)]}'
{
  "commit": "60eca2df18fcd47d71ab136e460efabbc2f4a044",
  "tree": "31ced95391a7fad0e7e31ab7bd45ef26a13a303e",
  "parents": [
    "edd6f7692ad1bb30ba48dc74e0488cef074e402c"
  ],
  "author": {
    "name": "Cody Maloney",
    "email": "cmaloney@users.noreply.github.com",
    "time": "Sun Nov 15 00:31:55 2026"
  },
  "committer": {
    "name": "GitHub",
    "email": "noreply@github.com",
    "time": "Sun Nov 15 00:31:55 2026"
  },
  "message": "gh-161145: Test more bytes simultaneously in find_first_nonascii (#161147)\n\nLong ASCII runs now OR eight aligned words and test the high bits once,\none branch per 64 bytes (32 on 32-bit) instead of one per word.  When\na block contains a non-ASCII byte, the existing one-word loop finds it.\nIt is plain C: GCC and Clang compile it to SSE2/NEON loads, other\ncompilers to ORed word loads with the same single branch.\n\nMultiple compiler and architecture codegen sample: https://godbolt.org/z/8bM83vx3a\n\npyperf, PGO+LTO, 2 interleaved rounds:\n\n| Benchmark | main | this |\n|---|---:|---:|\n| (b\u0027a\u0027 * 4096).decode() | 165 ns | 108 ns: 1.53x faster |\n| (b\u0027a\u0027 * 200).decode() | 28.9 ns | 26.5 ns: 1.09x faster |\n| b\u0027a\u0027 * 200 + \u0027é\u0027 + ..., .decode() | 60.9 ns | 56.0 ns: 1.09x faster |\n| b\u0027a\u0027 * 136 + \u0027é\u0027 + ..., .decode() | 53.8 ns | 51.3 ns: 1.05x faster |\n| 8-100 byte decodes, \u0027é\u0027 within 100 | | within 1.2 ns |\n\n```python\nimport pyperf\n\nrunner \u003d pyperf.Runner()\ntail \u003d \u0027é\u0027.encode() + b\u0027a\u0027 * 50\nfor n in (8, 100, 200, 4096):\n    runner.timeit(f\"(b\u0027a\u0027 * {n}).decode()\", \"s.decode()\",\n                  globals\u003d{\"s\": b\u0027a\u0027 * n})\nfor n in (16, 100, 136, 200):\n    runner.timeit(f\"(b\u0027a\u0027 * {n} + \u0027é\u0027.encode() + b\u0027a\u0027 * 50).decode()\",\n                  \"s.decode()\", globals\u003d{\"s\": b\u0027a\u0027 * n + tail})\n```\n\nCo-authored-by: Claude Opus 5.5 \u003cnoreply@anthropic.com\u003e\nCo-authored-by: Victor Stinner \u003cvstinner@python.org\u003e",
  "tree_diff": [
    {
      "type": "add",
      "old_id": "0000000000000000000000000000000000000000",
      "old_mode": 0,
      "old_path": "/dev/null",
      "new_id": "5d809624e25f42060979ee7b2d987c05f2bf1dd1",
      "new_mode": 33188,
      "new_path": "Misc/NEWS.d/next/Core_and_Builtins/2026-11-01-16-02-42.gh-issue-161145.5vLdrm.rst"
    },
    {
      "type": "modify",
      "old_id": "da1ba44dba812903d7c85b03b5818504ab6f5bd1",
      "old_mode": 33188,
      "old_path": "Objects/unicodeobject.c",
      "new_id": "0fe765c36848f25affeb5c06a3e12f2e4a636c41",
      "new_mode": 33188,
      "new_path": "Objects/unicodeobject.c"
    }
  ]
}
