What I meant by anomaly was for a program to run slower when the CPU is clocked faster. This apparently happens with the 8GB model on certain kinds of parallel workloads but not with the 4GB model.I hope that this makes at least some clarification about measured "anomalies".
Benchmarks are fun. But achieving the optimal performance it's a bit of an art beside lots of knowledge, tests, dead ends, heat, noise and balance.
But as always, YMMV.
While the solution is, of course, then don't do that. What I find interesting is to know what's going on with the hardware and why does it not affect the 4GB model.
Statistics: Posted by ejolson — Thu Dec 14, 2023 5:18 am
