ASUS P5N-E SLI 650i Thread

[cybershark5886]

Originally Posted by aleiro

It is possible that your voltage was too low, but i would set the +100mv to auto and manually set the vcore voltage. I have found that the higher you go with the +100mv on, say to 3.2 ghz, the vdroop becomes larger and larger so you have to raise the vcore voltage to compensate for the drop.

How does that work? I thought the whole pupose of the +100mV setting being enabled was to combat Vdroop. Are you saying that having it on at higher speeds (3.2GHz - actually that's my speed) makes it Vdroop more? If so, please explain to me how the voltages work a little more, as I would like to learn. After that incident I upped the Vcore by .015 (two settings upwards), but left the +.100mV on.

[aleiro]

I do think you are right, that the +100mv is suppose to reduce vdroop but when i had my e6400 at 3.2 i needed 1.45v to be stable with +100mv enabled but only 1.42v when i manually set the vcore. Its not really scientific, simple observations using orthos and a couple of programs to monitor the voltage in half second increments.

Originally Posted by cybershark5886

How does that work? I thought the whole pupose of the +100mV setting being enabled was to combat Vdroop. Are you saying that having it on at higher speeds (3.2GHz - actually that's my speed) makes it Vdroop more? If so, please explain to me how the voltages work a little more, as I would like to learn. After that incident I upped the Vcore by .015 (two settings upwards), but left the +.100mV on.

[TriBeCa]

hi everyone! I see we have some new faces in here. Sorry I been away for a while

Got me a new PSU and am testing my sig OC again. I'm sure none of your remember, but I was stable for a while at 3.4GHz which then became unstable after a week or so, so I went to my sig OC. That became unstable after another week or so. So I went to 3GHz, and then THAT became unstable after a week or so. I became convinced that my POS Ultra 500W PSU was going...and just put in a coolermaster 750W an hour ago. So wish me luck.

Originally Posted by cybershark5886

How does that work? I thought the whole pupose of the +100mV setting being enabled was to combat Vdroop. Are you saying that having it on at higher speeds (3.2GHz - actually that's my speed) makes it Vdroop more? If so, please explain to me how the voltages work a little more, as I would like to learn. After that incident I upped the Vcore by .015 (two settings upwards), but left the +.100mV on.

The easiest way to think about vdroop is in terms of the voltage formula: V = P/I where P is power and and I is current. Your P is ~constant (it's not actually, but it helps if you think about it this way), and when you load your CPU it increases the current draw. I goes up, P stays the same, and so V necessarily drops. That's vdroop.

Now that you've wrapped your head around that, P *does* actually go up under load, it just doesn't go up enough to keep V constant. So the reason the +100mV helps combat vdroop is because the +100mV from that setting are actually a pair of seperate mosfets from the rest of the CPU mosfets (see the picture on page 3 of the guide in my sig). So because you're reducing the load on the rest of the CPU mosfets and drawing it off those instead, your board is able to provide slightly more P and so V doesn't fall as much.

Originally Posted by Awesome

Not always... I had to reset the CMOS a few times because the voltage doesn't get reset when the mobo doesn't POST.

What is this "Pencil Mod"? Does it have anything to do with all these spam emails I keep getting?

1) I haven't read back...but how are you resetting CMOS? 95% of the time you can reset it just by pressing your reset button 10 times in a row without letting a full power-up cycle occur. Much, much easier than jumper+battery method.

2) The pencil mod is where you take a graphite pencil and 'draw' over two conections on the motherboard. People report very good results in terms of reducing vdroop on this board (e.g. from 0.08V to 0.02V).

[TriBeCa]

Originally Posted by cybershark5886

Well, perhaps I misunderstood your intention but I got the idea from the OP where you said "Set vcore to your CPU's stock setting minus 100mV (so, if your chip is 1.325V at stock, set it to 1.225), set memory voltage to the highest value less than your RAMs stock voltage, set NB voltage to 1.393". That is dropping the voltages. I was wondering why you would want to do that when overclocking?

For the CPU, you're not actually dropping the voltage, you're manually setting it to stock voltage with the +100mV option on. If your chip is 1.325V and you set it to 1.325V with +100mV on your board will supply the chip with 1.325V. You'll most likely increase this as you go on, but because we don't want vcore as low as possible while staying stable, you want to start at stock and only increase vcore as needed.

And with vdimm you're not really reducing voltage.... The board is only capable of supplying voltages at the vdimm settings you see in the BIOS, so you're just picking the highest one that doesn't violate the warranty on your RAM.

Originally Posted by cybershark5886

P.S. Being the supreme newbie that I am to overclocking, may I ask how you gained all your knowledge of how OCing works? Years of practice? A degree in Computer Science or Electrical Engineering? None of the previous esoteric options? I have an Associates in CS and am working toward my Bachelors degree. Any low-level info I can gain on PC hardware would be invaluable to me (and there aren't many good in-depth tutorials online - that I know about anyway). Like the MOSFET cooling deal, whoa... that's a bit exotic to me. You've really got to be a hardcore OCer to get into that. I'd be afraid I'd break something. So if you have any nuggets of knowlegde that you could impart to me (or point me to) I'd be eternally grateful (like an in-depth, low level coverage of OCing for starters).

~Josh

...actually, just 8 months of experience OCing That and the main P5N-E thread here, plus the odd thing read elsewhere on the internet. There's no substitute for practical experience, and talking to knowledgeable people on Overclock.net!

[C2H6O]

If I were to OC to about 3.0 GHz with a e6750, would anyone suggest whether or not I should add new NB and SB heatsinks? Trying to avoid filling up my case with HR-05's.

[TriBeCa]

trying to remember the default multi on the 6750. is it 8x? if so, then you shouldn't need a new NB cooler to hit 3GHz. You should be able to manage 325FSB at 1.39 vNB, though I suppose it's theoretically possible you'll need 1.56. Just reseat the NB with AS5 before you install the board and it should be fine.

[C2H6O]

yea, 8x I think. Good, I have some AS5 handy. Those little snappers on the NB are reattachable, right? THnx

[cybershark5886]

Originally Posted by C2H60

If I were to OC to about 3.0 GHz with a e6750, would anyone suggest whether or not I should add new NB and SB heatsinks? Trying to avoid filling up my case with HR-05's.

Just FYI, I have the e6750 and last night I sucessfully got it up to 3.2 GHz, with no serious heat issues. I doubt you'll need to modify anything.

Originally Posted by TriBeCa

trying to remember the default multi on the 6750. is it 8x? if so, then you shouldn't need a new NB cooler to hit 3GHz. You should be able to manage 325FSB at 1.39 vNB, though I suppose it's theoretically possible you'll need 1.56. Just reseat the NB with AS5 before you install the board and it should be fine.

Yes it is 8x (locked). And my NB is at 1.39V, and the FSB at 400 to accomplish 3.2 GHz. It ran fine. The only issue was my Vcore. I'll bring that up in my next post though...

[cybershark5886]

Originally Posted by Tribeca

The easiest way to think about vdroop is in terms of the voltage formula: V = P/I where P is power and and I is current. Your P is ~constant (it's not actually, but it helps if you think about it this way), and when you load your CPU it increases the current draw. I goes up, P stays the same, and so V necessarily drops. That's vdroop.

Now that you've wrapped your head around that, P *does* actually go up under load, it just doesn't go up enough to keep V constant. So the reason the +100mV helps combat vdroop is because the +100mV from that setting are actually a pair of seperate mosfets from the rest of the CPU mosfets (see the picture on page 3 of the guide in my sig). So because you're reducing the load on the rest of the CPU mosfets and drawing it off those instead, your board is able to provide slightly more P and so V doesn't fall as much.

Thanks for the datailed explanation. I actually picked up a bit of that from your other thread (that's where I got the idea to enable the +.100mV - thus why I asked why you "dropped" the Vcore voltage - but now I understand that you compensated). Now that being said, do you think the problem I had was due to VDroop (see last page). Why would it flash 0.00V (and only in PC Probe - not CPU-Z or SpeedFan)? And as a confirmation question: I suppose it is normal after all to see such a VCore fluctuation from 1.36V to 1.3141V as I did under a full load?

And while I'm at it, now that I've achieved a 400MHz FSB would I gain any benefit from downclocking my RAM to its stock speed (800MHz - I was able to safely run it at 900MHz, its current speed). Would the 1:1 ratio make up for any difference by running 100MHz slower with it? What are the pros and cons of doing that (I'm sure it goes both ways)?

Thanks.

[cybershark5886]

Originally Posted by TriBeCa

For the CPU, you're not actually dropping the voltage, you're manually setting it to stock voltage with the +100mV option on. If your chip is 1.325V and you set it to 1.325V with +100mV on your board will supply the chip with 1.325V. You'll most likely increase this as you go on, but because we don't want vcore as low as possible while staying stable, you want to start at stock and only increase vcore as needed.

And with vdimm you're not really reducing voltage.... The board is only capable of supplying voltages at the vdimm settings you see in the BIOS, so you're just picking the highest one that doesn't violate the warranty on your RAM.

When you said, "If your chip is 1.325V and you set it to 1.325V with +.100mV" did you mean to write the second "1.325V" as 1.225V instead, because I thought that 1.325V +.100mv would make it 1.425V. I assume you mean that you drop the Vcore setting but enable the offset which puts it right back at stock voltage. I tried to get my Dad to explain this to me (he's an electronics technician) and he mentioned something about how the offset would help combat Vdroop in relation to the voltage going negative, something about keeping it from going negative. I'm not sure if he was talking about the actual clock cycle (the mobo runs on AC right, thus having neg Voltage as well?) or if he merely meant negative in the sense of it dropping. Perhaps you could elaborate a little for me. How exactly is the offset different than just keeping the Vcore at stock voltage w/ out the offset?

...actually, just 8 months of experience OCing That and the main P5N-E thread here, plus the odd thing read elsewhere on the internet. There's no substitute for practical experience, and talking to knowledgeable people on Overclock.net!

Well, perhaps I can grow as savy. I'm already learning at a surprising rate here. Good stuff.

~Josh