This makes sense since the memory controller is now on the processor and not a separate chip. Although doesn't AMD's AM2+ allow a separate voltage plane for the cores and the memory controller (I know that AM3 socket will). Will overclocked memory damage the processing cores or just the memory controller (obivously either answer results in a useless CPU). I wonder if this will affect the general overclocking ability of the processor itself. I also wonder if Enthusiast Memory is going to go by the wayside. I think the only reason people needed overclocked memory with the old/current Intels was to match the increased FSB speeds (there are plenty of reviews here that show memory speed is rarely a bottleneck with modern DDR2/DDR3 dual-channel setups). I don't know any specs on Intel's method, but AMD's method of tying memory speed to processor speed usually means overclocking the memory is pointless. God, how many returns is Newegg going to have to process because of all the idiots out there who insist on cranking up their memory voltages to 1.9V+ regardless of how fast they plan to overclock it?

With the smaller and smaller transistors and chips there is more issue with crosstalk between components. Maybe the materials at this reduced size can not stand up to the heat and bond more easily?

"Let us not forgot that manufacturers have always recommended against over voltage and clocking in some form or another – but it does not mean you cannot do it. We are pretty certain that this shouldn’t cause a big issue in the long run. The only real difference is that novices will now be more prone to burning out their processors instead of their boards / memory first. Depending on which model of processor is being used, it could be a more costly mistake."

I agree, we have always been warned - but it never really means anything beyond that. I bet the only reason they are putting this warning out is to educated people that the CPU is at greater risk of damage with the controller being onboard now.

I bet we will soon be hearing about a lot of careless Core i7 system builders screwing up their newly built systems at the forums.We will soon hear "I am never building another Intel system".
It might be fun for some laughs for a while.

Personally, I find this information about as useful as a milk bucket under a bull currently. As mentioned in the article, John Q. Public hasn't even seen the first full review yet, let alone being able to purchase it. With that said, one wonders, will the tri-channel configuration bandwidth performance make enough difference to not really have to o'clock the DIMMs and rely strictly on the CPU bus speed for performance increases?

Hahaha, so true.

looks like the enthusiast ram industry is getting a kick in the pants to get in gear with better stuff

You comment was about as useful as the article, in your own words, of course.

I wonder if we'll see boards or modules with seperate power plugs soon? So the board would only supply 1,5v thru the cpu and addittional current thru a different means, or a converter between cpu and memory scaling hte 1,5v from the cpu up to 2v on the memory or something ...

neiroatopelcc

“According to Intel CPU spec..." Well, according to Intel spec, you shouldn't overclock due to possible CPU failure. Also, According to Intel spec you shouldn't mess with vDroop at all, but some people bypass this feature and most don't have issues while some do. Intel's just playing it safe.

?

The difference I suppose is, that you can buy retail memory modules that won't nessecarily work, and eventually kill the cpu, because those module specs require more power. That is, you can run your mainboard and cpu at stock speeds and still kill it by setting the vendor specified settings for the memory.

how much more performance do we get from an overclocked (and overvoltaged) DDR3 Tri-channel? 1% ? haha! This Overclocking is Ridiculous... if you put little more money, you'll get a better system, why overclock to damage the system and shorten it's lifetime? why overclock in the first place? there's a saying, Who pays more, get more soup!

Because OverClocking is life.

There's a saying "You, sir, are obviously full of fail".

There may be a warning, but I think me,IF I ever buy a Core i7 pc,I'd prefer not to try out how far I can up the voltage.
If intel says 1,65V, it means 1,7 most likely will still go,and I bet some might even get it to work at 1,75V!

However, seeing how costly the system costs,memory and all included,I am certainly not the first to try out overclocking on a machine like this!
Let's hope Core I7 will support overclocking without boosting the Voltage too much.

As for me,I'd probably focus on underclocking such a system to get the max on powersavings!

That aside from the OC'ing aspect, which is the #1 reason why I purchase any pricey hardware... Will we really "need" to OC with a 4-8core Nehalem running a psuedo-benched 30-50% faster (in some instances) than the fastest Yorks? I think OC'ing will always be around, it's just that the technological speed range is shrinking based on the current materials and layout used. A new and completely radical material, die design and layout is what I'm more excited about... again we're getting the metered out products (which is good for their business), but drags and slows overall technological innovations. I'm pretty sure for all the conspiracy folks: The SKUNK Works boys that fly out to Groom-Lake (govt still refuses to ack it's existence) have computer technology 10-15yrs beyond what we're seeing now... Too bad obsession with gross profits is keeping us in the slow lane.

That and the simple fact of "air cooling" using a metal sink and heatpipes seems rather low-tech Vs compact refrigerant units or TEC (Peltiers)... Cooling is def way behind what it could be: We've all see how much higher OC'ing can yield with sub-ambient cooling. Then again Air-cooling is cheap and effective Vs bulky liquid based units.

Sounds like intel is routing VTTDDR directly through the memory controller. Either way, sounds like poor engineering on thier part.

The reason I'm calling them out for poor engineering is; lets say, for example on a current system such as a modern 775 or AM2+ motherboard, the VTTDDR is fed via a linear regulator from the 5v plane. Also, lets project a scenario of a memory stick burning up and shorting out on one of these 775 or AM2+ motherboard; 5v then becomes short with ground, trips the breaker in your PSU, system shuts down with minimal damage (with a slight chance of frying the VTTDDR regulator on the motherboard if your PSU is a low-grade model). Lets project that same scenario on the new 1366 motherboard... VTTDDR, of which is routed through your CPU becomes short with ground. This will not trip the PSU breaker as there is still resistance in the line (your CPU). So your CPU is left to dissapate not only over 25 amps from the vcore, but a full 30 amps on top of that from the 5v line. I SMELL BACON!

It's not just about speed. It's a quest. The quest for more speed and more computing power one can squeeze out of a commercially available product. It's like why people build drag racers, put turbocharger on their 4-cylinder engines, and install NOx injection system on their cars. The engine might explode if done wrong, but it's heck lot of fun.

Can it run Crysis?

the diagram of the nehalem architecture shows the IMC on the new Core i7 chip and thus changing vdimm to over 1.65 will damage the chip itself. this was not a problem in prior architectures where the memory controller was on the chipset and separated from the chip

check this post for the diagram
http://www.augustinefou.com/2008/1 [...] vdimm.html

*Tsssst* Negative Houston... we're having a problem with Max-FSAA... *Tssssst* lol

What the hell is NOx injection? Last I heard ricers were still using the brand name "NOS" instead of Nitrous Oxide, or N2O..

I sort of forgot which x is used to boost engine power. Thanks for reminding me that x==0.5.

@ Grims : NOS is the company name, for one of the companies providing n2o upgrade kits for engines. It is NOT the only company doing so. So while NOx may be wrong, so is your indication that NOS is the right term.

well i have an intel dx38bt mother board and ocz pc10666 the default volt was 1.54 i had to change it to 1.7 because the system was unstable restarting whenever.. now it is stable and no issues