Question how does max boost clock work on epyc 7002 series?

[oortcloud]

Does max boost clock on the epyc 7002 series work like the 7001 series? My understanding is that on the 7001 series, if a numa node has 3 or less active cores then those cores can go into max boost (if thermal and power is good). The 7002 series has fewer numa nodes so I'm guessing it handles max boost clock differently, but I've been unable to find documentation on this.

Does anyone know more details or have some links handy?

 

[Topweasel]

Does max boost clock on the epyc 7002 series work like the 7001 series? My understanding is that on the 7001 series, if a numa node has 3 or less active cores then those cores can go into max boost (if thermal and power is good). The 7002 series has fewer numa nodes so I'm guessing it handles max boost clock differently, but I've been unable to find documentation on this.

Does anyone know more details or have some links handy?

Think that but better. It will be a lot more fluid. With max clocks being limited well below the any reasonable limitation of any single core. Basically the CPU will be constantly flexing the CPU around dependent on the temps/core limits/power. So less uniform on it but basically a core will clock up to its max within usage assuming to many for the package aren't also usage. If all are being used they all will run at the lets say 2.5GHz. There is probably some limitation on core clocks per CCD, but again these are clocking well below Ryzen, so even at max clocks, a single CCD or even several maxed out probably will be well under the power limits of the package. But in the end this is going to be a lot more secret sauce then any promotional material is going to lay out.

 

[oortcloud]

That's pretty interesting and explains a few things. The fluidness of it sounds like it could work well. I was wondering why I couldn't find much info on it.