That really looks like it's just sorting by memory bandwidth. Particularly with Cascade Lake's results vs. Skylake at the ~10 core number, since there's not really any other major difference between those two.
Being that memory bandwidth constrained, and then not being NUMA aware, is also quite likely why the dual-socket results are a disaster.
That really looks like it's just sorting by memory bandwidth. Particularly with Cascade Lake's results vs. Skylake at the ~10 core number, since there's not really any other major difference between those two.
Being that memory bandwidth constrained, and then not being NUMA aware, is also quite likely why the dual-socket results are a disaster.