I’m creating a CUDA program for finding combinations of elements in 9 arrays. My idea is to represent each array with a dimension. Array 1 = Block x, Array 2 = Block y…etc. Between blocks and threads, I’m limited to 6 dimensions. Is there a way emulate more dimensions or should I go about this a whole different way?
One possible solution is to convert the 9 dimension indexes, into 1 linear index and then converting the 1 linear index back into 6 dimensional indexes.
That makes my head hurt, but I think it’s a good plan. I’ll see what I can do.