Another Q6600 OC

[Hudzy]

I know there's a few of these threads around, but I don't want to hijack. So...

I'm going for 3.6 on an Asus P5Q Pro with a G0. This is what I've set so far:

FSB: 400
Multi: 9
DRAM: 400 (1:1)
Default timings: 5:5:5:15

I've also disabled CPU Spread Spectrum and PCIE Spread Spectrum on the advice of a few other threads.

Voltages... I've set the memory to 1.8, it's rated 1.8-1.9. VCore is apparently my issue.

I started with my VID of 1.2875. I knew that would never be enough so I've been increasing it until I could successfully boot, which happened in the late 1.3 range. Prime however has failed on the first minute, again expected.

I've been increasing it slowly, until 1.4250, after which I jumped to 1.5 just to see if it would ever get past the first test. It's 15 mins stable at 1.5 (I know, I know, it needs more testing) but the temps are hitting 97 so I stopped it for now. Is it me or are those temps a bit high? I'm using a Thermalright Trueblack with a Noctua so I was expecting a little better.

Anyway, the next step is resuming small increases from 1.4250 until I get favorable prime results, but I've got some questions.

How hot is too hot? It's always varying answers with that...

Is increasing volts to the memory something I should be doing here? To 1.9 say. Or lowering timings? I'd rather not, but if it's unrealistic for it to work at default with this overclock then so be it.

I've read 1.5v is the limit for these 6600s and with 97 degrees I can believe that.

Anything I'm obviously doing wrong? It's been a while since I dabbled.


Quick update: Temps are still at 80 even at 1.43v, and it needs to go higher to get past the first prime test

 

[Johnny|lucidcomposure]

Theres something massively wrong if your at 97 Degrees surely? I wouldnt wanna run your cpu much past 65!

 

[Hudzy]

That's pretty much what I thought, I'm not really sure what to expect with a 1.5 vcore though.

 

[mattyj1085]

What cooler are you using ive got 1.5v volts through mine 24x7 clocked at 3.75ghz even had up to 1.6v to try and get 4ghz stable it boots but lasts about 5 seconds of prime

mines watercooled though but still gets up to high 70's, 90's are a bit toasty though i'd invest in a better cooling solution

 

[Hudzy]

Thermalright Trueblack. It should be pretty damn good for £60 I'd say. I bought it about a year or so back and it came with a lot of good reviews.

It could be seated incorrectly I guess but it gives me a headache even thinking about reattaching it given how long it took me last time.

Stock idle is 25 degrees though, which seems fine.

 

[Big.Wayne]

So you got the Q6600 at 3.6GHz (9x400) using 1.50vCore . . . being cooled by a Trueblack with a Noctua and the Prime95 load temps are hitting 97°C?

Is the heatsink fitted properly? . . . does the chip have a massive "concave" IHS or is your Q6600 a really low VID chip or something? . . . what [VID] does it have using CoreTemp? . . .

Depending on the answers to these questions it may be the case you need to back-off the OC a little!

 

[WingZero30]

Big Wayne op has stated in his first post that his VID is 1.2875v.

OP apart from the heatsink re-seating issue, did you change your 'FSB strap to North Bridge' to 400MHz. Also did you up the north bridge vcore otherwise known also as MCH Vcore?

 

[Hudzy]

VID is 1.2875 according to CoreTemp.

I don't recall anything unusual about the chip itself, it might well be a little concave. It's starting to look like I need to get the heatsink off and have a look at reattaching it anyway, but I've been happy with stock temps so I always assumed it was seated correctly.

Maybe I'll even dust it too But it's not terrible, I've seen a lot worse for dust.

I can always back off the OC like you say, but I'd like to hit stable 3.6 if at all possible. If it is the heat causing prime to fail, I guess that would invalidate all the previous test runs right? It might be stable at a lower vcore with lower temps?

 

[Hudzy]

Big Wayne op has stated in his first post that his VID is 1.2875v.

OP apart from the heatsink re-seating issue, did you change your 'FSB strap to North Bridge' to 400MHz. Also did you up the north bridge vcore otherwise known also as MCH Vcore?

I didn't touch either of those. I can give them a go now.

EDIT: What kind of voltages should I be looking at for the northbridge?

 

[Hairybudda]

97c at 1.5v? Something is definitely not right, especially with that cooler.

 

[WingZero30]

I didn't touch either of those. I can give them a go now.

EDIT: What kind of voltages should I be looking at for the northbridge?

What is your northbridge voltage currently? If it is by default set at say like 1.1v, then up it to say like 1.25-1.3v and see if it helps.

 

[Hudzy]

97c at 1.5v? Something is definitely not right, especially with that cooler.

I think I'll be looking at reseating the thing.

I'll measure the stock temps a bit tonight and see if they're any good.

What is your northbridge voltage currently? If it is by default set at say like 1.1v, then up it to say like 1.25-1.3v and see if it helps.

It's default at the moment. I can't see my temperatures going down though adding more volts to things so I'll try that once I've had a look at the cooler.

 

[Hairybudda]

I think I'll be looking at reseating the thing.
It's default at the moment. I can't see my temperatures going down though adding more volts to things so I'll try that once I've had a look at the cooler.

Nothbridge won't really heat up too much with a little more volts, it's probably passively cooled on your board anyway.

[edit] You could always whack your VTT up a little notch and/or increase GTL 1-2% on the failing cores to give you a bit of headroom. You really want to keep it under 75c though because you'll be risking damaging your chip though.

 

[C64]

Try 3.5 ghz at 1.4 volts

It shouldn't be hitting 80 under prime, intel burn test yes prime no

which fans have you got on the true ? need two fans in a push pull for it's best work

 

[Hudzy]

I actually bought two Noctuas for it back when I got it, but it only came with one set of clips so I'm only running the one fan. The side without a fan is right next to the exhaust though, which is another Noctua.

Stock prime temps are steady at 55, is that about normal?

 

[Big.Wayne]

I started with my VID of 1.2875

op has stated in his first post that his VID is 1.2875v.

Yup thanks but I wasn't sure if Hudzy meant [VID] or he was referring to vCore . . . some people don't know the difference hence why I specifically asked and linked an app to confirm!

So a Q6600 with a [VID] of 1.2875v is lower than average and even though it should be able to overclock using less vCore than a higher [VID] chip it will run hotter . . .

did you change your 'FSB strap to North Bridge' to 400MHz. Also did you up the north bridge vcore otherwise known also as MCH Vcore?

Neither of those settings have anything to do with getting a Q6600 stable at 3.6GHz (9x400) with the memory running [1:1] sync? . . . modern LGA775 motherboards can run 400MHz-FSB with [Auto] vNB and straps on [Auto] . . .

I can always back off the OC like you say, but I'd like to hit stable 3.6 if at all possible

Hudzy what your doing is jumping in the deepend without any voltage scaling knowledge? . . . in your boots I would have worked the overclock up from 3.0GHz to 3.2GHz to 3.4GHz etc . . . this would allow you to see how much extra vCore your chip needed between the various steps and given you some indication of its "sweetspot" . . .

As it stands all you know is the chip appears to run stable @ 3.6GHz with 1.50vCore but is hitting 97°C . . . thats too high . . . if you run IBT/LinX those temps will fly over 100°C which is not really doing your processor any favours!

I'd inspect your cooler as something may not be right? . . . the TIM may have dried up etc . . . if you remove the cooler you can inspect the TIM layer a bit like a old-hag reads tea-leaves to see what kinda contact your heatsink has had with the processor . . . . if its all good then I suspect you will need to back-off your overclock or invest in Water-cooling!

 

[Hudzy]

Hudzy what your doing is jumping in the deepend without any voltage scaling knowledge? . . . in your boots I would have worked the overclock up from 3.0GHz to 3.2GHz to 3.4GHz etc . . . this would allow you to see how much extra vCore your chip needed between the various steps and given you some indication of its "sweetspot" . . .

Guilty

I'll start working up from 3 GHz. I liked 3.6 as a goal though to keep 400 FSB and 1:1 with the memory divide. It's been fun at least

As for water... I've been there and it got too much of a pain to upgrade parts. It's why I invested in a solid air cooler.

 

[C64]

Put your cpu PLL voltage up 2 notches and north bridge voltage up two at least.
Enable Load line calibration too.

 

[Big.Wayne]

C64, he doesn't need to touch vNB? . . . if he takes that off [Auto] he will also need to start tinkering with vTT . . . 400MHz-FSB is a stock speed and is "easy" to run using stock [Auto] voltages? . . . the memory is [1:1] and he is using the processors native [x9] multi so the NBcc (real FSB) is cruising along at 400MHz . . . i.e a walk in the park

Once he has done some voltage scaling tests the time to start adjusting PLL may come . . . but adding PLL is only going to increase temps!

 

[redshadows]

3.6Ghz, 1.5v @ 97c load....with a Thermalright Trueblack.
Must be a TIM issue, a reseating issue, needing lapped, or dodgy sensors.
Sensors are very unlikely

I have a true mark one and I peak at 72 load with those volts at 3.6GHz

Trust nothing else is the issue all it can be.