I7 4Ghz OC Stability...

Sounds good. I really want to clock mine, hopefully the store will get stock in this weekend. Mainly waiting on a cpu block

Bed time for me I think mate, it's too late. Catch you both later


^nice, £40 for the upgrade is good. If you'ev had it for a while you might have made it back in electricity bills. Perhaps
Water is awesome though, its a lot more fun than a new bios
 
Found this......

http://www.hardwarecanucks.com/foru...te-ga-ex58-ud5-x58-motherboard-review-14.html

I had an absolute nightmare getting a stable OC without LLC enabled so done some more research and found this article also tested myself and found no spikes with LLC on using OCCT. It seems like with most things its depends on the manufacturers implementation.

2009-08-14-14h26-VCore.png
 
Last edited:
Look at the resolution on the X axis. This is why you see no spikes.


It isn't related to the circuity, its a very direct consequence of how electricity flows. If you cause abrupt changes in voltage, which is needed to counter the drop in voltage when going to load, you get ringing. The sharper the change, the more it rings. It drops out of the mathematics, you can watch it on oscilloscopes. You are not going to see it on a graph which plots every second or so because it happens many times a second.

Do as you like mate, if none of the above is persuasive nothing will be
 
Did you read the review in the link. They put a dmm on it and found no spiking either. I understand the concept now, but if you are using low voltages then spikes are not going to be such a problem. I7 intel specs are 0.8v to 1.375v vid range and a maximum voltage of 1.55v. So with x58 mobo's using better caps and voltage regulation than before the spiking should be only within a 0.2v range under the worst circumstances. I'm still researching this, but pretty much everyone seems to enable LLC with i7 chips but not 45nm quads. Which could be something to do with the way I7's are put together or interact with the voltage etc.
 
Not a chance that the ringing is less than 0.2V. Where do you get the 1.55V maximum from?

A dmm won't detect it. If you've ever used one you'd know they don't exactly update thousands of times a second. The voltage regulation circuitry is improved on x58 boards, but this in no way changes that the voltage accross the processor will drop with load, and that llc forces it not so, so forcing an increase in the voltage fed to the processor which is what causes the ringing which puts me off using it.

There may very well be thousands and thousands of people out there using llc with i7. They can still all be wrong, especially so on these boards where a hell of a lot of them bought it preoverclocked.

The arguement is whether or not the ringing is harmful enough to matter, not whether or not it exists. And the arguement against llc is that just setting higher voltages at idle solves the same problem but doesn't introduce the ringing
 
JonJ678 said:
Not a chance that the ringing is less than 0.2V. Where do you get the 1.55V maximum from?

A dmm won't detect it. If you've ever used one you'd know they don't exactly update thousands of times a second. The voltage regulation circuitry is improved on x58 boards, but this in no way changes that the voltage accross the processor will drop with load, and that llc forces it not so, so forcing an increase in the voltage fed to the processor which is what causes the ringing which puts me off using it.

There may very well be thousands and thousands of people out there using llc with i7. They can still all be wrong, especially so on these boards where a hell of a lot of them bought it preoverclocked.

The arguement is whether or not the ringing is harmful enough to matter, not whether or not it exists. And the arguement against llc is that just setting higher voltages at idle solves the same problem but doesn't introduce the ringing
Click to expand...



Straight from the intel spec sheets......

vid.jpg


transientlimits.jpg


Next door chap had a look at the anand graph and said it used arbitary figures not real world ones and that transients are unpredictable and you get them whenever electrical load changes whether you use vdroop or not. Oh and he says that he does know how quickly a cpu changes it voltage as he works for a semicondutor company. He said the amount of voltage in the chip will limit the transient spike and it is extremely unlikely that it would exceed the limits of the cpu and if you are using voltages which are well below the spec maximum you have nothing to worry about. Its only if you are overvolting or close to the maximum that it would cause a problem.

But anyway he also agreed that if you can get low voltages/temps anyway there is no reason to enable it. I now have a stable 4Ghz OC with 1.275v with LLC off so am pretty happy.
 
Good tables, thank you.

I agree with the fellow, though I don't think he was suggesting that the transients are the same magnitude with or without vdroop. It looks very much like the arguement is that if you aren't putting high voltages through the chip it wont hurt, but if you're using low voltages anyway you don't need it.

So, given this is overclockers and we are likely to put higher than stock voltages through it, doesn't this conclude the argument in favour of not using llc?
 
That's what I concludes as well. I have found that my stable load voltage is actually lower without LLC enabled thus temps are a degree or so lower under load. I ran IBT 16threads 20 passes on vhigh and occt and had no problems. So it's LLC off unless someone can prove it's better on.
 
I'm surprised it's lower than with it enabled, was expecting it to be the same. Perhaps there's something to the idea that stressing the motherboard more hurts stability.

Been good arguing with you, I hope you'll answer some questions when I finally start clocking my one :)
 
We think it's something to do with variation in the load which make it vary between 1.200 and 1.232 under load. We think that it has something to do with c1e.

It's been educational, my wife thinks I'm an obsessive but it's been fun. Any questions on your OC no problem.
 
ive never overclocked before

got my new i7 setup the other week and thought well ill have a tweak

[email protected] voltage 1.23
6gb kingston hyper ram@1750 voltage 1.54

prime 95 stable
passmark burntest pass

2x4890s overclocked 935 core 1050 mem

i got this very easy maybe a bit to easy...makes me think ive done somthing wrong?????lol
 
If its testing stable and not running insanely hot you're probably good. Ram voltage looks a touch low to me, but lower is always better if it stays stable.

Overclocking tends to be easy to start with, then start being difficult at higher speeds, then get so rapidly more difficult that I cut my losses and leave it there. Worrying about things like load line calibration probably counts as worrying too much :)
 
You must log in or register to reply here.
Back
Top Bottom