Core i7 and DRAM voltage above 1.65v??!

[003]

Hey guys. I have 3x 2GB sticks of CSX Diablo DDR3 2000 memory, with the only catch being that it needs 1.9v to do its rated speed. Now I have been reading reports that you will kill your Core i7 CPU if you use a RAM voltage above 1.65v -- WTF?!

What truth does that statement hold? Am I really going to be limited to 1.65v??? I have seen conflicting reports that that it is not true as well, but I am very confused right now, and nervous, as I paid a lot of money for this RAM.

My motherboard is an Asus P6T Deluxe. Am I going to have a problem above 1.65v?

[bowman]

No. Just make sure that the difference in voltage between uncore domain (i.e. L3, IMC) and the memory is no more than 0.5V, and you should be fine. This comes from Francois Piednoel (drwho on internet forums, including this one) himself.

While we were fearful about the memory overclockability of Core i7 CPUs we've found this not to be the case - we've hit 1,600MHz without even breaking a sweat on pre-production BIOSes and companies are already launching triple channel memory kits that are hitting 2,000MHz at a record low 1.65V. You can push the memory voltage, but Intel performance guru François Piednoel was keen to stress that the CPU voltage must be kept within a 0.5V potential difference to memory.

With CPU stock voltages at sub 1.2V this only allows an upper safe maximum of ~1.65-1.7V, however increase the CPU voltage to 1.4 or 1.5 (or more) with some extreme cooling (remember these CPUs are rated at a 130W TDP at ~1.2V) and you give yourself more breathing room to 1.9-2.0V on the memory.

That's not to say that exceeding the 0.5V rule will instantly kill a CPU, but it's like the effect of excessive VTT or PLL voltages to current Penryn CPUs - don't be surprised if they die without warning much sooner than you expect. By that rationale, AMD's Athlon 64s suffered the same fate with DDR2, but the effect was often exacerbated because the initial voltage difference was that much greater to begin with and greatly overvolting the memory to 2.3+V made it that much worse.

http://www.bit-tech.net/hardware/200...tecture-dive/3

[003]

When you say "uncore domain", that is the same as the CPU voltage? So if I set the CPU to 1.4v (on water), I will be good for 1.9v on the DRAM? If so that is a huge relief. The whole point of the DDR3 2000 was so I can try and break 4GHz without memory limiting me

[systemviper]

That is good to know, all i keep hearing is the 1.65v limit, So i ordered 3x1G of Kingston hyperX 2000

I wonder how many people by the new ram and are selling their perfectly good ram?

[bowman]

When you say "uncore domain", that is the same as the CPU voltage? So if I set the CPU to 1.4v (on water), I will be good for 1.9v on the DRAM? If so that is a huge relief. The whole point of the DDR3 2000 was so I can try and break 4GHz without memory limiting me

No, uncore domain is the section of the CPU where the L3 and the integrated memory controller resides. Each section of the CPU has its own voltage setting. You'll see it plainly enough when you get your board up and running and you get to key through the BIOS options.

And yeah, the uncore can run at a higher voltage than the CPU cores - at least I've seen people running it higher. The article I've linked to simply says 'CPU' but I can swear the intention was the uncore as that's the IMC. Oh well.

[003]

No, uncore domain is the section of the CPU where the L3 and the integrated memory controller resides. Each section of the CPU has its own voltage setting. You'll see it plainly enough when you get your board up and running and you get to key through the BIOS options.

And yeah, the uncore can run at a higher voltage than the CPU cores - at least I've seen people running it higher. The article I've linked to simply says 'CPU' but I can swear the intention was the uncore as that's the IMC. Oh well.

Hmm, well I'm confused again In your quote of Francois, it reads in my mind like he is saying that the difference between the CPU voltage and the DRAM voltage should not be more than .5v, with no mention of "uncore" voltage. Is there a separate setting for uncore voltage then? Do you know what it is called on the P6T Deluxe?

[Cliff Burton]

It is referred into most of BIOS as "QPI/DRAM Voltage", and it usually goes from 1.20v to 1.90v. It has nothing to do with the CPU voltage (core), noted as Vcore.

It is too early to make conclusions about max safe 24/7 QPI voltages before degradation, but it seems Vqpi=1.40-1.50 it's considered still safe with the right cooling.

If your sticks need Vdram=1.90v to reach their rated speed, then your Vqpi would be set to Vqpi=1.40v-1.45v. As a mate said, the difference between Vqpi and Vdram should not exceed 0.5v, if so your Bloomfield will suffer.

To make you a quick idea:

Vqpi=1.20v (min), Vdram=1.65-1.70v --> uncore safe
Vqpi=1.50v (max safe), Vdram=1.95v-2.00v --> uncore safe
Vqpi=1.20v (min), Vdram=1.90v --> uncore very stressed

[003]

Ah, I get it now. Thanks so much!

[003]

Wait, I'm confused again. I found this picture:


It has the QPI and IOH as separate voltages, and they are both different. Which is the one that I want to set at 1.4v? Or do both get set at 1.4v? Or are neither of those the ones I am looking for?

[dinos22]

QPI Voltage is the correct one

[003]

Thanks... now one final clarification. I know I asked this earlier, but due to the level of contradiction between (the same!) sources, I feel I need to ask again.

In this bit-tech article, it quite clearly says that the difference between the dram voltage and cpu voltage must not exceed 0.5v.

Intel performance guru François Piednoel was keen to stress that the CPU voltage must be kept within a 0.5V potential difference to memory.

But in this bit-tech article, it quite clearly states that the difference between the dram and uncore voltage must not exceed 0.5v.

keep the CPU uncore voltage within 0.5V difference of the DRAM voltage and there's no problem.

So how is it we know that it really is the qpi (uncore) voltage to increase and not the CPU voltage?

[dinos22]

i emailed richard about this
he confirmed the uncore/vdimm difference so that should be it

[003]

I'm confused again. I just looked over saaya's i7 overclocking guide and read the section on memory voltage. He makes no mention of QPI or uncore voltage, or anything about memory voltage being within 0.5v of anything. Instead, he says that it has to be 1.5x the VTT?? how can it only be one constant voltage like that? I didn't know i7 even had a VTT, or thought it was called something different. Now I am really confused again, because basically what saaya said on the topic does not seem to relate in the slightest to what you guys have said

[[XC]Atomicpineapple]

QPI/DRAM Voltage on the P6T is VTT voltage. Gigabyte are calling it QPI/VTT.

[003]

Ok, but what is that saaya is talking about saying the DRAM should be 1.5x the VTT/QPI, and stuff about 0.7v offset?? It's totally different from the explanation in this thread which is very simple: Just keep vDIMM - vQPI less than or equal to 0.5v.

That is easy to comprehend. But saaya is talking about keeping voltages in proportion in specific ratios with various multipliers and offsets... wtf is he talking about? Basically, do I just follow the above emboldened statement and I am fine or is there any merit to the much more complex and confusing (for me) explanation by saaya?

[003]

Ok, but what is that saaya is talking about saying the DRAM should be 1.5x the VTT/QPI, and stuff about 0.7v offset?? It's totally different from the explanation in this thread which is very simple: Just keep vDIMM - vQPI less than or equal to 0.5v.

That is easy to comprehend. But saaya is talking about keeping voltages in proportion in specific ratios with various multipliers and offsets... wtf is he talking about? Basically, do I just follow the above emboldened statement and I am fine or is there any merit to the much more complex and confusing (for me) explanation by saaya?

Can anybody clarify these discrepancies for me? I am really nervous I will end up using the wrong settings and damaging something unless I know 100%, and I'm sure it would be very helpful information for lots of other i7 owners as well

[003]

Ok, I received a well clarified PM from saaya. There are currently three theories floating around on what constitutes a safe vDIMM voltage on the Core i7 / x58 platform.

1. vDIMM - vQPI ≤ 0.5v

2. vDIMM - vQPI ≤ 0.7v

3. vDIMM ≤ 1.5(vQPI)

Of these, the most stringent is theory number one, which also happens to be the most "official" statement by intel on the matter. As such, I will be going with theory number one.

I hope this post can also help others looking for clarification on this matter.

[Demo]

When using air cooling like the TRUE120ex on the i7 how much would be the safe 24/7 QPI voltage? Intel rates it at 1.375v right ?

[r1rhyder]

There are newer threads than this one to ask in.