CLI API

The command line interface is implemented with Typer.

Typer command line interface for AI Hardware Benchmark Suite.

version_info()

Print package, Python, platform, and core dependency versions.

doctor()

Run installation health checks for the current environment.

benchmark(config_path=None, profile=None, model=None, backend=None, device=None, batch_size=None, precision=None, warmup=None, iterations=None, storage_root=DEFAULT_STORAGE_ROOT, report_dir=None, report_formats=None)

Run a benchmark session and persist the result.

profile_command(profiler=None, config_path=None, storage_root=DEFAULT_STORAGE_ROOT, report_dir=None)

Run a benchmark with profiling configuration recorded in the session.

generate_report(session_id, storage_root=DEFAULT_STORAGE_ROOT, output_dir=None, formats=None)

Generate reports for a stored benchmark session.

compare(baseline_session_id, candidate_session_id, metric_name='latency_mean_seconds', storage_root=DEFAULT_STORAGE_ROOT, output='table')

Compare one metric between two stored sessions.

export(session_id, storage_root=DEFAULT_STORAGE_ROOT, output_dir=Path('exports'), formats=None)

Export a session as JSON plus optional report formats.

config(config_path=None, profile=None, output='yaml')

Resolve and print the effective AIHW-Bench configuration.

completion()

Show shell completion installation guidance.

suite(action, benchmark=None, output_dir=Path('benchmarks'))

Inspect or materialize the official reproducible benchmark suite.

dashboard(storage_root=DEFAULT_STORAGE_ROOT, output_dir=Path('dashboard'))

Generate a standalone interactive dashboard from stored sessions.

assistant(session_id, storage_root=DEFAULT_STORAGE_ROOT, baseline_session_id=None, output_dir=None)

Explain a benchmark and optionally compare it against a baseline session.