Fix snapshot CPU mismatch: use v4 snapshots (4 CPU), reduce concurrency to 100

Daytona bakes CPU/memory at snapshot creation time. v4 snapshots have
4 CPU / 8 GB. Preflight now checks against 4 CPU per sandbox.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Bryan Helmkamp 2026-03-16 12:44:14 -04:00
parent 33a2d32fae
commit 4cd9e301ea
No known key found for this signature in database
2 changed files with 4 additions and 4 deletions

View file

@ -498,7 +498,7 @@ def main():
help="Output directory for eval results",
)
parser.add_argument(
"--max-workers", type=int, default=150,
"--max-workers", type=int, default=100,
help="Max concurrent eval sandboxes (default 100)",
)
parser.add_argument(
@ -516,7 +516,7 @@ def main():
setup_logging(args.output_dir)
# --- Preflight: check Daytona capacity --------------------------------
preflight_daytona(args.max_workers, sandbox_cpu=2)
preflight_daytona(args.max_workers, sandbox_cpu=4)
log.info("=" * 64)
log.info("SWE-bench Evaluation (Daytona)")

View file

@ -388,7 +388,7 @@ def main():
"--provider", default="anthropic", help="LLM provider",
)
parser.add_argument(
"--max-workers", type=int, default=150,
"--max-workers", type=int, default=100,
help="Max concurrent sandboxes (default 100)",
)
parser.add_argument(
@ -410,7 +410,7 @@ def main():
setup_logging(args.output_dir)
# --- Preflight: check Daytona capacity --------------------------------
preflight_daytona(args.max_workers, sandbox_cpu=2)
preflight_daytona(args.max_workers, sandbox_cpu=4)
log.info("=" * 64)
log.info("SWE-bench Evaluation")