The exclusive means for row pointer. I think segmented exclusive scan means we dont include the value at index of row pointer. For example, the vertical line the segment between out 0 and out 1 is at index 2, so the exclusive here means it sum up from 0 to < 2.
Is any of this even worth still spending time on considering many packages such as numpy, etc. handily out perform using such an implementation since the cost of setting up the sparse matrices isn't worth it? Thoughts?