NullCosmos

Ah! May I suggest we implement `get_configuration_warning` to validate the correct configuration?

I'm having the same issue. Worked fine for about 24 hours. Now I always get cudaError 30 exactly like above. Related to #135