead, since this can be used for both the 4-wide as well as\nthe 8-wide code.\n\nThe fidentity functions can also make use of these helpers with some\nslight tweaks (such as needing `sqrt(2) * 2` in the fidentity8x16_neon\ncase). Also use the existing shift helpers in the fidentity functions to\nmatch.\n\nChange-Id: Ia63f47c61141119f0a8fe0dcb8a9144f6e70fd08\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "0725f19d1c1db1c0f76bd4715d929d7768a6ad03",
      "old_mode": 33188,
      "old_path": "av1/encoder/arm/neon/av1_fwd_txfm2d_neon.c",
      "new_id": "311af05f13556ac2515aafe4cfef353a80891d68",
      "new_mode": 33188,
      "new_path": "av1/encoder/arm/neon/av1_fwd_txfm2d_neon.c"
    }
  ]
}