12900KS and 14900KS. HT on, e-cores off.
With HT, I never have to worry or wonder if an application is using the HT threads instead of the P-cores. Acording to all my tests over the years, it doesn't work like that (unlike e-cores).
GTR 1 (AKA GTR: FIA GT Racing Game) is a truly single-threaded game. It cannot gain from / use more than a single thread and is also massively CPU-limited. If you assign the game to only use 1 thread but choose a HT core instead of a physical core, you don't get any performance drop - nothing changes. So the physical core is still doing the work.
I also did many tests with later games like GT Legends, GTR 2, and RFactor 1. These games use evolved versions of the game-engine in GTR 1 but, unlike GTR 1, they make use of up to 2 threads instead of 1. Again, if you assign the game to only 2 threads but choose HT cores instead of physical cores, there is no performance drop - nothing changes. You'd expect a humongous drop in performance since HT's cores only have around 33% the performance of their physical core brothers but, no.
On the other hand, E-cores can create all sorts of performance-affecting issues. I decided to give E-cores another chance when I switched from a 12900KS to a newly setup 14900KS system. I open Cinebench R15 and get ridiculously low scores. I checked it's CPU affinity using Windows task manager and saw it was assigned to all E-cores but no P-cores, what the!!!??? So I assigned all the P-cores to it as well, re-did the test, and it was STILL only using the E-cores. After that, I was like "screw this" and went back to permanently disabling the E-cores.
Some people think Windows/software CPU affinity settings are the "final say" when they are absolutely not. It's clearly much deeper than that.
