Skip to content

Print bad default config if compute baseline fails #688

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Sep 26, 2025
Merged

Print bad default config if compute baseline fails #688

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e70327c
Print bad default config if compute baseline fails
yf225 Sep 25, 2025
78cd875
fix test
yf225 Sep 26, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 13 additions & 3 deletions helion/autotuner/base_search.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -115,9 +115,19 @@ def _compute_baseline(self) -> tuple[object, bool, Sequence[object] | None]:
"""
new_args = self._clone_args(self._original_args)
baseline_config = self.config_spec.default_config()
baseline_output = self.kernel.compile_config(
baseline_config, allow_print=False
)(*new_args)
try:
baseline_output = self.kernel.compile_config(
baseline_config, allow_print=False
)(*new_args)
torch.cuda.synchronize()
except Exception as e:
decorator = self.kernel.format_kernel_decorator(
baseline_config, self.settings
)
raise exc.InvalidConfig(
"Default config failed while computing baseline.\n"
f"Default config: {decorator}\n"
) from e
original_args_flat, _ = tree_flatten(self._original_args)
new_args_flat, _ = tree_flatten(new_args)
mutated = False
Expand Down
2 changes: 1 addition & 1 deletion test/test_atomic_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -278,7 +278,7 @@ def test_atomic_add_w_tile_attr(self):

@skipIfRefEager("Error only raises in normal mode")
def test_atomic_add_device_tensor_error(self):
@helion.kernel(static_shapes=True)
@helion.kernel(static_shapes=True, use_default_config=True)
def kernel(x: torch.Tensor) -> torch.Tensor:
for tile in hl.tile(x.size(0), block_size=128):
device_tensor = hl.zeros([tile], dtype=x.dtype)
Expand Down
Loading