import math
import pytest
import torch
from . import attri_util as attrs
from . import performance_utils as base
def _input_fn(shape, dtype, device):
yield {
"end": math.prod(shape),
"device": device,
"dtype": dtype,
},
if base.Config.bench_level == attrs.BenchLevel.COMPREHENSIVE:
yield {
"start": 0,
"end": math.prod(shape),
"step": 2,
"device": device,
"dtype": dtype,
},
@pytest.mark.arange
def test_arange():
bench = base.GenericBenchmark(
op_name="arange", input_fn=_input_fn, torch_op=torch.arange
)
bench.run()