2500K vcore does not drop in idle

[Rainwater]

Aida 64 Extreme, prime95, Realbench are three I use. Ibt is also good but runs hot so pay attention to your temps when running it. You didn't upset me man, I just felt maybe it was the right time to change my approach to see if something would click on your end. See if there is a bios update for your motherboard, it could potntially widen your vdroop or llc ranges. Yes if you've been testing with Doom this whole time then you've got some work ahead of you. What I meant as to more info is you have 2 vdroop choices which do you use. Well were you using either option when you said your idle vCore was 1.28v and 1.3v under load? If yes which one? But before you answer these questions run some tests using each of those settings so you can provide results for each of those settings before asking which one you should choose in the end.

Really all it takes is a ten second run with your vdroop settings to see where you're idling and what happens under load with each setting. Literally. As soon as you start a stress you will see what each setting does to your load vCore values.

 

[arkantos91]

First of all, regarding BIOS newer versions: I already have the latest. The next one enables my motheboard to support Ivy Bridge cpus too. But that's not my case.

I did some tests: here are the results.

There seems not to be any difference between low v-droop and auto v-droop in terms of voltage fluctuation.

For example, with 1.260 v-core, both with low v-droop or auto v-droop:

- minimum voltage (idling) read by cpu-z is 1.264.
- maximum voltage (stress testing with IBT) read by cpu-z is 1.304

So, the v-droop setting seems to be irrelevant in my case.

Now, this OC was made adjusting the cpu ratio to 45 (as you can see in the first photo in the previous page) and leaving Intel Turbo Boost disabled.

Another way to do the same is leaving cpu ratio to 33 and enabling Intel Turbo Boost. This allow to change the ratio for each of the four cores to 45.

Is one better than the other?

 

[wingman99]

If both achieve the same 4.5GHz clock with full load, they are both the same method ends to a means.

 

[Rainwater]

Well with these newer chips like the Skylake i7 I have with my Asus motherboard doesn't even have that option. They aren't separated like that. There isn't a multiplier and a turbo like you have and like I had with my 2600k. But when I had my 2600k I disabled turbo altogether and set my multiplier to 45 because I didn't want my cpu to ever clock down. With this generation I have to have my Windows Power Plan set to High Performance to avoid having my Cou clock down at idle. So really it's up to you. If you have C-states enabled it might still clock down either way at idle, I don't remember. I always have them disabled but if you don't have them enabled and you have your multiplier stock and just your turbo overclocked then it will still clock down at idle to whatever the multiplier is set to. So it's really up to you. I don't really think either setting will impact stability too much but I think it just makes it simpler all around just to disable turbo and set your multiplier to where you want to overclock it.

 

[arkantos91]

Ok, thanks.

Now in regard to the lowest stable v-core what I've found is that even a decrease from 1.260 to 1.250 V causes instability.

In fact, during IBT a message popped up saying the system wasn't stable even if nothing strange happened.

But later I got a blue screen when the system was idling lol.

1.255 V gives same result. No problem in IBT (oddly) while after half an hour of DooM it freezed.

So I think 1.260 is the way to go for 4.5 GHz and my chip. Even if it's not really 1.260 because as I said it's always higher then that according to cpu-z.

I'll do some gaming with Hw Info which also reports average values and see what is it for the v-core. I bet it's going to stay on average to 1.280 or something like that. I think that 1.3 V peaks can be seen only with torture tests.

Would be interesting also proving if I can achieve 4.5 GHz with v-core lower than 1.260 playing with other BIOS settings but I don't think so because, in the end, it's really a matter of multiplier and voltage, isn't it?

 

[Rainwater]

Yes, and not all that common for 4.5 Ghz being stable under 1.26v Personally I would try something else other than IBT. Or at least something with it. Not at the same time but use another utility. Stable means no freezes, no blue screens. Gaming, idling, benching, stressing, encoding, or any task that the pc might be doing. No other settings are going to matter if voltage isn't high enough. If you reduce to voltage given to a light bulb, it dims. I say at least 4 hours with Aida 64 or p95 version 26.6 or earlier. Most people would agree. Some will say two. But regardless two hours on one of those utilities is the minimum. Or even Realbench, but I run at least two different utilities for a few hours to see. If you can't go a couple hours with Aida, p95 and maybe Realbench then you will have blue screens and freezes occasionally.

 

[arkantos91]

Gotcha.

I'm going to do some extensive tests and report back.