TestMathBitsXPU issues

[daisyden]

🐛 Describe the bug

• accuracy issue of test_neg_view_nn_functional_rrelu_xpu_float64 - PASSED on
  torch 2.6.0a0+git64ccebd
  torch-xpu-ops 3b245e2

• RuntimeError: value cannot be converted to type float without overflow
  TestMathBitsXPU has 2 cases with RuntimeError: value cannot be converted to type float without overflow - v2.6
  "test_conj_view_addbmm_xpu_complex64",
  "test_neg_conj_view_addbmm_xpu_complex128", - duplicated with [Evaluated] MKLDNN implementation of addbmm does not support complex and real data type #436

Versions

latest version

[daisyden]

test_ops.py

RuntimeError: value cannot be converted to type float without overflow

    "test_conj_view_addbmm_xpu_complex64",
    "test_neg_conj_view_addbmm_xpu_complex128",

[daisyden]

no this case any more based on latest pytorch, test_neg_view_nn_functional_rrelu_xpu_float64, on db80b98ec460ca5b2fd84c1dfb6426925f64c8cc

[daisyden]

For test_conj_view_addbmm_xpu_complex64, we got "RuntimeError: value cannot be converted to type float without overflow", it could because the addbmm implementation of mkldnn has explicit cast to float like "alpha.to()", when the alpha and beta are complex, it causes the error. {'beta': (1+2j), 'alpha': (2+3j)}. @ZhiweiYan-96 please help follow up.