Files
sglang/test/registered/kernels/ops/elementwise/test_launch_carveout.py
T

46 lines
1.5 KiB
Python

"""L1 preference must retain default occupancy and leave other devices untouched."""
import pathlib
import sys
import pytest
import torch
from sglang.kernels.jit.utils import load_jit
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
def test_launch_carveout():
"""A MaxL1-only policy loses occupancy for shared-memory-heavy kernels."""
current = torch.cuda.current_device()
untouched = [
device
for device in range(torch.cuda.device_count())
if device != current and not torch._C._cuda_hasPrimaryContext(device)
]
cases = [(smem, mode) for smem in (0, 32768) for mode in range(3)]
module = load_jit(
"test_launch_carveout",
cuda_files=[str(pathlib.Path(__file__).with_name("launch_carveout.cuh"))],
cuda_wrappers=[
(f"run_{smem}_{mode}", f"check_launch_carveout<{smem}, {mode}>")
for smem, mode in cases
],
)
expected = torch.arange(255, -1, -1, dtype=torch.int32, device="cuda")
output = torch.empty_like(expected)
for smem, mode in cases:
run = module[f"run_{smem}_{mode}"]
for _ in range(2):
output.fill_(-1)
run(output)
assert torch.equal(output, expected)
for device in untouched:
assert not torch._C._cuda_hasPrimaryContext(device)
if __name__ == "__main__":
sys.exit(pytest.main([__file__]))