Q6600 Overclocking Advices Needed.

enyceexdanny

2[H]4U
Joined
Sep 22, 2003
Messages
2,618
:confused:Hey guys, I have two or so posts spread out relating to hardware, but somehow it got into overclocking. So instead of trying to post to all those threads, I made this one post to help me progress in my overclocking.

My settings are below.
I have it overclocked at 3.2ghz. After about 14 mins in to prime95. one of the cores had an error. Not sure if I should raise the voltage more..

* EDIT 3:51 | 8/24/07 *

Raised the vcore to 1.3550v. Lowered all the rest except vdimm to defaults.
Vdimm is at 1.850v.

Currently a tad longer than an hour on prime95 with no errors on any of the cores as of yet.
Temps via coretemp are 63,61,63,61 on cores 1,2,3 and 4 -respectively.

What should I do next? You think I can go more without killing my cpu early?

Also, for some reason cpu-z shows my vcore as lower.. When in fact, I raised it..


*END EDIT*


Prime95 small FFT Load

• 400 fsb x 8 multiplier
• VCore - on bios: 1.325v | on cpu-z 1.312
• DDR2 - 1.850v Running at 4-4-4-12 1T
• MCH - 1.29v
• PCI-E - 102mhz

My load temps currently are read via core temp 0.95 and averaging at 60c.
Idle being approx. 40c

=================================================================

First, I'm not too sure about the voltages and their limits per cpu, mch, and etc.
It would be great if I can get some insight on that.

Second, my next step would be 3.4ghz and finally my goal is 3.6
But I'm kind of worried about the temps. I'm not sure if 60c on load is really hot or not.
What do you guys think? I would be grateful if someone can provide how I should go on about the bios settings regarding the speeds and voltages.

Third, does it make a difference having the FSB running so high? I'm currently at 1600. Are there any benefits? The cpu is rated at 1066 right? So what effect does it have?

That's it for now.
Thank you in advance. >_<
 
Is it a G0 or B3 stepping? If it is a B3, then you might have peaked out. You could try upping the voltage a little bit to see if you can get rid of the error.
 
keep vcore <1.5v ddr2<2.3v mch<1.48v, then you will be fine

P.S.
400x8 @1.325 is pretty good (whats your vid?)
you can prolly get 3.6G prime stable


P.P.S.
ddr2 @1.8 is a little bit low
I think crucial ballistix default setting is 2.2v
set ddr2 to 2.1v and loosen the timing to 5-5-5-15 so you don't have to worry about the memory causing OC failure
OC the memory after you find your cpu limit

One last note
how is your PWM cooling? (your cpu temp is fine, ~60c after an hour prime test is pretty nice)
if you go beyond 3.2G and use stock mobo coolers
PWM is prolly gonna overheat (reach 80c or something like that)
 
Hey cloud. Thanks for the voltage limit recommendations =)

Vid is 1.2750v
Yeah I hope to get 3.6. But the heat is already concerning me.

Ok I'll try that with the ram voltage and timings.
Hmm, where's the PWM? Well there are no active cooling on the motherboard itself.
 
install uguru (it should be in the mobo driver cd, or you can google for the latest version)
you can check pwm temp there (usually hotter than cpu)
uguru can also be used to OC cpu and set vols in windows
if no additional fans installed on mobo
your pwm temp won't be very good (but a 40mm fan over pwm cooler can cool it down pretty well)
 
I installed uguru as soon as I had this setup running.
Doesn't seem too reliable for temps atleast. Showing much lower than coretemp.
I'm not sure how I can oc it directly from this app. Can't seem to find input area for the fsb and such. Seems like it's all preset. I tried to select a preset, and it crashed my pc a bit ago. Not sure now, since it's been stable running prime so far.

wow. on uguru. PWM temp is at 77c at the moment.

But I wonder once I place all this in the case (Lian-li V1200BPlus II), if the temps would get better. Since there are 120mm fans to help the airflow.

Cpu-z's showing my fsb and speed correctly, but coretemp is showing 3.6ghz (450x8) and the property window for vista is showing 3.6ghz also..
Wonder what's up.
 
also, which section of the mobo is the pwm?
Is it right near the pci-e slot? If so, my 8800 ultra is providing alot of heat.... >_>
 
pwm is next to cpu (under a blooming shape stock cooler)
77c sounds about right w/ stock cooler (hence a fan over this area is a must)

to oc in uguru, go to "setup" (a round button), then choose "OC guru"

cpu-z/core temp did funny things w/ my vista x64 before I upgrade them to the latest versions

if you wanna go for 3.6G, you prolly should disable C1E/EIST/VT in BIOS (find "cpu features" under "advance bios setting" or something like that)

After 1hour 40 mins. 1 core failed. >_>
What should I do now? Raise the vcore?
what are your fsb/vol settings?
I'd try to raise vcore and mch a little bit more at the same time (but since you are using stock cooler for NB, I'd keep mch <1.4v if thats possible)
how is your cpu cooling?
after almost 2-hour prime95 it might be the high temp causing errors
 
I disabled the C1E/EIST and VT in bios after you told me before. ^^

I'm on 400 x 8 multi.
cpu - 1.3550v.
Vdimm is at 1.850v.
All rest at default.

Cpu is being cooled by tuniq tower
 
Oh another question. Do I need to connect the power to the auxiliary four-pin molex connector?

the one for aux fans?
well, I've been told that aux fans might cause extra vdroop so I am not using mine (plus I have a 7-fan controller for all my fans)

I disabled the C1E/EIST and VT in bios after you told me before. ^^

I'm on 400 x 8 multi.
cpu - 1.3550v.
Vdimm is at 1.850v.
All rest at default.

Cpu is being cooled by tuniq tower
I told you before? Well...I guess that means I haven't changed my option since then :D
given your setup, I'd say 100 min prime stable is good enough to show that you can go for more
you can try 370/380/390/400 x9 (10~15 min prime stable for each step, then go higher)
if getting errors along the way, raise vcore one step higher (follow the preset voltages, if still getting errors, raise mch one step higher)
repeat until you are satisfied or can't make it prime stable w/ vcore<1.5 mch<1.45
then do a long prime stress test to make sure it's actually stable
after finding your prime stable cpu setting
if you want you can OC ram by changing the ratio and raising ddr2 voltage (you can prolly get 1:1.25 ratio @400MHz FSB w/ ddr2=2.2v, but you might have to loosen it to 5-5-5-15)
 
not the aux fans, the aux for pci-e i believe.

Isn't it better to run it at 400fsb? So that the memory can run 1:1 with ddr-800 at which it's rated for?

Is it really ok to raise the vcore that high? When Intel specification says max 1.372?
I don't mind, but if it's going to kill my cpu much sooner than needed, I don't know. lol
 
fsb 400 is ddr800's default setting
but you can always OC it if you want (mine running at 1100)
and you dun need to worry about vcore if its under 1.5v (of couse OC and raising vcore are prolly gonna shorten your cpu lifetime, but for me I prolly will replace mine after 2 or 3 years so its not a big deal )

I still dunno which aux 4-pin power you are referring to (maybe I missed it when I read my manual?)
Is it on your mobo or vid card?
 
I had mine plugged in for a few weeks, and just unplugged it now - haven't noticed anything different, but then I'm not going poking around looking for problems. It's still completely stable without it, at least.

I was unplugging to make room for a PCI slot exhaust fan :)
 
Yeah i figured it was for crossfire.
Guess I need to unplug mine. lol

Anyways, upped the vcore to 1.3650, but cpu-z is showing 1.216.... Why so low? I'm lost.
Also, I upped the DDR2 to 2.1v, should I just set it to 2.2?
It was actually fine at 1.8

Well, I'm about an hour in p95. Hopefully it'll go smooth. Then up to 3.4 I go. >_>

shrugs..

man, overclocking is hard work!
 
Tried to do 3.6 for the hell of it.
Failed, not miserably - but still failed.

2 of the cores fail p95 right away, the other 2 go on for a tad longer.

What I'm wondering is if the voltages are getting applied correctly..


Vcore : 1.4850v
DDR2 : 2.10v
MCH : 1.33v

wonder what I should try next..
 
i think you are doing it right. The only thing left to do is raising the vcore if you are brave enough. my q6600 was running 1.45 @3400. I upped the vcore to 1.48.. now running at 3500. i am sure if i up it to 1.5 to 1.55 i can do 3600. cpu scales with vcore. It just up to you to decide WHEN to stop. My load temp is at around 70c now. i am comfortable up to 80c. so i still have some more room to work with. good luck with your.
 
life span??? who give a shit. 80c is under prime 25.3 stress testing ( mine 6600 under 120 ultra extreme doesnt get that hot, but if it does, i am okie with it as long as it's stable (8hours of prime 25.3). But under normal everyday use, you will never ever get close to that. My last E6400 is running 24/7 for over a year. idling at 55c and load at 75c. Running just as well as the day i installed it. I said it last year and i will say it again. High temp is overrated..dont be afraid of it. If your computer can pass all those stress tests, high temp will not kill it. High temp is only a problem for those people that runs HOT and not Prime or Othos stable but insisted that they are 'fine' and 'stable'

P.S... High voltage...as in 1.5 or over...now that's a problem i can see....but if you keep your voltage under 1.5, a high temp but stable computer is perfectly fine with me.
P.s Life span....ask yourself this...how many people have a dead cpu because of heat? or even a dead cpu at all?? the cpus of today are alot tougher than you think.
 
Vcore : 1.4850v
DDR2 : 2.10v
MCH : 1.33v

wonder what I should try next..
keep vcore at 1.4850v
set ddr2 to 2.2v
mch to 1.41 or 1.44
raise all other vols by one or two steps
if still not working
try vcore 1.5025 and mch 1.48 (but these vol are a little high for stock coolers imo)
 
My PWM temps get past 80, even close to 90, reported by uguru.
So what I did was, when it gets that high I just stop prime95.. Afraid something would get toasted.. >_>

Also, why is it that when I set the volts - all the apps report it as much lower? Except uguru of course...
 
I think they fixed that, didn't they?

Also, should I keep going with the prime run even if uguru reports my pwm at such high temps.
 
keep vcore at 1.4850v
set ddr2 to 2.2v
mch to 1.41 or 1.44
raise all other vols by one or two steps
if still not working
try vcore 1.5025 and mch 1.48 (but these vol are a little high for stock coolers imo)


When you say all other voltages, you mean all the ones on that page?
 
Cloud, I tried those voltages. Still no good. Either 1 or 2 cores fail pretty much immediately.

I set it back at 3.2 for now. And will try other options once they come along.

=/
 
Do you think it's ok to keep my memory at 1t command rate?

Was playing graw2 and it bluescreened my pc.. wondering what may be the cause of that.

The error said something about ha10kx2k? I searched it and it belongs to creative hardware abstract layer driver..
Wonder why that's happening though. Can it be the overclocking? or my memory being at 1t?

Any ideas?
 
I've decided to try to run p95 blend mode with my current settings at 3.2ghz to test the maximum stability.
Seems to run fine with small FFTs. But when I run it as Blend mode, the pc just becomes unresponsive. and I have to hard reset.

Maybe it can be my memory, so i tried relaxing the timings - still the same..
Wonder what can be the problem.

*EDIT*

Something is wrong. I set everything back to stock and it still does that....
 
Was looking thru the event viewer.
When it crashed my pc last night, the error code was as follows.

Fault bucket 48619850, type 5
Event Name: RADAR_PRE_LEAK_32
Response: None
Cab Id: 0

Problem signature:
P1: PRIME95.EXE
P2: 25.3.1.0
P3: 6.0.6000.2.0.0

But it doesn't give that error code all the time. It did that just once.

The thing is, I'm running the blend test now. With my antivirus (kaspersky) off, and it seems to be fine...

Can an anti-virus cause instability with prime95 blend stress test?
 
I guess if vcore 1.5v and mch 1.44v don't work
it means your G0 can't go for 3.6G
it's time to settle for less (3.4 or 3.2, I'd go for 3.2 now and may OC to 3.4 when its winter time)

and yes
when I said "all other vols" I meant all other vol setting on that page

not sure about the prime95 problem
I assumed that you always used "blend mode" for stress test
I think you might be right about KAV causing the problem
I usually do OC and stress test on a freshly installed XP SP2 system with nothing but drivers and testing softwares to max my OC potential
 
hmm, it wasnt my anti-virus.

What I did was disable, antivirus, sidebar and the deathadder mouse icon tray.
I tried one by one so far, and it's not antivirus or sidebar.

Going to try the mouse tray icon now.

will keep you posted.
 
Ok, weird.. It's not any one of them. Definitely not the mouse tray icon, since it's just an icon. Closing it doesn't close the actual service.

I turned off both my antivirus AND sidebar, it runs fine..


*EDIT*

I stopped the test, opened up sidebar - and kept KAV off, and restarted the test. It's running fine still....
The culprit seems to be KAV. Is that normal?


Oh, by the way - do all the components stay cooler inside a case (e.g. lian-li v1200bplus II) or just laid out on a table. Because, my case didn't arrive yet (be here monday), so I have everything setup on a table. With padding on the bottom of course. If the case would help bring my temps down, I guess I can try to hit 3.6 then.. Which will be tomorrow. >_>
 
putting them inside the case is prolly gonna add 5~10c depending on your air flow
 
Another thing, does 1t make much difference on the c2d platform?

And, for overclocking the ram - higher clock & laxed timings? or lower clock & tight timings?

Also, Is there really a stability/performance cost not doing 1:1?

thanks =)
 
What is mch & pwm?? Are these mobo specific as I have an ASUS P5N-e.
Are these something I need to change for my Q6600 B3?
I am currently running mine at 3GHz with the Tuniq Tower.
Idle temps are 40 cel.
 
Back
Top