That last point is an interesting one I hadn't thought of until now. We've focusing on getting as much parallelism and utilization as possible without thinking how memory performance comes into play. Our algorithms in this lecture provide the most benefit for very large datasets if the memory system accompanying them can keep up.
That last point is an interesting one I hadn't thought of until now. We've focusing on getting as much parallelism and utilization as possible without thinking how memory performance comes into play. Our algorithms in this lecture provide the most benefit for very large datasets if the memory system accompanying them can keep up.