Just got a Q6700

[error8]

Jaredpace, you love to see people push their hardware over the edge.

I'm not sticking to 60 C, since I've already hit 72 C in Linpack, so only 18 C until Tjunction. I'm not extremely comfortable with that, so I don't know if it's safe to shoot for a 3.6 ghz. I need someone, to tell me if I should go further. But not you of course, I know that even if I'm touching 100 C you'd still advise me to go further. :laugh:

 

[n7]

If it gets too hot, it'll shut off

I didn't run LinPack much back with my Q6700, & as it can be 10C or higher than P95, i have no doubt my LinPack temps woulda been very warm.

70C doesn't worry me.
Actually 80C doesn't even.

Majority of stock OEM systems out there easily hit those temps under load...they run just fine as that's how they are designed...(shitty airflow & coolers)

 

[evolucion8]

I'm quite in the same situation, my CPU under prime95 using small fft can reach up to 89C and the first two cores reaches 86C and the last ones reaches 92C, the Vcore is set at 1.42V which was the only way to keep it stable, but since I moved to x64, my overclocks are no longer stable, why? I'm expecting to receive the Sin-Etsu G-751 Thermal Compound and apply some to see if it can help me, currently I'm using a white grease compound that came with my Sunbeam Core Contact cooler, why these Qxxx series are so vcore hungry??

 

[Tempered81]

Originally posted by: evolucion8
why these Qxxx series are so vcore hungry??

They come from a different production line. Different specifications, batches, wafers, process sizes, production week, stepping, different TDP, thermal design, Tjmax, etc...

Check out these TJmax:
http://www.xtremesystems.org/f...3380398&postcount=2400

The QX series chips have different release dates and TJmax values. Yours in particular could be as low as 75C.

 

[evolucion8]

Originally posted by: jaredpace

Originally posted by: evolucion8
why these Qxxx series are so vcore hungry??

They come from a different production line. Different specifications, batches, wafers, process sizes, production week, stepping, different TDP, thermal design, Tjmax, etc...

Check out these TJmax:
http://www.xtremesystems.org/f...3380398&postcount=2400

The QX series chips have different release dates and TJmax values. Yours in particular could be as low as 75C.

I found in an Intel Whitepaper that the max Tj was 90c, and there it states 80c, that would explain everything, may be I will have to move to Alaska or Canada so I can keep the temperatures in control, this CPU will be great as a heater in my room in those countries.

 

[error8]

Originally posted by: evolucion8
I'm quite in the same situation, my CPU under prime95 using small fft can reach up to 89C and the first two cores reaches 86C and the last ones reaches 92C, the Vcore is set at 1.42V which was the only way to keep it stable, but since I moved to x64, my overclocks are no longer stable, why? I'm expecting to receive the Sin-Etsu G-751 Thermal Compound and apply some to see if it can help me, currently I'm using a white grease compound that came with my Sunbeam Core Contact cooler, why these Qxxx series are so vcore hungry??

So you're feeding it 1.42 V from the bios, or that is the voltage under load?

 

[error8]

Originally posted by: n7

70C doesn't worry me.
Actually 80C doesn't even.

Majority of stock OEM systems out there easily hit those temps under load...they run just fine as that's how they are designed...(shitty airflow & coolers)

Yeah, but the stock OEM systems don't have their cpus overvolted. So even if they hit 80 C, they're doing it on stock vcore and maybe that saves them from early death.

 

[Idontcare]

Originally posted by: jaredpace

Originally posted by: evolucion8
why these Qxxx series are so vcore hungry??

They come from a different production line. Different specifications, batches, wafers, process sizes, production week, stepping, different TDP, thermal design, Tjmax, etc...

Check out these TJmax:
http://www.xtremesystems.org/f...3380398&postcount=2400

The QX series chips have different release dates and TJmax values. Yours in particular could be as low as 75C.

Jared I believe some of your TJmax are wrong because the source (Intel) was wrong at the time of publication.

Intel corrected the data in a Nov revision to their official publication on TJmax.

See:

Originally posted by: Idontcare
See slide 15 of this official Intel document regarding DTS and TJmax.

From this thread.

Q6700 (and Q6600) TJmax is 90C.

 

[error8]

I just tried a quick Linpack test at 1.5 V in bios and at 3.6 ghz, loads at 1.42 V. It passed only 1 out of 10 tests. I could probably get it stable at this speed if I'll fiddle around with VTT and NB,SB voltages, but the temperatures are getting really high, 10 C until Tjunction. I even had the impression that it smells like burned plastic when it was tested under Linpack and PWM makes a snake sound, that got pretty loud. I think I'm going to stay at 3.5 ghz, it feels much safer. Besides, I don't even have warranty for this chip, so no point in killing it.

 

[Tempered81]

Yah, I see you're right, Idontcare. Oh yeah, sure enough those old charts have different data for Tjmax. It says, "Corrected on Nov 18th" in the slide. Evolucion8 has a QX6850 (Tjmax=80C), I thought he was talking about his chip being hot/vcore hungry.

So Q6600/Q6700 should behave relatively the same then - i guess.

 

[evolucion8]

Originally posted by: error8

Originally posted by: evolucion8
I'm quite in the same situation, my CPU under prime95 using small fft can reach up to 89C and the first two cores reaches 86C and the last ones reaches 92C, the Vcore is set at 1.42V which was the only way to keep it stable, but since I moved to x64, my overclocks are no longer stable, why? I'm expecting to receive the Sin-Etsu G-751 Thermal Compound and apply some to see if it can help me, currently I'm using a white grease compound that came with my Sunbeam Core Contact cooler, why these Qxxx series are so vcore hungry??

So you're feeding it 1.42 V from the bios, or that is the voltage under load?

Both, I use Load Line Calibration, and in idle and full load uses the same 1.42V in the Vcore. For some reason, when I selected the Vcore by myself, I couldn't get it stable no matter what, and when I set it to Auto, it selected the same 1.42V that I chose before, so that's odd.

I used the Shin-Etsu thermal paste and the difference between it and the AS5 were dramatically different. Now at full load it never goes beyond 72C at the main core, and before it used to reach 82C to 85C, and in every core, it uses to run like 82-82-90-90 and now it runs at 67-67-72-72. And is quite strange because I applied the AS5 the same way, using the Pea Size method. So if is true that my Tj max is 80C, there no point then to overclock further, 3.60GHz on air on this CPU is quite a feat considering that I live in a tropical island were the room temperature hoves between 78F to 92F loll

PS: I heard that the Q6600 G0 stepping is pretty much a locked QX6850, so that's why lots of people can reach up to 3.60GHz with that CPU, the same results as my own CPU.

 

[E4300]

45nm duallies and quads (stock voltage and speed) will run okay @ 75C in those HP/DELL gumball machines. Overclocking a chip will lower the optimum stability threshold to about 65C. You could run hotter, but the chip will need more Vcore to pass Prime (brute force method).

For me, the sweet spot is 1.36Vcore or 65C @ 80F ambient, whichever comes 1st. Don't need to pump an extra 70-90mV to add another 100MHz of overclock. If the PC is not 100% stable with some applications, then I simply back-off 30-50MHz of core speed. Why torture my rig with LinX?

A good chip will hit 4.0GHz with minor bump in Vcore. A bum chip will struggle to break 3.5GHz. Folks spend way to much $ on fancy cooler and board only to squeeze an extra 70MHz out of a chip. That's a lousy return on your investment.

 

[error8]

Originally posted by: evolucion8


Both, I use Load Line Calibration, and in idle and full load uses the same 1.42V in the Vcore. For some reason, when I selected the Vcore by myself, I couldn't get it stable no matter what, and when I set it to Auto, it selected the same 1.42V that I chose before, so that's odd.

I used the Shin-Etsu thermal paste and the difference between it and the AS5 were dramatically different. Now at full load it never goes beyond 72C at the main core, and before it used to reach 82C to 85C, and in every core, it uses to run like 82-82-90-90 and now it runs at 67-67-72-72. And is quite strange because I applied the AS5 the same way, using the Pea Size method. So if is true that my Tj max is 80C, there no point then to overclock further, 3.60GHz on air on this CPU is quite a feat considering that I live in a tropical island were the room temperature hoves between 78F to 92F loll

PS: I heard that the Q6600 G0 stepping is pretty much a locked QX6850, so that's why lots of people can reach up to 3.60GHz with that CPU, the same results as my own CPU.

I guess you probably had a bad mounting, since I don't think that a thermal paste could do such a dramatic difference. Because if it does, I'll go tomorrow and buy a kg of that stuff. :laugh:

What have you used for stability testing?

 

[evolucion8]

Originally posted by: error8


I guess you probably had a bad mounting, since I don't think that a thermal paste could do such a dramatic difference. Because if it does, I'll go tomorrow and buy a kg of that stuff. :laugh:

What have you used for stability testing?

Probably, but I remounted that thing like 4 times with no difference, I used AS5 and the white thermal grease that came with the Sunbeam and made no difference, I use Prime95 25.9.40 64-Bit version and I select the small FFT tests which tends to work harder on the CPU because the data remains within the cache, compared to the Large FFT which data moves forth and back in the RAM. http://www.hardocp.com/article...w0LCxoZW50aHVzaWFzdA== <<That link shows some interesting results with different thermal paste, so give it a try and see if you can get some improvements. http://www.svc.com/g-751.html <<That is the link where I bought them, it took 8 days to arrive here but I live in P.R., so the shipping should be way faster in the Continental U.S. (If you live there)

 

[error8]

I live in Europe, so it will be hard to get that paste. Anyway, if you hit 67-72 with Prime95, you're probably going into 82 C in Linpack, so you are in the danger zone.

 

[evolucion8]

Originally posted by: error8
I live in Europe, so it will be hard to get that paste. Anyway, if you hit 67-72 with Prime95, you're probably going into 82 C in Linpack, so you are in the danger zone.

Heck, it was even worse!! I hit 92C and my system crashed within 45 second after I started to run Linpack, what the heck is this program? Runs very strangely, with lots of spikes in the Task Manager, and hence, is still able to almost melt my CPU. Now I'm checking if it's because of heat issues or If I can lower the Vcore to make it stable, If I'm not able to stabilize it, I will just leave it as is, currently there's no application that can tax the CPU that hard, not even at the Prime95 level which is less taxing for the CPU compared to Linpack. As it was never had a single issue running Folding@Home SMP Client etc.

 

[error8]

Originally posted by: evolucion8

Heck, it was even worse!! I hit 92C and my system crashed within 45 second after I started to run Linpack, what the heck is this program? Runs very strangely, with lots of spikes in the Task Manager, and hence, is still able to almost melt my CPU. Now I'm checking if it's because of heat issues or If I can lower the Vcore to make it stable, If I'm not able to stabilize it, I will just leave it as is, currently there's no application that can tax the CPU that hard, not even at the Prime95 level which is less taxing for the CPU compared to Linpack. As it was never had a single issue running Folding@Home SMP Client etc.

Yeah, I know, Linpack takes everything to the edge. It's the Furmark of cpus, but it's the ideal stability tester. Sees any weakness in just a couple of minutes, while Prime95 can work for hours without any issues. And, probably nothing on this Earth will load up your cpu as Linpack does, but for me, if the cpu doesn't passes at least 10 tests in Linpack, it means that it's unstable and I have the fear that someday, I'll get a BSOD in some program, because my cpu wasn't 100% stable.

Anyway, be careful, since 92 C is over your Tjunction, if I'm not mistaken.

 

[error8]

Got it stable at 3.6 ghz,
Vcore 1.52 V in bios , 1.46 V under load
VTT 1.4 V
NB 1.45 V
SB 1.8 V
DDR @ 720 mhz
Max temp is 77 C in Lipack, 69 C in OCCT.


Now my dilemma is what is faster: 3.5 ghz and DDR2 @840 mhz ( 5-5-5-15) or 3.6 ghz and DDR2 @ 720 ( 5-5-5-10) ? Anyone having an idea about this?