[gooddog]

I can easily get the i7 to 3.77 with vcore of 1.225.
Multiple of 26
Bus at 145

Runs CPU burn for 10 runs. (I know it is not a long term stability, I will get to that later).

The thing is that if I try to bump up the bus to 150, it takes a lot of voltage to get stable - as high as 1.28. Which other settings should I be experimenting with? Should I be playing with multiple rather than the bus?

Second note. I am not sure what the VID is on this. Running CPUZ it has the v-core fluctuating from 1.06 to 1.18. Any ideas on why that is happening? I assume there is some power management thing going on (speedstep is disabled)

Thanks

 

[mothergoose729]

Where did you get your i7?

EDIT: Unless you know something I don't....

http://www.pcstats.com/NewsView.cfm?NewsID=71157

 

[error10]

He got it at the store, just unlike everybody else.

I'm going to suspect, based entirely on totally unsubstantiated rumors, that you'll want to do more with the multiplier than the bus speed.

I'd imagine the Core i7 can take more voltage than you're giving it, but the first thing you want to do is monitor your temps, and for that you need to know Tjmax. Now the question is, have they updated CoreTemp/RealTemp/etc. to take into account the Tjmax of Core i7, which can vary from chip to chip?

For that you will need to get some tools to read the CPU MSR. The only one I could find from a quick search is RW-Everything. Install it, restart computer, then run it. Once you run it, go to Access menu and choose Command. A small command window will open. Type in "rdmsr 0x1a2" and hit Enter. Get the 3rd and 4th digit of the Low 32 bit(EAX) value, convert it from hex to decimal, and if all the stars aligned right, that should be your Tjmax. (This only works on Core i7!)

I'd have no hesitation about pushing 1.35V through the chip simply based on it being a 45nm process. Anything beyond that, I probably wouldn't. But until we have hard data from Intel in another week or so (hopefully) I probably wouldn't leave it OC'd that far 24x7.

 

[mothergoose729]

I see, my apologies.

I think you can probably give it a little more juice. Your sig says you are waiting on a replacement heatsink, so when it arrives try bumping it up to 1.3v and seeing how far she'll go. I don't know man, you are pretty much in uncharted territory here. More then likely you will find a overclock that works and be teaching all the rest of us when we finally get our own i7's

 

[gooddog]

Quote:

Originally Posted by error10 He got it at the store, just unlike everybody else. I'm going to suspect, based entirely on totally unsubstantiated rumors, that you'll want to do more with the multiplier than the bus speed. I'd imagine the Core i7 can take more voltage than you're giving it, but the first thing you want to do is monitor your temps, and for that you need to know Tjmax. Now the question is, have they updated CoreTemp/RealTemp/etc. to take into account the Tjmax of Core i7, which can vary from chip to chip? I'd have no hesitation about pushing 1.35V through the chip simply based on it being a 45nm process. Anything beyond that, I probably wouldn't. But until we have hard data from Intel in another week or so (hopefully) I probably wouldn't leave it OC'd that far 24x7.

Thanks. Apparently the new realtemp does get the right Tjmax on the i7 but I will check with your suggestion. How far away from Tjmax should you stay anyway?

I will bump up beyond 1.3 - I was just surprised that it was taking bigger and bigger increases for less gain.

+rep

 

[gooddog]

Quote:

Originally Posted by mothergoose729 I see, my apologies. More then likely you will find a overclock that works and be teaching all the rest of us when we finally get our own i7's

uhhh...you really don't want to learn from me

 

[OverclockTheStock]

Meh ima wait till it becomes "main stream" before I jump into it.

 

[error10]

Quote:

Originally Posted by gooddog Thanks. Apparently the new realtemp does get the right Tjmax on the i7 but I will check with your suggestion. How far away from Tjmax should you stay anyway? I will bump up beyond 1.3 - I was just surprised that it was taking bigger and bigger increases for less gain.

If I'm reading this stuff right, it's safe to stay 20 degrees below the Tjmax shown by that MSR. That's the point at which the stock CPU fan should kick into full speed and start cooling off the chip. E.g. if it shows 90° then you stay below 70. IF it shows 100, then you stay below 80.

And, any chip has a point at which you have to pump more and more voltage into it for diminishing returns. That's just the nature of the beast.

 

[gooddog]

Thanks

Yeah, I thought of diminishing returns too, but I didn't expect it at that point. That is why I assume that there are other settings I need to be adjusting.

I got my 9650 to 3.9 without much trouble.

 

[error10]

Oh, apparently the latest version of CoreTemp gets Tjmax right on these chips, but I'll wait to see if you can confirm it.