After many weeks of tinkering, I do believe I have something that is stable for an overclock.
Mostly asking for advice to see if I have left any stone unturned.
At current my overclock is 5.6 on 2p, 5.5 on 2p, 5.4 on 4p, 5.0e. CPU Core Voltage 1.36 DLVR disabled.
d2d and ngu is 2.8 I have a rather poor/mediocre sample and it refuses to go faster than this.
Reason I’m looking to make sure I’m leaving no stone unturned is I’m about to go to a custom loop for the 290k.
One thing I noticed with mine is that 1t workloads tended to benefit from raising the 8P active clocks, at least under Windows. I found it odd.
You’re about 10% better off than me on Cinebench, but I’m currently just running 200S Boost. I also noticed that there’s quite a spread between Geekbench Windows and Geekbench Linux; Gigabyte Technology Co., Ltd. Z890 AORUS ELITE WIFI7 ICE - Geekbench from the last time I ran it, and I think that was XMP On + stockish power limits + 5.4 on all P (1/2P at 5.5) with Es either left alone or 4.8. Seeing how relatively minimal the headroom really is, I like having relatively stable settings that are about 3 clicks from stone default more than trying to eke everything possible out of it.
Yea, the overwhelming problem with my cpu is it will not run 200s, 2.8 is all it has in the ngu+d2d department. Hence trying to get a 290k especially if they are packing them at the 18a fab.
Reason I’m faster In cinebench is because of my ecore clock being 5.0 and the goofy setup on my Pcores, most likely geek bench isn’t grabbing either if the two cores that are 5.6.
Not that it would matter much without my d2d/ngu being higher.
I’m going to make poor financial decisions and preorder it when it becomes available.
