No theory, actual throughput tests
I did the same.
Swings and roundabouts.
Gains on the speed.
Lose on the timings.
No theory, actual throughput tests
I've yet to see lower frequencies beat even a lax 3000-3200. But you've obviously done a lot of testing![]()
Quite a bit higher? Than 1.016v?
Doubtful, I would make sure you are testing stability correctly by loading up all 16 threads with HCI. I doubt you'll be able to get SA stable with less than .9v at 3000 and above.
Sarcastic? All you've been doing is saying how crap the platform is because you aren't having much joy with overclocking, so honestly I'd aim your suggestions closer to home, you've been saying a lot of things which don't seem to make much sense of late lol
If your clocking higher than 2666 on mems switch to 125 BCLK strap as this is always more stable.
All the GSKILL I have will do 3200+ but I have higher binned kits than yours. I would leave all settings on Auto in memory and try pushing the speed with 1.35vdimm. Then tune down the timings once you stabilize higher speed.
I think the 3000mhz Dom Plats are sadly Samsung IC which means about C13 at 3000 will be the best they can do.
Is it safe to run higher memory voltage levels pack, say 1.5v for 24/7?
1.5v is the absolute max Intel allow for XMP certification at the moment. There are a lot of grey areas with memory overclocking. Life time warranty anyway, but you're looking for someone to tell you what you want to hear, and you just won't at the moment fella