IntelPython / dpnp / 12710043140 / 1

72%

master: 72%

push

github

Update `TestSpacing` and  `TestTensordot` (#2251)

This PR suggests using `assert_allclose` instead of `assert_equal` with
a resolution-based tolerance in `TestSpacing::test_zeros` to fix test
failures on CUDA caused by excessive precision for float32.

And updates `TestTensordot::test_axes` and `TestTensordot::test_linalg`
to increase a factor parameter in `assert_dtype_allclose` for scaling
tolerance

4559 of 11468 branches covered (39.75%)

Branch coverage included in aggregate %.

16948 of 21579 relevant lines covered (78.54%)

19827.08 hits per line