"Value of threads per SM for entry <x> is out of range" ptxas errors since v1.3

[SFrijters]

#587 made changes to the way CUDA is compiled, but I'm now getting errors of the type

ptxas error   : Value of threads per SM for entry <redacted> is out of range. .minnctapersm will be ignored

Any hints on what needs to be changed in the calling code to make this work again? Reverting that single commit works around the issue but it's of course not a real solution for us.

[SFrijters]

Nevermind, we had the issue in our code all along, but for the previous occa version this error was silently ignored.
So we need to fix an issue on our end.