Why are we increasing the tile size here? Is it improving the performance of the program in some way, or is it just illustrating that Spatial makes it easier to tile?

I would assume it's probably the latter (Spatial makes it easier to tile) since the red arrows seem to imply that given where it's coming from. I believe we skipped this slide in lecture though / didn't really talk about it. We could probably improve performance of the program by playing around with tile size since it gives a sense of granularity in each operation. We probably don't expect huge variation though since the operation is just + and isn't something where we'd see irregular work levels that might lead to one slow tile becoming a bottleneck. This is maybe why we didn't talk about this as much when looking at ways to speed up the code later in lecture.