I am just learning CUDA. Trying to learn curand. I need to be able to call random int of a designated range many times in 1 CUDA thread but no need to store them or return them to host. They can be discarded after use. For example I may need to have a random int 1-25 and then do some things then have another 1. I may need to do this thousands of times in that thread but I don’t need to store those numbers. I only ever need 1 at a time and when it’s used I can toss it and get the next one. Can anyone tell me if this is even possible onboard the device and how to do it please? I am using C++ btw.