Skip to content

[NVVM] Guardrail test_program for version mismatches between driver and PTX compiled binary #1204

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
leofang merged 3 commits into from
Oct 31, 2025
+8 −2
Merged

[NVVM] Guardrail test_program for version mismatches between driver and PTX compiled binary #1204

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4403b0a
add exception for mismatching PTX and supporting CUDA Driver versions
abhilash1910 Oct 31, 2025
05cac12
add exception for mismatching PTX and supporting CUDA Driver versions
abhilash1910 Oct 31, 2025
dd78838
[pre-commit.ci] auto code formatting
pre-commit-ci[bot] Oct 31, 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
10 changes: 8 additions & 2 deletions cuda_core/tests/test_program.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,13 +2,14 @@
#
# SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE

import re
import warnings

import pytest
from cuda.core.experimental import _linker
from cuda.core.experimental._module import Kernel, ObjectCode
from cuda.core.experimental._program import Program, ProgramOptions
from cuda.core.experimental._utils.cuda_utils import driver, handle_return
from cuda.core.experimental._utils.cuda_utils import CUDAError, driver, handle_return

cuda_driver_version = handle_return(driver.cuDriverGetVersion())
is_culink_backend = _linker._decide_nvjitlink_or_driver()
Expand Down Expand Up @@ -317,7 +318,12 @@ def test_nvvm_program_creation_compilation(nvvm_ir):
assert program.backend == "NVVM"
assert program.handle is not None
obj = program.compile("ptx")
ker = obj.get_kernel("simple") # noqa: F841
try:
ker = obj.get_kernel("simple") # noqa: F841
except CUDAError as e:
if re.search(r"CUDA_UNSUPPORTED_PTX_VERSION", str(e)):
pytest.xfail("PTX version not supported by current CUDA Driver")
raise

@leofang leofang Oct 31, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Q: Shouldn't this be

Suggested change
raise
else:
raise

@rwgk rwgk Oct 31, 2025

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The else isn't needed: pytest.xfail exits the test right at the call site.

But the else doesn't do harm either.

I tend to fully rely on the pytest behavior, i.e. not add redundant flow control, for more compact code.

program.close()


Expand Down
Loading