does it hurt to leave everything but vcore on auto

Associate
Joined
31 Dec 2008
Posts
1,400
hi

with my overclock i have a reather modest 3.2 running stable at 1.25 vcore and my ram at 1.64, with my i7. But everything else is et to auto. my question is will this be harming the processor over time by raising other voltages too high. im not sure what the qpi or ioh voltage should be.

thanks
 
For a modest overclock like 3.2ghz, i was running ioh/ich at 1.1v, qpi at 1.21v, ram 1.64v and vcore at 1.18, with theese settings i was stable upto 3.6ghz.
 
Last edited:
Its fine to leave everything on auto, just most people like to tweak it all themselves, as u can usually get away with far less doing it urself then what auto will do.

That being said its more then fine to leave stuff on auto if your not comfortable messing with them.
 
i leave most of my setting on auto the only ones i change is the v core and the memory volts
 
^ as bifday2k says you should be fine on auto, temps may be slightly higher as auto tends to give more voltage than is required, i remember on my previous s775 asus p5q deluxe board, if certain settings were left on auto or lowest manual setting the board would overvolt theese settings to quite high levels, thankfully i havent heard any reports of this on the x58 range.
 
great thanks, looks like i can get the vcore down a bit to 1.22 rather than 1.25 then, ill try that. also what program can i use to check what the ioh/ich and qpi voltage is actually running at? - thanks oh and whats this about your qpi needing to be within .5 of your dram volts? and 1 more silly question - my gtx 280 went bust litterally a week after i did this overclock on my cpu, it wasnt the overclock that was unstable because it was still bust when i reset it. am i paranoid in thinking the overclock on the cpu may have had something to do with the graphics card going bust?
 
Last edited:
At 3.2ghz, it won't hurt. Even vcore on auto won't matter on such a conservative clock. If you start aiming anywhere north of 3.6 I'd seriously reconsider this approach, not only are motherboards historically fond of overvolting, they're not always very good at getting the balance right. Sometimes voltages have to be decreased to find stability.

It's just about possible that this is no longer true, as with setter I haven't seen any particularly damning reports on x58 boards overvolting. I consider it unlikely that since P45 this has been perfected.

qpi does need to be within 0.5V of the ram voltage, so at 1.65V keep qpi at least 1.15V. A greater difference is considered to degrade the chip. Errors inherent in measurement and voltage control give the more conservative estimate of keeping them within 0.45V of each other. The main effect is to justify going over 1.65V on the ram. As most people end up with a qpi around 1.3V for 4ghz, and this allows 1.75V on the ram, and few people are willing to go that far over the intel guideline, it's something of a non-issue.

The 280gtx dying sucks, but it was probably unrelated to the processor overclock. A tenuous link is the power supply, if it is inappropriate for the rest of the system then the additional load from the processor may have been enough to drive it out of spec with a knock on effect on the card. This is unlikely. What power supply are you using? In what way did the card die?
 
For a modest overclock like 3.2ghz, i was running ioh/ich at 1.2v, qpi at 1.21v, ram 1.64v and vcore at 1.18, with theese settings i was stable upto 3.6ghz.


You use 1.2v for IOH and ICH??

For 4.2GHz I have IOH at 1.140v, ICH is something I don't even bother adjusting.. well that's a lie, I take it off auto and set it to 1.100v..

From what I have found with these i7 processors, the only 2 voltage settings that seem to affect the temps, are Vcore and Vtt, none of the others seem to have any impact on temperatures at all..

I doubt it was your CPU that killed your GTX280 mate, the 0.5v rule is to prevent damage to the processors IMC iirc..

So long as you keep the distance between Vtt and VDimm at .5v, you can use more than 1.65v for VDimm, I haven't used anything higher than 1.68v for my RAM, but there are plenty of people out there pushing much higher voltage than 1.65v through their DIMM's, and there doesn't seem to be any ill effects from doing this.
 
You use 1.2v for IOH and ICH??

From what I have found with these i7 processors, the only 2 voltage settings that seem to affect the temps, are Vcore and Vtt, none of the others seem to have any impact on temperatures at all..

Not quite accurate. Core and vtt will affect your cpu temp but don't forget the rest of the mobo. When I was playing around at 4.5 I needed the whiny little fan on the 'Northbridge' for it to play ball.
 
Typo on my part davy:o, should be 1.1 at 3.6ghz, i only upped it to 1.2 when i was testing at 3.8ghz.
 
At 3.2ghz, it won't hurt. Even vcore on auto won't matter on such a conservative clock. If you start aiming anywhere north of 3.6 I'd seriously reconsider this approach, not only are motherboards historically fond of overvolting, they're not always very good at getting the balance right. Sometimes voltages have to be decreased to find stability.

It's just about possible that this is no longer true, as with setter I haven't seen any particularly damning reports on x58 boards overvolting. I consider it unlikely that since P45 this has been perfected.

qpi does need to be within 0.5V of the ram voltage, so at 1.65V keep qpi at least 1.15V. A greater difference is considered to degrade the chip. Errors inherent in measurement and voltage control give the more conservative estimate of keeping them within 0.45V of each other. The main effect is to justify going over 1.65V on the ram. As most people end up with a qpi around 1.3V for 4ghz, and this allows 1.75V on the ram, and few people are willing to go that far over the intel guideline, it's something of a non-issue.

The 280gtx dying sucks, but it was probably unrelated to the processor overclock. A tenuous link is the power supply, if it is inappropriate for the rest of the system then the additional load from the processor may have been enough to drive it out of spec with a knock on effect on the card. This is unlikely. What power supply are you using? In what way did the card die?

ok thanks - so with the dram set to 1.64 (as the mobo wont let me set to 1.65) will the qpi on auto set itself to 1.15 or slightly above? was wondering if there was a programe to monitor these voltages as i cant find 1?

my psu is a coolermaster m850. (850 watts) the card died by crashing with a pink screen when it was cool - heating up would fix the issue - seemed to be quite a common issue with 2 other people with 280s on this forum having the same issue at the same time so yeah its probably just paranoia! right?
 
Last edited:
No, the qpi cannot be expected to intelligently set itself within the 0.45 range. You'll have to do that yourself, but it's not so bad, the Asus bios is very friendly.

That psu is fine, no worries.

The 280gtx probably needs to be reballed. You can mimic this using an oven if you're fairly brave, or get a computer repair shop to do it for you. Assuming its out of warranty/hasn't been thrown away anyway.
 
Card sounds like dry joints / duff caps.....

if it's out of warranty and you looking at a dead loss (no pun intended) first check the caps look good (no leaking / bulging etc) then cook your card in the oven to reflow the solder. Surprising how many times this seems to work.

Alternativley get thee soldering iron out and have a go, just leave the surface mount stuff well alone unless your very good.

thats was quick Jon....
 
No, the qpi cannot be expected to intelligently set itself within the 0.45 range. You'll have to do that yourself, but it's not so bad, the Asus bios is very friendly.

.

ok thanks for that, but would you only need to set the qpi yourself if you want to overvolt your ram for whatever reason? as most people dont bother touching that voltage for overclocks less than 3.5ghz ish.

my gtx 280 has since been replaced by the 5870 in my sig just dont want the same thing happening again heheh
 
Back
Top Bottom