FHElium

Appearance

fhelium.benchmarks.suite.runner ​

Sequential suite execution, correctness qualification and run checkpoints.

collect_platform ​

function View source

python
def collect_platform(device: torch.device) -> dict[str, Any]: ...

Collect hardware and runtime fields used to compare complete runs.

measure ​

function View source

python
def measure(prepared: PreparedWorkload, device: torch.device) -> dict[str, Any]: ...

Measure a qualified callable; failed checks never produce latency rows.

run_suite ​

function View source

python
def run_suite(device: str, output_path: Path, *, progress: Callable[[str], None]=print, selected_cells: Sequence[SuiteCell] | None=None) -> dict[str, Any]: ...

Run the suite and checkpoint every result.

selected_cells is a diagnostic subset and produces a partial report. The public suite command always executes the full specification.