I sold my friend a 3.6 Q6600 (rig)...

[soccerballtux]

that's way to freaking high for a Q6600, the spec from Intel for the dual core chips of that generation specifically states 1.5v and 72.3C

a temperature and voltage which I ALWAYS failed stress tests after hitting.

you need a better cooler
and running it that hot would definitely damage the chip. At that level of entropy and voltage the electrons are tunneling across the dielectric into the gate. Some of them jump and embed themselves directly into the dielectric, reducing capacitance on the gate, which is permanent damage

what an idiot

this is why people shouldn't overclock

 

[dave_the_nerd]

Max thermal spec on a Q6600 is only like 70C. The spec didn't change to the common-today 105C until, like, Sandy Bridge or something.

 

[SPBHM]

I would have tried 3GHz (FSB 333) with some undervolt...

 

[jacktesterson]

I would have tried 3GHz (FSB 333) with some undervolt...

Yep

He needs a cooler or just run it stock for now.

 

[TurtleCrusher]

If I could run a Q8200 OC to 2.8Ghz @ 1.05/1.1V, the Q6600 should be able to hit 3.2Ghz at 1.35V. My old EP45-DS3L couldn't push past 480FSB very well but even then voltage didn't mean a thing on the quad. Even @ 457FSB (3.2Ghz) I only needed 1.2V.

 

[VirtualLarry]

Max thermal spec on a Q6600 is only like 70C. The spec didn't change to the common-today 105C until, like, Sandy Bridge or something.

Tcase is like 72C, but Tjunction/throttling temp is 100C for Q6600.

 

[VirtualLarry]

If I could run a Q8200 OC to 2.8Ghz @ 1.05/1.1V, the Q6600 should be able to hit 3.2Ghz at 1.35V. My old EP45-DS3L couldn't push past 480FSB very well but even then voltage didn't mean a thing on the quad. Even @ 457FSB (3.2Ghz) I only needed 1.2V.

Huh? That chip does 3.3 @ 1.2v. Not sure why someone brought up 1.5v.

Anyways, the cooler wasn't totally clean, and is probably due for a fan replacement. We were slightly pressed for time, so I didn't replace them.

 

[dave_the_nerd]

Tcase is like 72C, but Tjunction/throttling temp is 100C for Q6600.

And any halfway decent OC'er with a Q6600 and a load temp over 65C would back off their OC.

To quote Jeff Goldblum, "Yeah, yeah, but your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should."

 

[TurtleCrusher]

Huh? That chip does 3.3 @ 1.2v. Not sure why someone brought up 1.5v.

Anyways, the cooler wasn't totally clean, and is probably due for a fan replacement. We were slightly pressed for time, so I didn't replace them.

Then it shouldn't have heat issues at all. Unless you're using a stock cooler, something is wrong.

 

[VirtualLarry]

And any halfway decent OC'er with a Q6600 and a load temp over 65C would back off their OC.

To quote Jeff Goldblum, "Yeah, yeah, but your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should."

That's like saying anyone with a load temp of over 65C on a 4790K should underclock.

Q6600 CPUs simply ran hot. So do Haswells.

 

[Ketchup]

And any halfway decent OC'er with a Q6600 and a load temp over 65C would back off their OC.

To quote Jeff Goldblum, "Yeah, yeah, but your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should."

Eh, I think you are exaggerating the situation, a lot. It's just a computer for pete's sake. Goldbloom's character was talking about genetic manipulation.

If I can infer: the friend knows it's old, knows it's overclocked, knows it won't last forever. If they want to pay (assuming) for more out of it, that is their decision.

 

[aigomorla]

he needs a better cooler.

even tho the Q6600 is a tank, its not superman.

it wont last long at those temps, nor will the board and vregs from all that generated heat.


And ive seen MANY X48 boards fail.
Ive killed SEVERAL during my wolfdale/Yorkfield testing days.

Those older boards were not made with the same vregs which bloomfield introduced.
Bloomfield is where vendors got smarter and used better mosfets from all the kentsfield puqing the boards.

 

[VirtualLarry]

he needs a better cooler.
it wont last long at those temps, nor will the board and vregs from all that generated heat.

Better than a Tuniq Tower 120? That thing is freaking huge.

And it won't last a 43-50C idle, 83C load, Tjunction temps?

I ran my E2140 CPUs @ 85C Tjunction under load for years, with no apparent degradation.

 

[aigomorla]

Your E2140 is not a dual die cpu.

Did u forget what is under the lid in a Q6600?

Also if ur running a tuniq tower and its still hitting over 80C you need better fans.
The only way ur gonna get quiet and lots of overclocking is with a custom water setup.

That is why i always watercooled my Q's when i shot for 400FSB+.


Also RAM @ 400fsb was very unreliable.
Unless u really played with that memory divider, and had them under active cooling, your ram would end up dying faster then you can say OMGWTFBBQ.

Remember this pic?

those were all the ram i used while testing, and almost every one of them went though RMA's.

 

[dave_the_nerd]

That's like saying anyone with a load temp of over 65C on a 4790K should underclock.

No it's not - they're completely different chips, with different acceptable temp ranges.

 

[VirtualLarry]

No it's not - they're completely different chips, with different acceptable temp ranges.

TJmax is 95 or 100C for Q6600, 100 or 105C for Haswell. That's not different enough for me to personally consider 65C as a temp to back off of with a Q6600 and not a Haswell. These aren't AMD chips, where greater than 70C often causes crashes.

 

[aigomorla]

TJmax is 95 or 100C for Q6600, 100 or 105C for Haswell. That's not different enough for me to personally consider 65C as a temp to back off of with a Q6600 and not a Haswell. These aren't AMD chips, where greater than 70C often causes crashes.

65C was the sweet spot for longivity.

This is why people use this number like a bible.
You can do 70C even, assuming that the user will not have it on a folding machine or distributed computing.

But again, 80+C if i saw a user with that, id tell him no... no again... and save money, your getting a new pc soon. :awe:

 

[VirtualLarry]

Also RAM @ 400fsb was very unreliable.

Nonsense. X48 officially supports 400FSB, and also DDR2-800 RAM (with a 2.0 RAM multi). I've run my 775 rigs at 400FSB and RAM at DDR2-800, on P35 and X48, without issues for years. (Mostly running DC, which can often be RAM-intensive.)

Edit: But I never ran "overclocker" RAM, that wanted to be overvolted.

 

[aigomorla]

Nonsense. X48 officially supports 400FSB, and also DDR2-800 RAM (with a 2.0 RAM multi). I've run my 775 rigs at 400FSB and RAM at DDR2-800, on P35 and X48, without issues for years. (Mostly running DC, which can often be RAM-intensive.)

Edit: But I never ran "overclocker" RAM, that wanted to be overvolted.

Then you got a lucky board.

Ive had X48's which had a pretty bad FSB hole, meaning at a set FSB it would not respond.

I had to counter it by going over the hole..
Example on a DFI i used it had a hole from 375-405...
If i used it on that FSB it would not boot... if i went 406-450fsb it booted fine.

The faster you push your FSB, the faster u need to push your ram.
But there's another thing which happened with ram during those times.
It was called Timings..

It was more benifical to have tighter timings then overall ram speed during that era, so to get those nice tight timings and fast ram speed you needed to push voltage.

However 7/10 vendors during DDR2 era all OVERVOLTED there ram.
The ever so infamous Micron D9's are a good example of said ram, and also why i have 8 sticks of them in that picture....

I can go on and on about LGA775 and LGA1366 as those were the 2 platforms i did extensive tests on for my sponsor.

But im telling you now, 80+C is TOO HOT.
Replace the fans on the tuniq or add another one near the exhaust in a push/pull config to get more air though it.
80+C setup for a friend is not being a friend.
Its you trying to get greedy and fueling your friends greed at your friends expense.

 

[VirtualLarry]

But im telling you now, 80+C is TOO HOT.
Replace the fans on the tuniq or add another one near the exhaust in a push/pull config to get more air though it.
80+C setup for a friend is not being a friend.
Its you trying to get greedy and fueling your friends greed at your friends expense.

That I agree with. It was running too hot. Then again, hardly any real-world program gets to the load that OCCT Linpack puts on the CPU.

I would have gladly replaced the Tuniq's fan, or replaced the whole thing with an OCZ Ventetta, I've got a few of those stockpiled away. But we didn't have time.

 

[soccerballtux]

TJmax is 95 or 100C for Q6600, 100 or 105C for Haswell. That's not different enough for me to personally consider 65C as a temp to back off of with a Q6600 and not a Haswell. These aren't AMD chips, where greater than 70C often causes crashes.

set your voltage for 1.5v after vdroop and run prime95 until you hit 74C. It will fail prime like clockwork.