Describe the issue:

After updating to 2.4.2, we have noticed that np.linalg.norm returns a different value on the ubuntu-24.04-arm runner.

In particular, the code below runs successfully on the ubuntu-24.04, macos-15, macos-15-intel, and windows-2025 runners. On the ubuntu-24.04-arm runner, it raises an assertion error and says that norm = nan. Other test cases have returned numeric values that differ significantly from those returned on the other runners. Downgrading to 2.4.1 resolves the issue.

The test file can be found here: https://github.com/denialhaag/numpy/blob/8a9f9f55f14e65382aea791f209041652308562a/numpy/tests/test_array.npy

Reproduce the code example:

from pathlib import Path

import numpy as np

test_array_path = Path(__file__).parent / "test_array.npy"
test_array = np.load(test_array_path)
norm = np.linalg.norm(test_array)
print(f"norm = {norm}")
assert np.isclose(norm, 1.0)

Python and NumPy Versions:

Python versions: 3.11, 3.12, 3.13, and 3.14
NumPy version: 2.4.2

Runtime Environment:

No response

How does this issue affect you or how did you find it:

No response