```python import numpy as np, dpctl.tensor as dpt n, m = 10**2, 10**3 a = np.hstack((np.arange(n, dtype=np.int32),) * m) b = np.flipud(a) c = dpt.asarray(b, device='cpu') # crash ``` This regressions was introduced with merge of gh-1198. Filing on behalf of @antonwolfy who discovered and triaged the issue.