Calling kernel based on conditional

New to CUDA and C code, trying to wrap my head around unrolling loops vs. the loops/vectorization in R-Project…Any hints/tips/pitfalls for launching an RNG type kernel based on a conditional applied to array elements other than some of the guidance in the programming guide (P. 79).

Unfortunately can’t go into more detail give the private nature of the code.

TIA for the extraordinary support, Vince