Debug why unary polygamma tests are skipped in upstream CI but pass when run.

[dllehr-amd]

🐛 Describe the bug

Need to find out why these passing tests are skipped in upstream CI for test_unary_ufuncs.py

test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float16 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float17 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float18 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float19 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float20 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float21 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float22 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float23 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float24 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float25 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float26 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float27 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float28 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float29 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float30 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float31 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float32 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float33 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float34 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float35 (main.TestUnaryUfuncsCUDA) test_unary_ufuncs test_reference_numerics_hard_polygamma_polygamma_n_2_cuda_float36 (main.TestUnaryUfuncsCUDA)

Versions

ROCM 4.5.2