i7 920 overclocking thread

[BlackHoleSun]

Has anyone gotten to 4 GHz on a 920 stable here on the forum.

I saw the review at XBit and thats what put me over the edge. I'm ordering my 920 Ci7 shortly, but it won't go in until Christmas break for me.

[amitashr]

Just got an MSI Eclipse+i7-920...Now i cant Overclock because the voltages are set to auto and greyed out in the bios cell menu and i cant change them
HELP!!

[TFL Replica]

Originally Posted by amitashr

Just got an MSI Eclipse+i7-920...Now i cant Overclock because the voltages are set to auto and greyed out in the bios cell menu and i cant change them
HELP!!

There's a bios option that unlocks the voltage protection so you can OC. Consult the manual.

[eflyguy]

Yeah, I'm happy to run stock under the Intel cooler (V8 coolermaster is back at store, it was a piece of junk!) and get all my toys installed for now. As soon as I get the parts for my CPU WC loop, I can start messing with OC again - by then, hopefully, there'll be a little more info around here on what works and what doesn't..
..a

[HardwarzFan]

Originally Posted by BlackHoleSun

Has anyone gotten to 4 GHz on a 920 stable here on the forum.

Thats what I wanna know too

[003]

Originally Posted by {core2duo}werd

it changed with nehalem intel said the new max is 1.55

No.... no, it didn't change. It has been 1.55v all along, most likely even higher. Are you aware that Nehalem uses the exact same 45nm high-K manufacturing process as Penryn and Yorkfield?

Therefore, Nehalem, Yorkfield and Penryn all share the same maximum safe voltage limit, which is likely even upwards of 1.6v.

Furthermore, since maximum safe voltage and temperature share an inverse relationship, and since Nehalem runs significantly hotter than Penryn and Yorkfield, it can be concluded that the maximum safe voltage for Penryn and Yorkfield is even high than that of Nehalem.

Although, it has been quite amusing all this time watching everybody get their panties in a bunch whenever the idea of *gasp* running a 45nm CPU above 1.3625v is even mentioned in passing.

EDIT:
And you guys are aware that the reason Intel specified such a low max voltage for Yorkfield and Penryn was simply to make people shy away from overclocking and be suckered into buying the very expensive extreme edition CPUs.

EDIT2:
Proof to back up Core i7 using the same manufacturing process as Core 2 here.

[003]

Oh, and {core2duo}werd, I just learned some useful information myself. There is no problem with setting your DRAM voltage above 1.65v, you just have to make sure that the difference between your DRAM voltage and your Uncore voltage (typically called "QPI/DRAM Voltage" in the BIOS) is no greater than 0.5v.

So you would want to set your Vqpi to 1.4v, and then your DRAM voltage will be good up to 1.9v (1.9 - 1.4 = 0.5).

EDIT:
Source, thanks to bowman and Cliff Burton on XS.

[{core2duo}werd]

Originally Posted by 003

No.... no, it didn't change. It has been 1.55v all along, most likely even higher. Are you aware that Nehalem uses the exact same 45nm high-K manufacturing process as Penryn and Yorkfield?

Therefore, Nehalem, Yorkfield and Penryn all share the same maximum safe 24/7 voltage limit, which is likely even upwards of 1.6v.

Furthermore, since maximum safe voltage and temperature share an inverse relationship, and since Nehalem runs significantly hotter than Penryn and Yorkfield, it can be concluded that the maximum safe voltage for Penryn and Yorkfield is even high than that of Nehalem.

Although, it has been quite amusing all this time watching everybody get their panties in a bunch whenever the idea of *gasp* running a 45nm CPU above 1.3625v is even mentioned in passing.

EDIT:
And you guys are aware that the reason Intel specified such a low max voltage for Yorkfield and Penryn was simply to make people shy away from overclocking and be suckered into buying the very expensive extreme edition CPUs.

dude none of that makes any sense... first of all they said the max of yorkfield and penryn was 1.45v and nehalems max is 1.55v... second the 1.3625v max 24/7 was set by overclockers after seeing chips degrade and die. Who says nehalem is made with the same manufactoring process? and even so who says they didn't tweak SOMETHING to make it so the chip could take more voltage? i don't know all that much about making cpus, and i'm guessing you don't either, so when the people who do say that the max voltage is higher... i believe them because how else am i supposed to know without killing some cpus?

I agree that the 1.3625v thing is probably a little low, but i have personally seen a chip degrade after running it for 7 months at 1.425v.

saying that intel made the voltage limits to keep people away from overclocking is rediculus seeing that you can get plenty of 45nm chips to 4Ghz+ on air...

but seriously even if you don't agree with a statement (I wasn't trying to say 1.55 max for 24/7) doesn't mean you should mock it.

Originally Posted by 003

Oh, and {core2duo}werd, I just learned some useful information myself. There is no problem with setting your DRAM voltage above 1.65v, you just have to make sure that the difference between your DRAM voltage and your Uncore voltage (typically called "QPI/DRAM Voltage" in the BIOS) is no greater than 0.5v.

So you would want to set your Vqpi to 1.4v, and then your DRAM voltage will be good up to 1.9v (1.9 - 1.4 = 0.5).

EDIT:
Source, thanks to bowman and Cliff Burton on XS.

thanks for that, but i still don't know... we will see after a few months.

[zlojack]

003, I want to believe you because it would make overclocking even more fun.

But you're making statements without backing them up with links or proof.

[003]

Originally Posted by {core2duo}werd

but seriously even if you don't agree with a statement (I wasn't trying to say 1.55 max for 24/7) doesn't mean you should mock it.

Not mocking your statement personally, just the whole paranoia uproar about degradation and whatnot in general. 1.3625v was not agreed on by the NCOO (National Commission of Overclockers), it was dictated by intel as the max safe VID, and then fantastically overblown by the overclocking community.

Some of the very early 45nm chips had problems and did die or "degrade", but this is no longer the case. Current "degradation" (which is now all but nonexistent) is more likely caused by the use of crappier PWM and MOSFET caps on the cheaper motherboards, and cheaper power supplies that put out more ripple and dirtier power.

My Q9450 has been running now at 1.425v 24/7 with LLC enabled for the past 7 months, and I have used up to 1.6v on benching sessions. It has sat like a rock since day one. Not the slightest bit of degradation.

My post makes perfect sense once you realize that Nehalem does use the exact same high-K 45nm manufacturing process as Penryn and Yorkfield (as part of their Tick Tock release schedule). Max safe voltage remains more or less the same for a given manufacturing process.

And to clarify more, Nehalem runs hotter than Penryn and Yorkfield, and the maximum safe voltage decreases as temperature increases (the inverse relationship), and vice versa. So, since Penryn and Yorkfield run cooler than Nehalem, they can be run at a higher safe voltage.