P9: scaling on a RAM disk (bench --work/--clean, cloud task p9b)

bench/scaling.py

@@ -15,6 +15,7 @@ import argparse
 import datetime as dt
 import json
 import os
+import shutil
 import subprocess
 import sys
 import time
@@ -61,6 +62,8 @@ def main() -> int:
     ap.add_argument("--samples", type=int, default=0, help="use the first N samples (0 = all)")
     ap.add_argument("--simurg", default=str(Path(sys.executable).with_name("simurg")))
     ap.add_argument("--pure-python", action="store_true", help="run with SIMURG_PURE_PYTHON=1")
+    ap.add_argument("--work", type=Path, help="folder for the GTC outputs (e.g. a tmpfs); default: --out")
+    ap.add_argument("--clean", action="store_true", help="delete each run's GTCs after timing it")
     a = ap.parse_args()
     a.out.mkdir(parents=True, exist_ok=True)
     idats = subset(a.idats, a.samples, a.out / "idats")
@@ -68,7 +71,7 @@ def main() -> int:
     env = {**os.environ, "SIMURG_PURE_PYTHON": "1" if a.pure_python else "0"}
     runs = []
     for t in (int(x) for x in a.threads.split(",")):
-        out = a.out / f"t{t}"
+        out = (a.work or a.out) / f"t{t}"
         t0 = time.time()
         cmd = [a.simurg, "genotype", "--bpm", a.bpm, "--egt", a.egt, "--idats", str(idats), "--out", str(out),
                "--threads", str(t), "--no-cohort-qc", "--autocall-date", "1/1/2026 12:00:00 AM",
@@ -76,6 +79,8 @@ def main() -> int:
         rc = subprocess.run(cmd, env=env).returncode
         wall = time.time() - t0
         setup, phase = phase_seconds(out / "logs" / "genotype.log")
+        if a.clean:
+            shutil.rmtree(out / "samples", ignore_errors=True)
         runs.append(
             {
                 "threads": t,