Closed
Description
x(asarray) is created as float32 on gpu device without double type support and its printable representation shows with dtype which causes this test to fall.
The example code from WSL:
def test_print_repr(self):
...
x = dpt.asarray([np.nan, np.inf], sycl_queue=q)
> assert repr(x) == "usm_ndarray([nan, inf])"
E AssertionError: assert 'usm_ndarray(...type=float32)' == 'usm_ndarray([nan, inf])'
E - usm_ndarray([nan, inf])
E + usm_ndarray([nan, inf], dtype=float32)
E ? +++++++++++++++
Suggested action:
Add
if x.sycl_device.has_aspect_fp64
to check for double type support