[SYCL][CUDA] Comping SYCL code with subgroups failing

[pvelesko]

ptxas fatal : Unresolved extern function '_Z17__spirv_GroupFAddjjd

Are there plans to add sub-group support for CUDA? If possible?
If not, compiler should throw an error during compilation.

[bader]

Related PR: #2587

[bader]

@pvelesko, #2587 is merged. Does it address your request or anything else is required?

[bader]

Closing this issue. @pvelesko, feel free to re-open or leave a comment if the issue is not resolved on your side.

NOTE: #2623 adds shuffle support.