Core i7 920 rev D0 stepping 5 :)

[DerekT]

Who mentioned or spoke of guaranteed stability? What are you arguing about?

I answered this in my post, I'll state it once more just for you:

Prime95 tests integers of the form (2^n) -1 for primeness
Linpack solves systems of linearly independent equations. As to "stability",
there is no such thing as a program that guarantees stability.
And that's no bullsh**.

[emertX]

Have a look at this:

Better cooling equals better voltages!!!

[bluehaze]

And what if you have passed prime for at least 12 hours, and your computer crashes? What then? Could still be your OC.
You can pass 100 hours of linpack and see a BSOD the next day when you boot up and go into email.

In my experience over the past 10 years or so of OC'ing If i've got my system stable for over 12 hours of prime then my OC is stable. If it does crash it has always been drivers, buggy games/apps that need patches etc...never because of my OC.

[roller11]

[QUOTE=DerekT;3752864]Who mentioned or spoke of guaranteed stability? [QUOTE]

ReverendMaynard mentioned a "stability tool", presumably a tool that determines "stability".

[Splave]

GIGO = Accounting Term people learn in College, used to describe many data related analysis.

To each their own? Now your just missing the point entirely and being ignorant to the fact that you are grossly incorrect. Linx is NOT a Valid program to Test 24/7 settings for your i7, you will fail Prime95 in UNDER 10 MINUTES every time.

I just have to ignore him now and wish him the best of luck. You really can't spend that much time or energy on someone who can't grasp on to the big picture. In the working world, these tend to be people that are let go in slow times or never promoted. They latch on to the small words, the small ideas, instead of grasping the big picture: Data Validity.

Nothing in my signature is unique and can't be replaced or upgraded to something new. My Sig is nothing more then a snapshot in time for others who may have questions about building something similar. Feel free to hit PM or Email me with any questions you may have about my current build.

The ignorance in this thread is sickening.

You have someone that knows nothing about i7s trying to belittle people and force his ways of stability and another guy who reads the dictionary for fun a bit too much.

I guarauntee that if Im even 20x linx stable with a decent chunk of mem that I can run prime for over 5 hours blended atlteast.

Mr definition of stability over here might want to load up his bios, choose load fail safe defaults and put his tinfoil hat on. If you want true stability you should have gone with a workstation board and no overclocking.



Back to D0

@emertX, very nice did you put it under water or something?
Was suprised to see that it only took 1.475v on phase to hit 4.6 bench stable on this ordinary c1 where as 4.4 took 1.55 ish on water
http://valid.canardpc.com/show_oc.php?id=515465

[crashmax]

Max　Bclk
BIOS　233*19

Impressive!!

Can you post some voltages please? QPI/Vtt, VCore, IOH, ICH?

Thanks!

[haya_gann]

Impressive!!

Can you post some voltages please? QPI/Vtt, VCore, IOH, ICH?

Thanks!

BIOS
PCIE Frequency　108
QPI/Vtt　1.42V
VCore　1.325V
IOH 1.2V
ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

[drnip]

Nice.

[gunbo13]

The ignorance in this thread is sickening.

You have someone that knows nothing about i7s trying to belittle people and force his ways of stability and another guy who reads the dictionary for fun a bit too much.

I guarauntee that if Im even 20x linx stable with a decent chunk of mem that I can run prime for over 5 hours blended atlteast.

Mr definition of stability over here might want to load up his bios, choose load fail safe defaults and put his tinfoil hat on. If you want true stability you should have gone with a workstation board and no overclocking.

I will have to agree to the above.

I see the argument for "stability" pop up constantly. Why? Usually people use it to back out of silly arguments like using prime over linx. It is an arbitrary argument. You can't toss a dictionary result of the word "stability," slap some makeup on it, and have it hit the town. It is a circular argument that nets nothing.

So lets look at it in a way that is rare in this world, objectively. You are overclocking. Overclocking itself is a singular variable of change. I'm not going to bring in the scientific method here so lets latch on to the concept of changing variables. Encapsulated in overclocking is a plethora of variables that only further add to the analysis for stability.

What does all this mean? Unless you hit default settings and go back to square one your idea of stability will never be concrete. Is the earth's environment stable? Hey we are breathing and living right so it is stable. We apply an exponential amount of chemicals into this environment, similar to stressing the environment, and we are still living right? While that may be stable to some, it certainly may not be stable to others.

Now that the the arbitrary argument is defined here comes the objective viewpoint. We have a set of tools that are mathmatically based; there is no stress component type software. But do you have any other tools all of you stability wiseguys?

You use what you have and test accordingly. You have the next best thing so use it. Linx for i7 has been proven to be more stressful on the new architecture. Use it. Simple, done, nothing more to say. To argue otherwise is just going to have you running in a circle with your head down until you finally stick it up your own ***.

[ReverendMaynard]

BIOS
PCIE Frequency　108
QPI/Vtt　1.42V
VCore　1.325V
IOH 1.2V
ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

Hey man, is that much QPI/VTT necessary?

[overcrash]

Hey man, is that much QPI/VTT necessary?

Ofc its necessary, otherwise bsod and no booting

[roller11]

Unless you hit default settings and go back to square one your idea of stability will never be concrete.

There is nothing "concrete" about the so-called "default" setting of 2.67 Ghz in the case of a i7 920. 2.67 has no technical significance, it's just another setting on the continuum of settings from zero to infinity and in fact, the CPU has no default because Bclk is generated by the motherboard, not the CPU. The word "stability" as it's commonly used is actually a way of saying that the CPU is within it's AC limits. The AC limit is the clock value beyond which the flip flops can no longer change state fast enough to propagate the correct value (0 or 1)
to the next stage in worst case situations. Notice that there is mention of software here because any software would run properly if the CPU is within it's AC limits. Any choice of "stability tool" is strictly arbitrary because no software can create every worst case situation (aka 'race condition') inside the CPU. So there is no validity in claiming "stability" just because your CPU has run 100 passes of linpack because another program might cause your computer to crash.

[ben805]

I will have to agree to the above.

I see the argument for "stability" pop up constantly. Why? Usually people use it to back out of silly arguments like using prime over linx. It is an arbitrary argument. You can't toss a dictionary result of the word "stability," slap some makeup on it, and have it hit the town. It is a circular argument that nets nothing.

So lets look at it in a way that is rare in this world, objectively. You are overclocking. Overclocking itself is a singular variable of change. I'm not going to bring in the scientific method here so lets latch on to the concept of changing variables. Encapsulated in overclocking is a plethora of variables that only further add to the analysis for stability.

What does all this mean? Unless you hit default settings and go back to square one your idea of stability will never be concrete. Is the earth's environment stable? Hey we are breathing and living right so it is stable. We apply an exponential amount of chemicals into this environment, similar to stressing the environment, and we are still living right? While that may be stable to some, it certainly may not be stable to others.

Now that the the arbitrary argument is defined here comes the objective viewpoint. We have a set of tools that are mathmatically based; there is no stress component type software. But do you have any other tools all of you stability wiseguys?

You use what you have and test accordingly. You have the next best thing so use it. Linx for i7 has been proven to be more stressful on the new architecture. Use it. Simple, done, nothing more to say. To argue otherwise is just going to have you running in a circle with your head down until you finally stick it up your own ***.

+1

couldn't have said it better myself....nuff said anyway.

[Hornet331]

ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

very interesting.

[ether.real]

Suggesting that Prime overnight is one thing.....suggesting that 30 minutes of Blend is better than 20x Max Linx....thats just crazy talk.

[crashmax]

BIOS
PCIE Frequency　108
QPI/Vtt　1.42V
VCore　1.325V
IOH 1.2V
ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

You must have a very good CPU! I have no chance to do any higher then 220BCLK

[jcniest5]

You must have a very good CPU! I have no chance to do any higher then 220BCLK

Not even on your D0 CPU? That's odd...I thought D0 will break the 220 bclk barrier...

[fallwind]

This is where I'm at right now with my W3520. I thought I had it stable earlier at 1.296v but got an error after 6 hours of P95! So I bumped it a notch and hopefully it's still running in the morning.

The good news is my DS4 is one rock solid board. All other voltages are at stock(even QPI) and there is absolutely no vdroop!

[emertX]

BIOS
PCIE Frequency　108
QPI/Vtt　1.42V
VCore　1.325V
IOH 1.2V
ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

Did anyone test this with another Board?

[ben805]

This is where I'm at right now with my W3520. I thought I had it stable earlier at 1.296v but got an error after 6 hours of P95! So I bumped it a notch and hopefully it's still running in the morning.

The good news is my DS4 is one rock solid board. All other voltages are at stock(even QPI) and there is absolutely no vdroop!

Nice load temperature, is that on water?

[TinTin]

D0 920 4.6G, 1.411V SPI 32M passed

[fallwind]

Nice load temperature, is that on water?

Yes, see sig for details.

[PytonOrm]

BIOS
PCIE Frequency　108
QPI/Vtt　1.42V
VCore　1.325V
IOH 1.2V
ICH 1.15V

To raise BCLK, PCIE Frequency becomes important.

Max　BCLK　228/PCIE Frequency 103
Max　BCLK　230/PCIE Frequency 105
Max　BCLK　233/PCIE Frequency 108

Windows did Boot by these settings.

good find mate i try last night up to a 116mhz and booting into winodows without problem at all. give me more Bclk to

[roller11]

TinTin,

Nice clocks! Are those load temps accurate? Only 51C under load?
Anyhow, looks like you've got a lot of headroom left, could validate well above 4.7G with more
voltage. Great job!

[Falkentyne]

There is nothing "concrete" about the so-called "default" setting of 2.67 Ghz in the case of a i7 920. 2.67 has no technical significance, it's just another setting on the continuum of settings from zero to infinity and in fact, the CPU has no default because Bclk is generated by the motherboard, not the CPU. The word "stability" as it's commonly used is actually a way of saying that the CPU is within it's AC limits. The AC limit is the clock value beyond which the flip flops can no longer change state fast enough to propagate the correct value (0 or 1)
to the next stage in worst case situations. Notice that there is mention of software here because any software would run properly if the CPU is within it's AC limits. Any choice of "stability tool" is strictly arbitrary because no software can create every worst case situation (aka 'race condition') inside the CPU. So there is no validity in claiming "stability" just because your CPU has run 100 passes of linpack because another program might cause your computer to crash.

^^
Not trying to throw gasoline onto a fire, but I have to agree here.
I can pass 50 passes of linpack on a P4, at a certain speed, but run a 2D directdraw game (like Maplestory) or a 3D game (both which cause far less stress on the CPU) and the computer will bluescreen in less than 30 minutes, if the vcore is too low.

Haven't seen this happen on core2 architecture.
Taking RAM or chipset stability issues out of the picture, you never know what can cause a flip-flop to fail if you are on the edge of stability. Sometimes the strangest combination of transactions can cause a failure.

20 minutes of linpack plus a battery of other tests is a good way to get a well rounded result, though.