FHElium

Appearance

fhelium._cli.main ​

console ​

data View source

python
console = Console()

cli ​

function View source

python
def cli(ctx): ...

version_command ​

function View source

python
def version_command(): ...

Print the version of fhelium

benchmark_group ​

function View source

python
def benchmark_group(ctx, benchmark_file: str | None) -> None: ...

Open the benchmark TUI or use a non-interactive subcommand.

benchmark_list ​

function View source

python
def benchmark_list(ctx) -> None: ...

List registered benchmarks and profiles.

benchmark_run ​

function View source

python
def benchmark_run(ctx, benchmark_name: str, profile_name: str | None, overrides: tuple[str, ...], output: str | None) -> None: ...

Run one benchmark without opening the TUI.

benchmark_suite_run ​

function View source

python
def benchmark_suite_run(device: str, threads: int | None, output: str) -> None: ...

Execute every configuration, workload and planned state in the suite.

benchmark_recommend_group ​

function View source

python
def benchmark_recommend_group() -> None: ...

Run benchmark workloads with measured execution recommendations that recommend execution policies.

benchmark_recommend_ntt ​

function View source

python
def benchmark_recommend_ntt(suite: str, preset_name: str, device: str, warmup: int, runs: int, repetitions: int, backends: tuple[str, ...], seed: int, output: str | None) -> None: ...

Recommend an NTT backend without changing any default.

cuda_group ​

function View source

python
def cuda_group(): ...

Inspect CUDA devices and peer topology.

cuda_info ​

function View source

python
def cuda_info(device_id: int | None, as_json: bool) -> None: ...

Show a concise CUDA inventory and optional device details.

cuda_topo ​

function View source

python
def cuda_topo(bandwidth: bool, as_json: bool) -> None: ...

Show CUDA peer access and optional measured bandwidth.