VoxelForge/Standalone/Research/WorldAssemblerM3/Linux82/RunL82TNSlowdown.py

#!/usr/bin/env python3
"""Five interleaved TN LOD0-2 readiness pairs: 90378a5 vs 9cb5e53."""
from __future__ import annotations

import json
import re
import statistics
import subprocess
import time
from pathlib import Path

ROOT = Path(__file__).resolve().parents[4]
OUT = ROOT / "Saved/Linux/l82/tn-regression"
DATA = ROOT / "Standalone/data/owner_world.json"
BINARIES = {
    "90378a5": ROOT / "Saved/Linux/l76/verify/bin/VoxelForgeCoreHeadless-verify-clang",
    "9cb5e53": ROOT / "Saved/Linux/l81/bin/VoxelForgeCoreHeadless-l81-clang",
}
CAMERA = (19.166015625, -14.2001953125, -753.3740234375)
PIPELINE = re.compile(r"^\[StreamPipeline\] (.*)$", re.M)
FIELDS = re.compile(r"([a-zA-Z0-9_]+)=([^ ]+)")
NEAR = re.compile(r"^\[WorldAssemblerNearLODReady\] lods=0-2 tiles=(\d+)/(\d+) "
                  r"cold_ready_s=([0-9.]+)$", re.M)


def run(version: str, rep: int) -> dict:
    OUT.mkdir(parents=True, exist_ok=True)
    log = OUT / f"{version}-r{rep}.log"
    camera = ",".join(f"{v:.12f}" for v in CAMERA) + ",0,0"
    command = [
        "taskset", "-c", "0-11", "prlimit", "--as=6442450944", "--",
        "env", "VOXELFORGE_L81_PASSAGE_BOUND_PADDING_VOXELS=0", "timeout", "1740s",
        str(BINARIES[version]), str(DATA), "--world-assembler-prototype",
        "--stream-benchmark", "--stream-pipeline", "--workers=8", "--clip-radius=3",
        "--max-clip-level=8", "--max-level=2", f"--camera={camera}",
        "--profile-lod-work",
    ]
    start = time.monotonic()
    with log.open("w", encoding="utf-8") as stream:
        result = subprocess.run(command, cwd=ROOT, stdout=stream,
                                stderr=subprocess.STDOUT, check=False, timeout=1740)
    output = log.read_text(encoding="utf-8", errors="replace")
    pipeline_match, near_match = PIPELINE.search(output), NEAR.search(output)
    if result.returncode or pipeline_match is None:
        raise RuntimeError(f"TN run failed: exit={result.returncode} {log}")
    pipeline = dict(FIELDS.findall(pipeline_match.group(1)))
    whole_tiles = int(pipeline["whole_tiles"].split("/")[0])
    if near_match is not None:
        lod02_ready_s = float(near_match.group(3))
        lod02_tiles = int(near_match.group(1))
    else:
        # 90378a5 predates the explicit near-LOD milestone. With max-level=2,
        # the complete 0-2 view is the whole-view milestone.
        lod02_ready_s = float(pipeline["whole_view_ready"].removesuffix("s"))
        lod02_tiles = whole_tiles
    if lod02_tiles != 1344 or whole_tiles != 1344:
        raise RuntimeError(f"TN tile scope mismatch: lod02={lod02_tiles} whole={whole_tiles} {log}")
    row = {
        "version": version, "pair": rep, "exit_code": result.returncode,
        "wall_s": round(time.monotonic() - start, 3),
        "lod02_ready_s": lod02_ready_s,
        "whole_view_ready_s": float(pipeline["whole_view_ready"].removesuffix("s")),
        "whole_tiles": whole_tiles,
        "log": str(log.relative_to(ROOT)),
    }
    print("[L82TNRegressionRun] " + json.dumps(row, sort_keys=True), flush=True)
    return row


def main() -> int:
    for path in (*BINARIES.values(), DATA):
        if not path.is_file():
            raise SystemExit(f"missing input: {path}")
    rows = []
    for rep in range(1, 6):
        order = ("90378a5", "9cb5e53") if rep % 2 else ("9cb5e53", "90378a5")
        for version in order:
            rows.append(run(version, rep))
            (OUT / "runs-partial.json").write_text(json.dumps(rows, indent=2) + "\n",
                                                   encoding="utf-8")
    by_pair = {rep: {row["version"]: row for row in rows if row["pair"] == rep}
               for rep in range(1, 6)}
    paired_ready_delta = [by_pair[rep]["9cb5e53"]["lod02_ready_s"]
                          - by_pair[rep]["90378a5"]["lod02_ready_s"]
                          for rep in range(1, 6)]
    paired_whole_delta = [by_pair[rep]["9cb5e53"]["whole_view_ready_s"]
                          - by_pair[rep]["90378a5"]["whole_view_ready_s"]
                          for rep in range(1, 6)]
    summary = {
        "pairs": 5,
        "90378a5_lod02_median_s": statistics.median(
            row["lod02_ready_s"] for row in rows if row["version"] == "90378a5"),
        "9cb5e53_lod02_median_s": statistics.median(
            row["lod02_ready_s"] for row in rows if row["version"] == "9cb5e53"),
        "paired_lod02_delta_s": paired_ready_delta,
        "paired_lod02_delta_median_s": statistics.median(paired_ready_delta),
        "90378a5_whole_median_s": statistics.median(
            row["whole_view_ready_s"] for row in rows if row["version"] == "90378a5"),
        "9cb5e53_whole_median_s": statistics.median(
            row["whole_view_ready_s"] for row in rows if row["version"] == "9cb5e53"),
        "paired_whole_delta_s": paired_whole_delta,
        "paired_whole_delta_median_s": statistics.median(paired_whole_delta),
        "scheduling": "both binaries used their default scheduler settings; 90378a5 predates the policy flag",
        "runs": rows,
    }
    (OUT / "summary.json").write_text(json.dumps(summary, indent=2) + "\n",
                                      encoding="utf-8")
    print("[L82TNRegressionSummary] " + json.dumps(
        {k: v for k, v in summary.items() if k != "runs"}, sort_keys=True), flush=True)
    return 0


if __name__ == "__main__":
    raise SystemExit(main())