My new OC'ed Q9450 and voltage

[Witt78]

Built my new system recently on an Asus P5E Deluxe motherboard.

Yesterday I received my heatsink and AS5 from Newegg. (Zalman 9700 LED)

I used 600/800/1000/1500/2000 grit sand paper and lapped my Q9450 to a mirror finish.

The Zalman heatsink already had a perfectly flat mirror finish on it, so I did not lap the heatsink.

I then put a medium overclock on my system. I'm running at 400MHz X 8 FSB, which puts my CPU at 3.2 GHz.

I'm running about 32C idle and 50C on load. (Prime 95 Small FFT's)

Here is my question though.

----------------------------

When I did my overclock I took my Bios VCore setting off of "Auto" and set it to the same voltage that "Auto" was running my CPU at. This is a voltage of 1.25

That seems very high to me... What do you guys think?

- Witt

PS - I am still running tests to verify stability but have found no prime 95 errors yet. and I have not had any time to start lowering the voltage and retest. Time will tell. Your advice will help!

 

[Witt78]

PPS - Keep in mind. Though Intel for some unknown reason will not tell us the actual voltage tolerances of their chips.....

The 45mn architecture will not handle as much voltage as the other chips... This is why I think 1.25 VCore is a bit high.

- Witt

 

[1haplo]

Intels VID Range for the Q9450 is 0.85V – 1.3625V. 1.25 Volts is great. The only thing you must look out for is the VTT termination voltage.

Here is a quote from AnandTech
"Let this be a warning – do not go over 1.4V maximum for 24/7 use! We are certain that the high VTT voltage and extended testing was the cause of death, as we made no other major or obvious changes within the BIOS that could have instigated a failure. Obviously, we tried to boot the processor in a number of other motherboards without success before we decided to post our results up."

Here is the link
http://www.anandtech.com/weblog/showpost.aspx?i=428

A good rule of thumb is only go 10% over max voltage listed on the box. Each CPU is different. Even each batch is different. You may get lucky and get a great batch that will overclock to 3.4 on stock voltage, or you may have to bump it up to 1.3V to get 3.4.

1Haplo

 

[Witt78]

I wish the bios on various Motherboards would use standard terminology for various settings.

It seems that each brand has it's own language for this stuff. This has led me to be a bit confused as I'm somewhat new to this.

What is the difference between VCore and VTT?

I see the following settings in my bios

------------------------------

Vcore - set to 1.25V (Info screen says 1.1 Min and 1.7 max)
CPU PLL Voltage - Set to Auto (Info screen says 1.5V-2.78 Max)
FSB Termination - Set to Auto (Info screen says 1.2 - 1.5 Max)
DRAM Voltage - Set to Auto (Info screen says 1.8 - 2.66 Max)
North Bridge Voltage - Set to Auto - (Info screen says 1.25-1.75 Max)
South Bridge Voltage - Set to Auto - (Info screen says 1.050 - 1.2 Max)

Please describe the difference between VCore and VTT. What does VTT stand for and how do I determine mine?

- Witt

 

[Witt78]

Okay, Just did some hunting for information on this...

VTT is my "Front Side Bus Termination Voltage"

So if I take it off Auto, What should I set it at?

- Witt

 

[Dopekitten]

Leave VTT alone, you don't need to tweak it unless you have stability problems. (Basicly keeping it on auto is fine)

 

[Witt78]

Leave VTT alone, you don't need to tweak it unless you have stability problems. (Basicly keeping it on auto is fine)

So if I leave it on Auto, it won't do something silly like spike the voltage for some silly reason and fry my chip like chicken?

I'm a tad bit nervous about leaving it on Auto... Nervous enough to turn my PC off while I went to work today...

On the other hand, having my PC on for 8 hours, off for 8 hours for a while will help my Arctic Silver 5 to set up...

- Witt

PS - If VTT is left on Auto, what will it do if I..... Say..... Run Prime 95 Small FFT's for 24 hours?

 

[Dopekitten]

The VTT won't spike like that on auto, thats not what auto does. I believe that auto simply sets the VTT at a predetermined voltage, then leaves it at that, it doesn't increase or decrease it under high/low loads. You shouldn't be nervous, i mean, people run the QX9770 at stock speed, 3.2ghz (its just a higher binned Q9450) with VTT on auto because they don't even know how to get into bios. You don't see those chips failing.

So, to answer your question, if you ran prime 95 for all of enternity, VTT wouldn't change (as a setting in bios).

If you still feel nervous, set the VTT to 1.2V. If that compromises stability, set it to 1.3, or tweak it. Just don't go above the max CPU voltage (1.3625) specified by intel. Essentially the VTT sends extra voltage through the CPU when it's needed, so if its above the max specified CPU voltage, then you are slowely killing your CPU (what happened in that Anandtech article).

So, i'd put it at 1.2V for now. If you want to overclock higher, and you have stability issues, you can tweak both that VTT voltage, and the GTL voltages to get certain cores more stable. (Worked for me, was trying to get 3.7ghz stable on my Q9450, and i found a point where increasing the GTL 2/3 voltage by .05 would stop Core 0 from failing in prime, but then Core 1 would start failing, and if i decreased it, the opposite would happen)

 

[Witt78]

Is there any way I can find out what the "Auto" Setting for my VTT is actually setting the voltage too?

CPU-Z will tell me what my VCore Voltage is, but I can't find any utility that will tell me what my VTT is running at?

When I took my VCore setting off "Auto" I just used CPU-Z to determine what my Voltage was and then dialed in that same voltage... That way I could be certain that my CPU voltage was under as little flux as possible.

I would like to do the same with VTT if I can find out what voltage it's currently running at.

- Witt

 

[Dopekitten]

As far as i know, you can't determine the voltage that "auto" sets it at. But i believe that for 45nm processors, the default is 1.1V (That's what my motherboard told me)

 

[Lupiron]

Wouldn't it be best to find out what your starting voltage is supposed to be, before deciding if 1.25 volts is high, or not high?

This is an easy thing to do. Run either Core Temp, or Real Temp, and list what either say your VID is. (Real temp may have two VID fields, a min and max, the max is the listed VID.)

Once you know the VID, you can see if that voltage is high or not. Since the range is 1.1000 to 1.2500 volts, a 1.1000 VID chip would need hardly anything to reach 3.2 Ghz. While a 1.2000 VID may see the slight increase, if your board on Auto was setting it higher because you OC'd it.

Only one way to find out!

Take a peek and see!

I would say I hope you have a 1.1000!! But, since its at 1.25 already, no chance of that!

Waaahhhhhh!!

--Lupi

 

[Dopekitten]

Eh, its not that bad, i've got 3.6ghz on my Q9450 stable with 1.28V or something like that.

Edit: Erm, forgot to say that i had a 1.25 VID.

 

[Witt78]

Mine is 1.25 I didn't know about that VID listing in Real Temp. lol, thanks. I guess 1.25V isn't high.

- Witt

 

[Dopekitten]

1.25 is normally what most people get, seems to be the most common.

 

[Lupiron]

That is mostly true, Dope!

As 1.2500 is the worse one, and the one that will attain the worse over clock at a given voltage, Intel is assured that the mainstream over clockers will only get moderate chips.

That doesn't mean they will over clock badly! It just means that some can over clock better!

Q9450 starts at 2.66 Ghz, so all of them should be able to reach 3.6! 3.8, etc. It's just the lower the VID goes, the higher the clock you should get at a given voltage! Especially since by starting with less voltage, it will need less voltage to be stable at a given speed.

(As in, the lower the VID, the better the Mhz from the amount of extra voltage you are giving your chip! All I can use as an example is the q6600 series. The lowest VID, 1.2000 needs .1 volts more than that while loaded to be stable at 3.6 Ghz. The worse VID, 1.3250, needs 1.45 while loaded to remain stable.) When looking at the voltage, the 1.3250 needed an increase of not only the difference in VID. But an extra .0250 volts for no reason at all, other than the fact that having more voltage makes the voltage less efficient! So you have to again use more!

It's like a neat lil spiral.

--Lupi