Previous | Next --- Slide 9 of 46
Back to Lecture Thumbnails
Tengjiao

It seems that in the graph, higher performance is achieved on the right hand side. This means we still need a huge core for executing the sequential-oriented code and reserve a tiny part for parallel code. Does that mean in today's world, most of the workload is still sequential?

EggyLv999

We can see in graph d especially that the best speedups are actually achieved with a balance of small cores and large cores. Additionally, this graph has nothing to do with workloads in real life. It's simply a simulation of "If I plug in different sequential vs parallel ratios, what kind of performance do I get?" Real workloads can be anything, and there are definitely mostly sequential and mostly parallel solutions out there.