This is a bit of a long shot, but when I run my simulation multiple times with the same input, the output diverges after some variable amount of time. I expect it’s a problem on my end (e.g. unitialized memory being used somewhere), but has anyone experienced similar problems?
EDIT: It seems the problem was re-using shared memory. I forgot to add a __syncthreads() between reading the values from shared memory and then writing new values to it.