When can Voltage Increases be Destructive?

[scorchingice]

I am trying to Overclock my CPU way beyond the normal. Closer to the 4Ghz Mark.

I have a E8850 Conroe 1333 FSB processor with an EVGA 780i SLI motherboard.

In the BIOS I have the following Voltage settings available to modify. When modifying the voltages, there are 3 colors. White (Normal) - Green (Increased above normal but still reasonably safe) - & Red (Use extreme caution using this setting)

Here are the voltages available to me to set.

CPU V-core: Currently set at 1.5v (Default 1.3v Max Safe 1.5v Max 1.8v)

FSB: Currently set at 1.4v (Default 1.2v Max Safe 1.4 Max 1.5v)

Memory: Currently set at 2.2v (Default 1.8v Max Safe 2.2v Max 2.5v)

PCI-E (SPP): Currently set at 1.45v (Default 1.35v Max Safe 1.45 Max 1.55v)

nForce (MCP): Currently set at 1.65v (Default 1.5v Max Safe 1.65v Max 1.75)

HyperTransport (SPP<->MCP): Currently set at 1.40v (Default 1.2v Max Safe 1.4v Max 1.55v)

My Questions are about raising those voltages into the red settings to try to get more over clock. Is this ok to do? Is there any guide to how much is too much when setting all of these temps ?

I want to go into the red on some but not sure if doing this will cause something to fry. How much is too much? When to stop ? When to try and see without too much risk blowing my Motherboard or CPU?

Any help would be great.

Thanks

 

[Evilsizer]

if the board is actually putting 1.5v into that cpu now, that is already to much voltage. this is after all a 45nm cpu, at that voltage right now, its like putting 1.6-1.7v into a 65nm cpu. which isnt good, the most voltage i would put into a 45nm lga775 cpu is 1.45-1.5v under water. the other issue is to get the most out of the cpu you need to change voltages which you dont have control over in the 780i board.

the main thing with NV based boards is for high fsb's you need to up the northbridge voltage. given that they run hot from the start doesnt help matters much.


The only time voltage increase becomes dangerous, is when you use a low PWM based board to oc. which in that can will kill the board and the board might take the cpu with it. in you case you run the risk of killing the cpu since your already pumping alot of voltage into the cpu as it is.

 

[jason4207]

Evil, that's an E6850; 65nm.

1.5v should be fine with good cooling. Those chips are pretty bullet-proof.

Scorchingice, just take your time. You might want to lower the CPU multi to 6 and lower your RAM to 1:1, so you can focus on the FSB speed. Start with all your voltage settings at default (except give your RAM it's spec'd voltage and give the CPU 1.35-1.4v for now), and start raising the FSB 10-20MHz at a time. Test w/ P95 Blend for 5-10mins and once you start failing or can't boot try testing each voltage setting 1 at a time. If a setting helps you then use it, and if it doesn't put it back. Only give it voltage where it needs it. That board might have FSB holes, so you might also want to just try a slightly higher FSB speed if you hit a problem. Keep pushing up the FSB while repeating those systematic steps, and try to get to 450. Make sure to test longer once you get dialed in. I like for the test to run a full cycle which you can tell by seeing the fft size reset; usually overnight.

Once the FSB is good you can up the CPU multi 1 tick at a time. Use P95 small-fft to test, and use vcore to stabilize.

You can also try to increase RAM speed following the same procedure; use Blend.

You might also want to google "e6850 780i" to see others results and BIOS settings.

 

[Evilsizer]

doh, i didnt click the link, thought it was some new E8000 intel release to clear out old stock... thought maybe his sig was out of date or something..

 

[doz]

Im a firm believer in higher voltage than max rated spec is OK as long as temps are kept inline. Temps are more important at higher voltages than the actual voltage. I ran 1.53v on my E8400 for 6 months, and its still running today (over a year later) at 1.38v @ 4ghz in my brothers rig (and it ran 6 months @ 1.38v before running 1.53v).

 

[jason4207]

On the 65nm CPUs it is generally accepted that 1.65v is OK on water, and 1.55v is OK for good air for 24/7 usage assuming temps are kept in check.

If you get much over 70-75*C you're going to have a hard time keeping it stable at 3.8GHz+ anyway, and you will also have diminishing returns on speed past 1.4-1.45v in my experience.

I got my brother-in-law's E6750 stable at 3.8GHz @ 1.4v.

 

[scorchingice]

On the 65nm CPUs it is generally accepted that 1.65v is OK on water, and 1.55v is OK for good air for 24/7 usage assuming temps are kept in check.

If you get much over 70-75*C you're going to have a hard time keeping it stable at 3.8GHz+ anyway, and you will also have diminishing returns on speed past 1.4-1.45v in my experience.

I got my brother-in-law's E6750 stable at 3.8GHz @ 1.4v.

My cooling setup is great. As you can see by my signature I am using : Cooling System = 5 120mm Case Fans + LAPPED ZALMAN CNPS9700 LED + ARCTIC COOLING ACCEL X8800 Fluid Dynamic VGA Cooler

2 of the case fans are Scythe Ultra Kaze (The 2 intake), the other 3 are coolermaster stock (The 3 exhaust).

My temps stay at the following: (This is at 3.6GHz *FSB = 400MHz* *Memory 450MHz*

*At Idle* (CPU temp = 34c) (Core #1 = 27c) (Core #2 = 28c) (SPP = 56c) (MCP = 54c)

*Under Full Load* (CPU temp = 47c) (Core #1 = 48c) (Core #2 = 49c) (SPP = 57c) (MCP = 55c)


It will run fine at 3.8GHz but occasionally after a while encoding, Windows 7 64bit will blue screen. I do not believe this to be a heat problem as I never even hit 50c on the CPU. I believe this is due to not enough voltage to do the work. I want to try raising the voltages, but I am just unsure as to which are safe to raise past the (Green Safe Zone) Bios settings.

Also will having too low of wattage Power Supply Unit cause an issue? My PSU is a (750 Watt) COOLER MASTER Real Power Pro RS-750-ACAA-A1. I have only 1 8800GTX Video Card.

I will eventually Try jason4207's suggestions and run my ram at 1:1 and raise FSB to 450MHz and slowly raise voltages and the multiplier till I can get it stable at a higher clock. Maybe my proccessor will just not run faster than 3.6GHz safely.

Anyhow thanks for the responses so far. Here we go trying to raise the speed

 

[jason4207]

You can go past the 'green safety zone' safely if you follow my instructions above.

Your PSU should be fine.

 

[EarthDog]

The braintrust has spoken.. you are in good hands..

But note I wouldnt call that zalman great. Its made for quiet, not top notch cooling performance. With that said, keep the temps in line and all will be well.

It just sounds like your system is not stable...

 

[Evilsizer]

as long as your load temps are low 60's, you will be fine. though like E.D. said the zalman isnt a great cooler i can think of a few others that are a touch cheaper nd better coolers.

 

[scorchingice]

But note I wouldnt call that zalman great. Its made for quiet, not top notch cooling performance. With that said, keep the temps in line and all will be well.

Please note to the Zalman Comments, This Zalman was Ordered from Newegg on 12/28/2007. Back then, this WAS the highest rated Air cooling for my CPU. Sure now there are better, But this Zalman still gets the job done.

Anyhow, as to your other comment:

It just sounds like your system is not stable...

What can I do to make it more stable ? Is it that my hardware just cannot run higher than the 3.6GHz norm overclock?

Right now I am stable at 450MHz FSB with an 8x Multiplier. that is still 3.6GHz but it seems to run faster than 400MHz FSB with an 9x Multiplier (and slightly hotter)

My CPU is locked so I cannot go above 9x multiplier. The settings are 6x - 9x capability.

I would just like someone to elaborate on the statement "My system is not stable" I understand it is not stable when it blue screens, but what will help make it stable so it can go above the norm?

I would like to get 3.8GHz - 4.0GHz out of this thing if I can.

Thanks (BTW I run Everest Ultimate Edition) to test my settings and Prime95 64bit to put my system under load.

Thanks again.

Brian

 

[EarthDog]

Well, 'stable' here is such an ambiguous term...LOL! You should be able to run Prime95 Small FFT and blend for about 8 hours each with no errors or dropped cores. If it errors out, you have to likely raise some voltages. Which ones those are, no idea. You would have to list all your major ones (CPUv/VTT/PLL/vDIMM/NB) Check out the guide from my signature.

 

[Evilsizer]

well if your testing the cpu stick to small TFF/TTF's after you get the cpu stable. run the blend test to make sure the whole thing is stable. the two tests do completely different things. blend tests a bit of cpu/ram/chipset, if its unstable in blend it would be a bit tricky to find the problem. if you did small first which focuses on cpu and you dont get errors then you know its not the cpu but chipset or ram.

 

[EarthDog]

Absolutely! ^^

 

[scorchingice]

Thanks for all the replies. After this current batch of encoding is done (About 2 hours left) I will try some of the things you said. I will try for 3.8GHz and use the testing method you stated to find out the culprit causing the blue screens. I know 3.6GHz is easily obtainable. It is getting beyond there that causes issues.

Can I go beyond 450MHz FSB since my multiplier is locked at 9x max and I do not have 1/2 marks in my bios I got 6,7,8,+9 settings (All even numbers no halfs)

I will also post all the Bios Voltage Settings along with other settings. I will try to be as thorough as I can.

Again I really appreciate all the replies and helpfull suggestions I am getting.

 

[scorchingice]

Here are my current test results:

3.6GHz / 1600MHz FSB / 400MHz Memory

3.6 GHz / 1800MHz FSB / 400MHz Memory

3.712GHz / 1856MHz FSB / 870MHz Memory

No matter how high I set the voltage, the 3.8GHz+ settings seem to blue screen on the FPU tests in Everest.

 

[jason4207]

Which voltage? Are you trying all the settings?

 

[scorchingice]

Which voltage? Are you trying all the settings?

My Bios Settings available to me are as follows :

CPU V-core: Set at 1.5v setting shows 1.45 current value even though set at 1.5 (Max Safe 1.5v Max 1.8v)

FSB: Set at 1.4v (Default 1.2v Max Safe 1.4 Max 1.5v)

Memory: Set at 2.2v (Default 1.8v Max Safe 2.2v Max 2.5v)

PCI-E (SPP): Set at 1.45v (Default 1.35v Max Safe 1.45 Max 1.55v)

nForce (MCP): Set at 1.65v (Default 1.5v Max Safe 1.65v Max 1.75)

HyperTransport (SPP<->MCP): Set at 1.40v (Default 1.2v Max Safe 1.4v Max 1.55v)

To try to get to 3.8GHz STABLE I have tried the following

Raising the CPU core all the way to 1.7 (NEver tried 1.8v was not going to risk it)
Raising the FSB voltage to 1.5

Raising the Memory to 2.25v

Raising the SPP to 1.5v

Raising the SPP<->MCP to 1.5v

Raising the MCP to 1.7v


It runs fine, but when I stress it to the max, I get a blue screen of death.

I have tried using a 9x multiplier with a lower FSB and an 8x multiplier with a higher FSB. Both blue screen though the lower FSB maintains lower temps.

 

[Evilsizer]

It runs fine, but when I stress it to the max, I get a blue screen of death.

it doesnt run fine if you can not get it prime95 or orthos stable.

sounds like the chipset needs more voltage for higher fsb's. forget temps, at this point i would find which clock is stable while being the fastest.

if you want higher, im afriad you need a new mobo...

 

[freeagent]

Each cpu is different. For example, I have no qualms about useing 1.5-1.6v on my e8600es, but I would have never used those kinds of volts on my last q6600, 1.4v was about the max for it with air. tho with my first q6600 I had no problems giveing it 1.6-1.65v heheh.