• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Run RAM Above 1.6v May Fry Your Nehalem

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

AngelfireUk83

Member
Joined
Nov 7, 2004
http://www.theinquirer.net/gb/inquirer/news/2008/10/03/performance-ram-damage-nehalem

WE GOT WORD that Intel is adivising mobo-makers and memory vendors alike about their new X58+Core i7 combo and to abide to a strict 1.65V limit on memory voltages, beyond which there is the risk of burning out the CPU.

The info first tipped up on the XFastest forums where the admin posted several photos of an Asus P6T Deluxe (retail-packed) mobo including one of a big sticker over the DIMM slots motherboard warning that anything over 1.65V will fry the CPU.

The DDR3 JEDEC specification is well within limits of this, however, starting off at 1.5V, but as Intel’s CPUs have made for such great overclockers, memory vendors have ever so happily catered to the enthusiast crowd with faster, higher-powered memory kit. You’ve got plenty of examples, OCZ Reaper PC3-14400 operates at 1.9V, Mushkin’s XP Series uses 1.9-1.95V, Corsair’s Dominator high-end takes you all the way up to 2.1V…

Asus' staff has been busy in meetings so we rang up several memory vendors here in €uroland, and considering it’s a public holiday in Germany, they were more than helpful.

Mushkin’s VP Steffen Eisenstein said that they are re-designing specifically for the X58/Core i7 combo, and that their kit should be out some time next month… in Tri-Channel packs. Other memory vendors across the globe are claiming their kits are still undergoing certification with motherboard vendors so they still don’t know whether they qualify or not.

We’re still waiting for Intel to explain why the memory voltage will burn the CPU, but pressing F9 in Outlook and constantly checking our phones isn’t making things any faster.

It looks like the DDR3 IC and memory vendors will have to go back to the drawing board if they are to create enthusiast-grade performance levels for Nehalem. If you want to get a Core i7 as soon as it hits the market, make sure you have the memory kit to go with it. You’ll have to underclock your kit to more reasonable (power compliant) settings in the meantime. µ

Update: Asus called us back and confirmed the limitation on the RAM voltages. They are currently running a 1.7V kit in their UK office, but beyond that, you're on your own

Pic of Sticker warning Here on ASUS Kit: http://www.xfastest.com/redirect.php?tid=14549&goto=newpost
 
umm they are kidding right? they dont know why it might fry the cpu? do we have to draw them a map? INTEL ADDED THE MEMORY CONTROLLER ON THE CPU!!!!

when you up the VDIMM the cpu is also going to see that voltage, to the memory controller.
 
umm they are kidding right? they dont know why it might fry the cpu? do we have to draw them a map? INTEL ADDED THE MEMORY CONTROLLER ON THE CPU!!!!

when you up the VDIMM the cpu is also going to see that voltage, to the memory controller.

Not suprising considering the source, not the brights crayons in the box.
 
I've had a feeling for a while that these CPU's will suck for OC'ing. They're not aimed at enthusiasts and definitely not aimed at gamers... from the info we've seen so far anyway.
 
this is not bs even though it is comming from the inq, who reported it late I might add. What most people think right now is that due to the integrated memory controller on the i7 overvolting the ram also overvolts portions of the cpu. The ram can take it while the cpu cant so the cpu dies as a result.

Basically this is the same thing that happened to the first runs of athon 64 cpus
 
oh ya this sticker will most likely ship on retail Asus mobos
 

Attachments

  • img8458ye2sw3.jpg
    img8458ye2sw3.jpg
    110.9 KB · Views: 290
Interesting, but I guess you can't say they haven't tried, isn't 1.65v on any part of a CPU asking for trouble usually?
 
I have a feeling this is also part of a ploy to get more sales from there extreme chips. With the limited memory voltage it is going to make the multiplier ever more important.

Interesting, but I guess you can't say they haven't tried, isn't 1.65v on any part of a CPU asking for trouble usually?
65nm chips could take it, but not the best for long life. 45nm chips would be on life support after being run north of 1.65 for extended periods.
 
Last edited:
umm they are kidding right? they dont know why it might fry the cpu? do we have to draw them a map? INTEL ADDED THE MEMORY CONTROLLER ON THE CPU!!!!

when you up the VDIMM the cpu is also going to see that voltage, to the memory controller.

That was my first thought too.

But no that does not make sense, elsewise voltage would have been killing socket 939 years ago...

Anyone remember BH-5?
 
Well, I wasn't holding my breath. Something always has to give. Looks like my 775 is gonna have to last a long while more. I wonder if it's true that the mobo manufacturers will redesign the boards or leave them be?
 
How do you mean? As in somehow altering the circuit to supply more voltage to the memory than CPU?



Somewhat. I mean if too much voltage to the ram will fry the cpu, then someway to downstep the voltage to the chip while raising the voltage to the ram @ a certain point.. Not sure if I'm maiking any sense, but you get the idea.
 
I gotcha ya. That is what I thought you meant. Shouldn't be to difficult, more or less a multiplier of sorts. I am sure they will figure something out. They have to much money to lose if no one buys high end boards because they cannot OC any better than an $80 board.
 
Wait though, how comes in CPU-z of nehalems the CPU voltages is less than 1V? That reading should be correct i guess, but yet DDR3 usually will not run below 1.5V at any clock speed. So maybe some ratio of like 1:1.5 may apply?
 
any idea if they will address this "issue" and fix it in future core i7 chips? I've been waiting to make the jump on the core i7's, so I want to know if it is advisable to wait it out and let the problem be resolved.
 
Somewhat. I mean if too much voltage to the ram will fry the cpu, then someway to downstep the voltage to the chip while raising the voltage to the ram @ a certain point.. Not sure if I'm maiking any sense, but you get the idea.

can't they do two options? One voltage setting for IMC and then one voltage setting for dimm slots?
 
I think a few of you are getting a bit confused. I may not understand it right either, but this is my working theory...

I don't think the 1.6+v kills the CPU core. It does kill the integrated IMC which happens to be on the CPU die. If the IMC dies then the CPU will no longer function.

The CPU core is still getting the vcore we all know well (~1v). The problem is the integrated IMC is not designed to tolerate much above stock voltage which is 1.5v for vdimm. Hopefully, they can find some way to step the voltage down b/n the RAM and the IMC.

If not, I still don't see this being a huge issue as far as CPU OCing. You should still be able to OC the crap out of the CPU. You'll just have to select a lower memory multiplier so that the RAM doesn't run so fast. You will be losing bandwidth, though, and the i7 needs it w/ the lower amount of L2 cache present.
 
Back