Hi everyone,
I'm trying to reach a OC of 3.6 with my q6600 (g0). Here is my rig:
 
q6600 (g0) - VID 1.25  
asus p5k premium
2x1gb patriot extreme ddr2-800
600W OCZ Stealthxtreme psu
500gb wd hd
xfx 8800gt vid card
TRUE with Scythe s-flex (1600rpm) cooling
ANTEC 900 gaming Case
 
 
Here's a quick summary what I found so far Using cpu-z / h-w monitor /coretemp and my combinations: My ram is running 1:1 i(5-5-5-15) in all cases.
 
@3.2 (355x9)--- vcore on auto (bios 1.43 , vdroop 1.34) - temps b/w 49-55 for the 4 cores - prime95 stable 10 hrs.
 
@3.3 (367x9) - vcore (bios 1.45, vdroop 1.36), temps b/w 50-58 -prime95 20 mins (will do more later)
 
@3.4 (378x9) - vcore (bios 1.5, vdroop 1.41), nb 1.55 temps b/w 51-59, prime95 FAILS after 15 mins. - Going to retry this tonight.
 
@3.6 (450x8) - vcore (bios 1.5, vdroop ?), nb 1.55, windows only recognized two cores, did not test further.
        (400x9) - won't load into windows at all. (vcore 1.5, NB 1.55).
 
As for 3.4, I used max voltages and I can't seem to get it stable (going to retest tonight), 3.6 seems out of the question as windows wouldn't recognize 2 of my cores. Also there is quite a bit of vdroop, I might just try setting vcore at vdrrop levels in bios and see what happens. What's more important towards chip stability/damage, bios vcore or OS vcore (from cpu-z/h-w monitor)?
 
Suggestions?
 
Are there any bios settings I should play with? C1e? CPU volt reference?  
 
Does anyone have a p5k premium board that could send me their OC profile (using the latest bios 0504)?  
 
Hopefully someone here can shed some light here.

Underclock ram to rule it out as possible source of instability. Run 9x400 for 3.6ghz, not recognizing 2 cores may be due to motherboard. Consider turning on Loadline Calibration to decrease vdroop, if your bios supports it. Is that board chipset p35 or x38? P35 fsb can only be reliably oced to 1600mhz (4x400), any more is shaky, and can cause problems.

Hey,
I just edited my post for the ram settings. It is underclocked currently except when I go to 450x8.  
 
The board is a p35, it has some voltage damper setting for Vdroop (some ppl say disable, some say enable, I don't know). I did run the board at 400x6 and it was 100% stable, so that should be fine. The x38's are out of my price range ($220+ in canada) so I bought the p5k premium.  
 
I tried 400x9, but I couldn't get into windows at all, hence I tried 450 to no avail.

 
Motherboard fsb is quad pumped, so it's always calculated by x4. x6-x9 is cpu multiplier. P35 should be stable up to 400mhz single, or 1600mhz(400x4) actual. Loadline Calibration should be turned on, unless you're already on high vcore (1.5+). The cpu can handle the resulting spike. It's better than having to run higher voltage constantly to compensate for large vdroop.

Yeah it's stable at 400 hz for sure. I think it's called voltage damper under my bios not loadline calibration (the description mentions reducing vdroop).  
 
I see you hit 3.6. What were your voltages (nb/vcore/fsb)?

 
X38, no nb overvolt, 1.43 vcore, 1600mhz(4x400)
 
The max prime95 stable clock I got was slightly above 4ghz at 1.65vcore. Tuned it back for day to day use. 1600mhz fsb, or 400mhz single, should be fine for p35. Have you tried 400x9 instead of 450x8?

I tried 400x9 and it wouldn't get into windows. I even used vcore of 1.5 and nb of 1.55 in bios. It would do a quick reboot after the loading screen.
 
What was your vid? Mine is 1.25 when I boot into windows on default bios settings. Maybe that's the diff?

Haha, Umm, Dignatec... I am not sure, but I'd say you put the CPU at that 400X9, 1.4625 VCore in the Bios. So it droops near 1.43 which should be more than enough for 3.6 on a VID 1.2500

On my 1.2625 G0 I need just over 1.40 in windows loaded to get 3.6, hours prime stable. It may be something else holding you back...

The VDroop is bad for OCing, but very good under stock settings because it makes sure they last intels warranty! And if you can defeat it any way while OCing, Its less voltage, period.

Like this, My Maximus formula has zero VDroop with LLC enabled. So I can set the VCore like 4 lower, because it doesnt droop. That means the Idle voltage is lower as well!

For instance. I had a 780i with a droop of .06 in windows when loaded. To test loaded, I needed to set that VCore like 4 notches higher, just to make up for the droop! And that only helps when testing! The rest of the time it sits there idle with all the voltage. If there isnt any up and down, I can set it at 1.40 and get 1.40 under load. I dont need 1.4350 so it can droop under load to 1.40. Get it?

--Lupi

Hey Lupi,
Was wondering when you'd post and help me figure this out...lol.
 
Maybe you misread what I posted? I did try 400x9 at 1.5 vcore (in bios) and NB voltage at 1.55 and it wouldn't get past the windows loading screen. Is there something I'm missing here?
 
As for VID (tell me if this is right): I loaded default bios settings and booted into Windows, ran cpu-z/everest and both reported 1.25 for vcore. This is the right way to find my vid right?  
 
I don't have LLC in my p35 board, but I have some voltage damper that I can enable. I'd like to ideally have the lowest voltage settings of course, only because of this massive vdrop/vdroop I'm having to set the bios value fairly high (e.g @3.3 bios vcore 1.45, vdroop vcore 1.36 from cpu-z, that's almost a 1.0 vdroop!)
 
I think I understand...What should be happening is that i set 1.40 in bios, windows idle should show 1.4 and windows load should show 1.4. However, my board is taking that 1.4 and changing it to 1.38 (vdrop)(for ex) under windows idle then vdroops to 1.36 under load. Correct?  
 
In my example, if the system is not on load, is the cpu getting 1.38 or 1.40?  
 
Man this stuff makes my head hurt...lol.

Thats correct. The feature is to make sure that the voltage never spikes over the VID. So a VID of 1.3000 will look like 1.275 or so in windows. Then under load, it draws alot of juice, so it droops!  
 
And your system gets the full windows voltage. If its 1.3000 in the bios, without LLC it will be 1.28 or a lil lower in windows. Thats what your cores get. BUT at a higher speed, with a bad vdroop. you'd need to place it on say, 1.3250 or more to make up for a droop, if it was bad!  
 
So on the 780i it was getting 1.45 volts continuous, when loaded it went down to 1.38 anyways! get it? If I could start at 1.38 and get 1.38, thats like 6 VCore notches that I dont need to set higher to offset the VDoop.

I made a mistake Lupi, my damn VID is 1.325 from coretemp/everest at default settings. That's probably why I'm stuck at 3.2 and can't get past it reliably. I was able to get 3.3, but I had to crank up the juice on all voltage param...
 
Also, I read an article from anandtech that makes a very strong case against LLC and why vdroop is actually a good thing...It sounded pretty convincing to me , check out the graphs:
 
http://www.anandtech.com/cpuchipse [...] i=3184&p=5
 
Using PLL inflates your max stable OC with prime95, but your system isn't truly stable under gaming conditions. What do you think?

I think LLC does what its supposed to, and no, I have a q6700 that was a low VID and I noticed that it may not be my board anymore. its just needs like 6 more VCore notches than I though, that's too much! But since the 6700 starts out 266 Mhz better, I got to see it first hand! Thats a great piece of info. It means that a q6600 will have an upper range of 3.6... after it to get stable in prime 95 you'll need a semi-large boost in VCore.  Subject to your VID of course!!

LLC is a wonderful tool when it works, and will be more than stable for gaming, if it can get stable in prime! I use it all the time, to various degrees. And the better the LLC, then better the OC has been for me! Simply because you dont need extra VCore to offset VDroop. Its what it says that it is when you boot and look. It will not go down from that CPUz stated core voltage. It may go up a tiny amount to promote stability even! Mine does!!

--Lupi

That is wrong. Unless you're already running on very high vcore, the spike won't destablize your system or kill your processor. Prime95 is a test for truly stable condition. The alternative is keeping a constantly high vcore to compensate for high vdroop.

BTW 1.3250 is the worse one! I have one of em I just got today, so when I test it, you'll know, and so will I, where yours should be able to go!
 
--Lupi

Hey Dagger, TBH, i don't know what to believe. I know that when I enabled PLL (voltage damper whatever it's called) I was stable at 3.4, but with it off, prime95 in a few seconds gives me errors.
 
Lupi: Yeah I figured as much, pissed off but not much I can do. Maybe you'll have better luck (since I'm an admitted noobie at this) and you can OC it better than I can. Thanks man.

I can tell you right now, try 1.5000 in the bios, trying for 1.47-8 in windows and just watch the heat. Thats about what it will take. And the only real problem becomes the heat, since you are within intels rated specs. 1.5 volt max.

As long as its cold, you could go up to 1.52500 in the bios, targeting 1.50 in windows. Some people would say thats to high, but it should last a good amount of time, even at that voltage as long as its nice and cool, and remember, you'll virtually never see a load like small ffts in a real life situation. So what you test at, you'll always be under it. Thats the very max.

I will give you a better 3.6 voltage, when I lap and re test it out to get screen shots for my collection of data. Once I am done testing them, someone is gonna be happy because I will put them up for sale on craigs list at less than retail, with optional overclocking. I'll already know where to put them! And I will write a small review of the data and post it here for everyone, with my findings!

Just wait till you read what I have to say about intel re branding their low VID chips as extreme chips, and selling them for 500 bucks more! Its nuts! I had to do a lot of weird research! Because so many factors went into why they OC well, down to economics!

Forthcoming...

--Lupi

Hey,
I did try 1.5 vcore (in bios) and NB voltage at 1.55, at 400x9 it would hang right after I got out of bios (had to shut down) at 450x8, it gets into windows but only shows two cores.... If you can get into Windows with all 4 cores, post your settings so I can try as well.
 
I agree small FFT's is literally pushing the system to max so it's not real life. My temps haven't broke 60 C so far, but I've read that once your voltages go high, even with good temps, there's a fair bit of electron migration (like putting your thumb over the end of the hose, adds too much pressure).
 
Do you mind trying it without lapping? I'd like the system to be comparable since we're both essentially starting from scratch (plus it will save you a few hrs...lol).  
 
Sounds good. I realized last night that intel must be doign the same thing, taking the low vid chips, OCing them themselves and calling it a new model....Leave it to intel with their shoddy practices...
 
Looking forward to your post, thanks buddy.

Sure, I will get on that a lil later today. I actually did pick up another q6600. Now I have the full set!!
 
1.2000, 1.2500, 1.2625, 1.3000, 1.3250!
 
I am thinking that it may be something else. But I will know for sure if I can get into windows and pass a while of testing, huh?
 
And yes, now that I have been studying WHY the processor OC's and some dont, I am begining to see intels tricks! Very smart business practices. Why let the OCers reap the speeds and steal our cash?? LOL!
 
--Lupi

Pretty much, I bet we make up maybe 5-10% of people who buy their products, ie. 90-95% of people won't know the diff b/c they don't OC at all...
 
Given that the q6600 costs $240, why would I go out and spend $1000 on a quad core that runs at 3.0 when I can get mine up to 3.2 safely (hopefully more after you post back!)>
 
What mb will you be using? Hopefully an ASUS p35 (p5k premium if u have one given taht u have so many procs) haha.  
 
Let's see what happens.