Overclocking help E6600/P5W-DH/Mushkin 8000

[RMSe17]

Hi, I got my new computer parts to work finally, after having to RMA my seasonic 600...

Anyway, E6600, P5W-DH BIOS 1305, 2Gb Mushkin Redline 8000, Big Typhoon in the CPU fan plug. (1288RPM Q-control set to performance). AS5 under all heatsinks.
Currently I have set
CPU FSB 400MHz @ VCORE 1.5375V
DRAM 800MHz (1:1) 5-5-5-11-6 @ VDIMM 2.30V
VMCH Auto
FSB Terminal Auto
ICH Auto

Does most tests fine, but Prime95 keeps dumping on one of the cores when I run 2 instances.
Core Temp: Idle/Full: 56-58C/66-69C

Questiosns that I have:
1) What excatly is VMCH, what does it do? When do I raise it, what does it cause to heat up, etc...

2) What does FSB Terminal affect? Stability? Faster FSB possibility? I never see people talk about this in their OC stats.

3) Should I even touch ICH setting if I dont want to mess with PCI /PCIe OC? (It's south bridge, right?)

4) My RAM can handle 1116 according to Anandtech, under 2.35V, but if I go to BIOS, and change DRAM to 1000MHz (4:5) the system becomes extremely unstable (most of the time, no XP boot) even under more relaxed timings than Anandtech posted and maxed VDIMM. Is this cause I need to change something else??

5) Also, I can not get a stable boot with FSB 410, is this a matter of VCORE? or memory timings/voltage? Or VMCH (no clue what that does, see previous question)


7)I see some people set their multiplyer on E6600 to lower number for higher FSB, but in my BIOS it is set to grey 9, with no options to change?


Also, why would only one core dump Prime95?!

Edit: Figured out #6.

RMSe17

 

[Canai]

I have a similar rig and I'm not sure about your questions, but I think that's an insanely high VCORE. I'm running 334fsb @ 1.375! I'm not sure that the VCORE would need to be set above 1.5, especially if you want to prolong the life of your chip.

Let me know if I'm way off the mark, as I will soon be pushing my rig to 3.2 or 3.4 and have just started messing with my VCORE. I'm also interested in replies to this topic, as I am unsure about those other voltages.

You should try updating your BIOS to 1407, I think that will unlock the multiplier.

ftp://dlsvr03.asus.com/pub/ASUS/mb/socket775/P5W%20DH%20Deluxe/P5WDH1407.zip

 

[The-Noid]

every 400 mhz past 3.2 ghz takes another 100 to 125 mV on air. With Vdroop he is probably only pushing 1.49. That is acceptable but you could probably take another 25-50 mV off. Don't look at Core Temp it is worthless check Asus Probe. And if you use the 4:5 multiplier you are overclocking the Northbridge (vMCH). Use 1.8v on the Northbridge if you want to go that high.

 

[Canai]

Originally posted by: Yoxxy
every 400 mhz past 3.2 ghz takes another 100 to 125 mV on air. With Vdroop he is probably only pushing 1.49. That is acceptable but you could probably take another 25-50 mV off. Don't look at Core Temp it is worthless check Asus Probe. And if you use the 4:5 multiplier you are overclocking the Northbridge (vMCH). Use 1.8v on the Northbridge if you want to go that high.

I had a lot of problems with crazy readings from ASUS Probe, and the temps it registers are not very accurate. I use Core Temp and Everest to moniter my temps and have found that they give very reliable data.

 

[Nightmare225]

Wait, the mutliplier is really unlocked in the 1407 BIOS? IF so, is it both up and down or jsut one-way?

 

[RMSe17]

Ok, so.. after posting that, I actually upped the VCORE voltage to 1.55 (still running 400MHz/3.6Ghz) and Prime95 now went from 1 minute before error to 36 minutes before error.
But the temperatures are starting to make me worried, I am getting 79 CoreTemp and ASUS Probe now shows 61C under full load. (Stock voltage, no OC, under load was like.. 27C)

And this is with Big Typhoon and AS5...

I tried dumping VCORE to 1.4 just to see if I was crazy about too much voltage, and the thing wouldnt even post at that voltage. And I mean... prime went from 1 minute to 36 minutes when I went from 1.537V to 1.55V so I am guessing voltage is correct for 3.6GHz?

Looks like 1407 answers my #7 question

I also tried maxing VMCH (1.85V) and umm.. that plastic thingy on the NB bubbled up (eh?!), WinXP promptly crashed. So I set it to lower.. 1.65 at the moment.

 

[Canai]

Originally posted by: Nightmare225
Wait, the mutliplier is really unlocked in the 1407 BIOS? IF so, is it both up and down or jsut one-way?

I'm pretty sure it's just down, to 6 I believe.

Originally posted by: RMSe17
Ok, so.. after posting that, I actually upped the VCORE voltage to 1.55 (still running 400MHz/3.6Ghz) and Prime95 now went from 1 minute before error to 36 minutes before error.
But the temperatures are starting to make me worried, I am getting 79 CoreTemp and ASUS Probe now shows 61C under full load. (Stock voltage, no OC, under load was like.. 27C)

And this is with Big Typhoon and AS5...

I tried dumping VCORE to 1.4 just to see if I was crazy about too much voltage, and the thing wouldnt even post at that voltage. And I mean... prime went from 1 minute to 36 minutes when I went from 1.537V to 1.55V so I am guessing voltage is correct for 3.6GHz?

Looks like 1407 answers my #7 question

I also tried maxing VMCH (1.85V) and umm.. that plastic thingy on the NB bubbled up (eh?!), WinXP promptly crashed. So I set it to lower.. 1.65 at the moment.

The clear plastic wrap on the mobo heatsink covers? You should prolly pull that off as all it does is impede heat transfer. If you're going to be running eerything really hot you should prolly also take off the metal covers on those heatsinks since they just get in the way of airflow. And if I remember correctly the highest operating temp that the Conroe is rated to is 60.5ºC, so you might want to be careful with that. Was there any throttling when the temps got that high?

 

[airfoil]

C2D's don't throttle till it hits 85C on Core Temp.

 

[RMSe17]

Hmm... could my heatsink be mis-installed? Or are these temperatures normal?
My system is still not very stable, like after a day on idle, I would come to see it locked up...

 

[Canai]

Originally posted by: RMSe17
Hmm... could my heatsink be mis-installed? Or are these temperatures normal?
My system is still not very stable, like after a day on idle, I would come to see it locked up...

Did you do this?
http://www.xtremesystems.org/forums/showthread.php?t=106563

(thanks phile for that)

I personally didn't but I hear it lowers temps nicely. You should also dl Everest and check what it says for your mobo temps. Also you may want to disable Q fan if you don't mind the noise. I personally prefer keeping all my fans on max to keep temps down as low as possible.

If the chipset temps are still too high you could also try wiring a couple little fans to them for a quieter solution than cranking up the other fans.

 

[RMSe17]

Yea, I did the cleaning and AS5 application under all heatsinks as stated in that link before I put everything together, the only thing I haven't done is the removal of the covers from the chipset heatsinks. I am not going to do hardware modding to up the possible vMCH though. Q Fan is disabled (give me 1300RPM instead of 1000)

I'll check out everest later today.

 

[The-Noid]

I don't think you have enough tension on the heatsink. At that voltage you shouldn't be seeing temperatures that high. Try reseating the heatsink. My guess is that between 1.475 to 1.525 vCore, 1.65-1.70 for vMCH, 2.25 on the ram, vFSB set to 1.4v, should get you 24 hours stable. Just use the 1:1 ratio for now and set timings to 5-5-5-15. Once it is stable at that you should work on memory timings. at 1.5 vCore on the typhoon with the 120mm fan I was getting 47 loaded. Possible it just needs to be reseated.

Take the cpu out and see how good the coverage is on the heatsink. If it is bad lap the heatsink and the processor and use a thinner layer of thermal compound this time to reduce the amount of friction the heat has to move through.

 

[RMSe17]

I will try to reseat it. How tight do you have the top nuts set (the ones that hold the H bracket that is holding the heatsink down to the chip)? I have about 5th of an inch of screws sticking out the top of the nuts. I can make them all tighter, but I wonder if it is going to break something? Or should they be loser?
I know I can slide the heatsink back and forth a bit with some pressure.

As far as the thermal compound, I used the razor blade to apply just enough so you can't see the CPU metal through the compound. Is that too much?

 

[Canai]

Originally posted by: RMSe17
As far as the thermal compound, I used the razor blade to apply just enough so you can't see the CPU metal through the compound. Is that too much?

Arctic Silver put out some new guidelines for applying their compound to Intel dual core chips... a line down the middleand then just stick the heatsink on.

http://www.arcticsilver.com/pdf/appinstruct/as5/ins_as5_intel_dual_wcap.pdf

I did that at first on mine but after having to take the heatsink off to put the mobo back in the case I looked at it, and I was just kinda nervous having the thermal compound only in a splotch down the middle of the chip. I reapplied it with a credit card, putting a drop in the middle and spreading it out as much as I could and putting another drop and so on until the compound barely reached the edge of the heat spreader.

Edited to make some sort of sense. I hope.