)]}'
{
  "commit": "2e124f462b87996c6d6ddb1fab843c04bdaa14a6",
  "tree": "25dc36754f80dd663a5c5aa9ff35cb1b84a7fde2",
  "parents": [
    "1e900cac01098ba2b996dab9683193daec971e2e"
  ],
  "author": {
    "name": "Gerda Zsejke More",
    "email": "gerdazsejke.more@arm.com",
    "time": "Tue Oct 28 14:51:07 2025 +0100"
  },
  "committer": {
    "name": "James Zern",
    "email": "jzern@google.com",
    "time": "Thu Nov 13 21:14:27 2025 -0800"
  },
  "message": "Optimize aom_quantize_b functions\n\nOptimize aom_quantize_b_neon, aom_quantize_b_32x32_neon,\naom_quantize_b_64x64_neon by removing the initial memset function\ncalls that write zeroes to qcoeff_ptr and dqcoeff_ptr. Instead,\nwrite those zero entries inside the processing loops.\n\nIn av1_quantize.c the aom_quantize_b functions are called with NULL\nvalues for qm_ptr and iqm_ptr. The unit tests also have the same\nbehaviour. When these pointers are not NULL, the\naom_quantize_b_helper_c function is called instead. Remove the NULL\nchecks for these pointer inside the quantize_b functions.\n\nChange-Id: I0032ac5346992fcb949ecaddf8214e3d0b62efd8\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "8d13e579a47ce9a3bca594e7c9490d33cd83db66",
      "old_mode": 33188,
      "old_path": "av1/encoder/arm/quantize_neon.c",
      "new_id": "e9c0ca761998c797ae872ff1309499e9edb59e00",
      "new_mode": 33188,
      "new_path": "av1/encoder/arm/quantize_neon.c"
    }
  ]
}