CUDA: unspecified launch failure on CI Windows

[szhengac]

Description

CI on windows keeps giving CUDA: unspecified launch failure. It has been retriggered many times.

http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Fwindows-gpu/detail/PR-17400/20/pipeline/

[ptrendx]

@haojin2 The first failing test is test_operator_gpu.test_np_nan_to_num.

[haojin2]

Please first disable then I’ll investigate.

[haojin2]

#17630 still fails on windows-gpu with test_np_nan_to_num disabled, the error starts to occur on the test immediately after test_np_nan_to_num, meanwhile there were other PRs passing the windows-gpu build, as a result, I suspect the problem is not with the test_np_nan_to_num itself.

[ChaiBapchya]

Another one for unrelated PR #17487 - http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Fwindows-gpu/detail/PR-17487/24/pipeline

[leezu]

This issue appears fixed by the toolchain updgrade #17962