Stride mismatch error with correct contiguity info

[IvanYashchuk]

When running the following code snippet I get a stride mismatch error. The error message says that the contiguity is [t, t] even though it's specified in the FusionDefinition to be [False, True]:

RuntimeError: Stride mismatch with contiguity info.  allocation domain: iS14{2}, iS15{2}: sizes: [2, 2]: strides: [3, 1]; contiguity: t, t; dim: 0; expected stride: 2; actual stride: 3
Exception raised from validateAllocationSizesAndStrides at /home/iyashchuk/dev/Fuser/csrc/tensor_metadata.cpp:262

# torch version: 2.7.0a0+git1335882
# cuda version: 12.8
# nvfuser version: 0.2.25+git6b0b17e
import torch
from nvfuser import FusionDefinition, DataType

def nvfuser_fusion_id0(fd : FusionDefinition) -> None :
    T0 = fd.define_tensor(shape=[2, 2], contiguity=[False, True], dtype=DataType.Float, is_cpu=False, stride_order=[1, 0])
    S1 = fd.define_scalar(0.00000, dtype=DataType.Double)
    T2 = fd.ops.gt(T0, S1)
    S3 = fd.define_scalar(0.00000, dtype=DataType.Double)
    T4 = fd.ops.where(T2, T0, S3)
    T5 = fd.ops.cast(T4, dtype=DataType.Float)
    T6 = fd.ops.set(T5)
    fd.add_output(T6, T0)
    fd.add_output(T6)

with FusionDefinition() as fd:
    nvfuser_fusion_id0(fd)

inputs = [
    torch.randn(5, dtype=torch.float32, device='cuda:0').as_strided((2, 2), (3, 1)),
]
fd.execute(inputs)

[IvanYashchuk]

Is the T0 definition correct for the given input?

[naoyam]

Looks like the contiguity vector gets modified somewhere:

contiguity: t, t;

@jjsjann123 Could you please take a look?