Energy

Energy modules for benchmarking

---

source

compute_energy_multi

def compute_energy_multi(
    model:torch.nn.Module, # model to benchmark
    sample:torch.Tensor, # input tensor (with batch dimension)
    devices:Sequence[str | torch.device] | None=None, # devices to benchmark (default: cpu + cuda)
    kwargs:VAR_KEYWORD
)->dict[str, EnergyMetrics]:

Measure energy on multiple devices.

---

source

compute_energy

def compute_energy(
    model:torch.nn.Module, # model to benchmark
    sample:torch.Tensor, # input tensor (with batch dimension)
    device:str | torch.device='cpu', # device to run on
    warmup:int=20, # warmup iterations
    steps:int=100, # measurement iterations
    offline:bool=True, # use offline emissions tracker
    country_iso:str | None=None, # country ISO code for carbon intensity
    measure_secs:int=1, # power sampling interval
)->EnergyMetrics:

Measure power consumption and carbon footprint using codecarbon.

---

source

EnergyMetrics

def EnergyMetrics(
    mean_watts:float, energy_wh:float, co2_eq_g:float
)->None:

Energy consumption and carbon footprint metrics.