q6600 temps seem high

[Karadhas]

I ordered a q6600 (B3, bugger) from newegg last week and got everything assembled this monday. Since coretemp 95 and 94 both reboot my computer I need to use speedfan with the +15 offset trick to guess at temps. I'm on all stock except for lowering the voltage to 1.2125V in the bios.

Running prime95 for just a minute or two, even with my tuniq tower's fan manually cranked up to max, gives me some pretty high temps for stock settings.

CPU: 49C
Core0: 58C
Core1: 57C
Core2: 56C
Core3: 58C

Idle temps are around 40-45C across the 4 cores. So do I need to redo the thermalpaste and reseat my heatsink or is this about what I should expect from a b3?

[bing]

I think that is close to calculation and normal with your room temp at about 30 Celcius ! :)

Why ?

Lets do some "rough" math . :D

Q6600 B3 TDP = 105 Watt (full loaded)
Tuniq Tower thermal resistance = 0.12 C/W (on full fan speed and properly mounted)
Ambient temp = 30 C (85 F)

So the CPU temp on full load "theoritically" is :

CPU Temp = Ambient Temp + (HSF Thermal Resistance x Watt) = 30 + (0.12 x 105) = 42.6 Celcius

Maybe remount ?

Remember, software based cpu temp reading never accurate.

[Sam__]

Lets do some "rough" math . :D

Q6600 B3 TDP = 105 Watt (full loaded)
Tuniq Tower thermal resistance = 0.12 C/W (on full fan speed and properly mounted)
Ambient temp = 30 C (85 F)

So the CPU temp on full load "theoritically" is :

CPU Temp = Ambient Temp + (HSF Thermal Resistance x Watt) = 30 + (0.12 x 105) = 42.6 Celcius




nice little equation...might have to note that down for myself....are most aftermarket HSs around the 0.12c/w mark??

[bing]

nice little equation...might have to note that down for myself....are most aftermarket HSs around the 0.12c/w mark??

That is the problem with today HSF makers even the high ends, they sort of "lazy" to put this very basic & crucial technical specification. Tuniq does have it and most high end HSF makers did "in the old days". :D

Intel LGA775 stock HSF officially stated they're at 0.2 C/W mark while Tuniq Tower 120 is at 0.12 C/W when on full fan speed.

Please remember this number means nothing if the HSF is not mounted properly with good TIM.

Most high end are around 0.10 to 0.15 mark, while on water cooling the good one can be reach down to < 0.1 C/W.

Selling HSF wihout these number its like selling an iron without mentioning it's wattage but put tons of claim sthat it can be hotter than a sun or better ironning certain fabrics or cloths, strange isn't it ? :D

[SuperDave1685]

Oh wow. Thanks Bing for the information and dandy equation. According toy our equation, my Thermochill PA120.2 has a .025C/W Thermal resistance mark with fans at 12v. I am going to be using 4, 114cfm Panaflows with this radiator, so according to your results:

Cpu Temp = 23C + (0.025x105) Equals a idle CPu temp of ~25C

Does that sound right for a QUAD core... 25C at idle?? Seems kinda low dont ya think? even with water cooling.... Is there a more concrete equation?

[Elluzion]

yes that does seem pretty high. have you pushed it to overclock at all? or just messed with the voltages?

[bing]

Oh wow. Thanks Bing for the information and dandy equation. According toy our equation, my Thermochill PA120.2 has a .025C/W Thermal resistance mark with fans at 12v. I am going to be using 4, 114cfm Panaflows with this radiator, so according to your results:

Cpu Temp = 23C + (0.025x105) Equals a idle CPu temp of ~25C

Does that sound right for a QUAD core... 25C at idle?? Seems kinda low dont ya think? even with water cooling.... Is there a more concrete equation?

No, there is no other more concrete equation for this case. :)

That 105 is on load, well when idle it will even cooler, but again, 0.025 c/w mark for that radiator, NOT for the "overall" system block + rad. I believe the number should be alot higher than 0.025 c/w. ;)

[Karadhas]

Still, for a tuniq tower on an undervolted cpu the load temps seem pretty close to that 62C Oh**** limit. Any chance that the load temps are from a botched thermalpaste application... like doing the AS5 suggested horizontal line thing wrong? That might explain why the fan speed doesn't seem to be making a difference to my temps. That or having the side a few centimeters away from the rear exhaust fan and exhausting directly into the psu fan.

[edit]

Running a double instance of orthos right now to stress all 4 cores and core2 apparently produces errors. That's the one that's always been a lot cooler than the others too. Everything else will be 48-50C and that one is 45C. I'm starting to think something isn't kosher in denmark.

[SuperDave1685]

lol bing.. well according to my waterblock, its C/W is about .033... my pump does a little more than 5 gpm, so with the block and radiator in the loop, I'm guessing the actual flow rate is about 3.7 or so. So it'd be:

CT = 24C + ([.025+.033] x 105) = ~ 30C

Still seems low lol

[bing]

That 0.025 c/w from radiator is for the coolant with certain flow rate, not CPU.

And that is not how to calculate "overall" Water Cooling's loop C/W by summing up the block's c/w with the rad. :D

Sorry OP, this is getting OOT, SuperDave, I suggest you start the same thread up on WC section to further discuss it. :D

[SuperDave1685]

lol sorry to the OP. Bing- can you pm on how to figure the C/W?

[Karadhas]

Np, ADD seems to come with technical skills >P

I've been reading around and a couple other people have the same problem I do with almost the exact same temps. One was already OCed to 3GHZ using a ultra-120 extreme (b3 stepping too). It seems we all erroneously followed Arctic Silver's instructions and put way too much TIM on the cpu. I'm going to redo mine and report back to see if the results have changed.

It's 75F in my apartment right now, theres no way in hell an undervolted core2quad should ever be able to have higher temps on a tuniq tower than a prescott P4 did on it's stock heatsink.

[SuperDave1685]

To the OP: You stated your Core temps were:

CPU: 49C
Core0: 58C
Core1: 57C
Core2: 56C
Core3: 58C

is that without the +15C trick for speedfan? If not, that means your CPU is about 72C on load :eek: What kind of fan do you have on your Tuniq at the moment? I highly doubt that a re-application of Thermal paste will make a hug difference. Some, but not jaw-dropping. how many cfm's is your fan rated for?

[jellysandwich]

Since coretemp 95 and 94 both reboot my computer I need to use speedfan with the +15 offset trick to guess at temps. I'm on all stock except for lowering the voltage to 1.2125V in the bios.

interesting, the fan specs arent listed (http://www.newegg.com/Product/Product.asp?Item=N82E16835154001)

[Karadhas]

that's WITH the 15C offset, and i'm using the stock tuniq fan. I've got the little ******* open in my room right now and I'm pretty sure I just fscked up the heatsink application. At least half the cpu's thermalpaste was still smooth, just like I left it when I screwed the tuniq down, and i'm pretty sure there was way too much on there because there was a fairly thick layer on the heatspreader AND the heatsink.

What really clued me in to something being wrong were my idle temps being as high as my prescott's load temps. Even a quaddie should not idle at 48C in a 70-75F apartment when the heatsink's fan is at max, and why on earth is that one core always so much cooler?

I really hope I don't need to lap that damn thing.

[edit]

No, they aren't. It's a pretty hefty fan though, I can feel significant air movement on the exhaust AND the intake sides whenever I put my hand in there while it's going. It's got enough going for it that it pretty much cleans itself, dust just gets sucked right through and blown out the exhaust.

[SuperDave1685]

It's 75F in my apartment right now, theres no way in hell an undervolted core2quad should ever be able to have higher temps on a tuniq tower than a prescott P4 did on it's stock heatsink.

You have to remember that the P4 was a single core architecture. Even the fastest P4 Prescott was no more than a 70W TCP design. Bing's comments got me searching :) Let me do a little math for you.

CPUT= ambient+ (HSF's C/W x TDP of your CPU)
CPUT= 23.89 + (.21 x 70C) = 38.59C for a P4 Prescott

A Q6600 B3's TDP is 105W. Soooo:
CPUT= 23.89+ (.21x 105)= ~46C temp for a Q6600.

Now this is all in theory of course.

http://i9.tinypic.com/6fr8vhe

here they have taken a P4 and oc'ed it 4Ghz. You can see the Tuniq does pretty well. Put that is just ONE core. We're talking about 4 of those puppies now. If I am right, the Tuniq was designed for single and dual core systems. As Nol pointed out to me, These Quad Cores are producing 50% more heat in the same surface area. And the good ole adage- theres no replacement for displacement is true. I do not think that without increasing the HSF's surface area (i.e making it bigger), you're not going to see a whole lot of improvement in temps. You could try a higher cfm fan and that might net you several degress because of more airflow. You could also try and lap the HSF's base and the IHS on the CPU and see if that could net you a few degrees. These are just my thoughts. If you look around, you'll see a lot of folks having heat issues with Quads simply because current HSF's aren't quite designed to handle such a large heat load efficiently.

-Dave

[SuperDave1685]

The Quaddies are hard to get a HSF mounted on correctly. As you can see from the diagram below, the cores are not perfectly centered under the HSF:

http://i17.tinypic.com/4kfb2is.jpg

This makes mounting a HSF, waterblock, TEC, etc, difficult. The IHS's are also known not to be very flat on the new Quads. It could be that the one core is making really good contact with the IHS and the other cores aren't. You can't be for certain unless you lap the IHS to make it smooth and then re-mount and check again. If you're using pressure springs to mount it, I'd suggest replacing the pressure screws with some screws the same size, but with wing nuts. That way, you can tighten down the side on which the cores reside a little more snug to ensure you have good contact on the off centered cores. Make sense?

[Karadhas]

Well since one of the heads on the screws of my tuniq tower just popped off it seems to be a moot point for the time being. That sort of mechanical fault could definitely explain odd temps, heatsink was loose on one corner.

[SuperDave1685]

That could do it lol. like I said, just take one screw out, go to home depot, and get a screw of matching width and get 4 of them and get a wing nut for each one. That should do it for ya :)

[Karadhas]

Yeah, i'm running with 2 diagonally atm and what is imho a much better thermalpaste application. I'm going to take what's left of the two-piece tuniq screws to home depot tomorrow and try to get replacements.

I'll also be sending an email to tuniq informing them their product is defective and should be recalled. What kind of idiot sells a heatsink with screws made of two pieces superglued together?

[edit]

Nevermind. I can't even boot up, the mobo emits a very loud long beep which raises to a higher. GOing to put the stock heatsink in there with some AS5

[dudleycpa]

Yeah, i'm running with 2 diagonally atm and what is imho a much better thermalpaste application. I'm going to take what's left of the two-piece tuniq screws to home depot tomorrow and try to get replacements.

I'll also be sending an email to tuniq informing them their product is defective and should be recalled. What kind of idiot sells a heatsink with screws made of two pieces superglued together?

[edit]

Nevermind. I can't even boot up, the mobo emits a very loud long beep which raises to a higher. GOing to put the stock heatsink in there with some AS5

That's Strange - Your not the first person who has problems with the screws that come with a Tuniq Tower.

[Karadhas]

Class action suit imho.

[edit]

shrill tone identified, plug in the 6-pin PCIE power. idiot. *headsmack*
Going to be running the fan on higher than normal speeds until I get replacement screws. Hopefully it will function at least as well as switching to the stock heatsink which with the tuniq backplate would be a PITA.

[edit2]

And running 4 instances of prime95, 1 per cpu, my temps are now 66, 66, 60, and 61 under load and 47, 45, 43, 46 when idle. So is it safe to use this thing or should I hold off on anything except desktop use until I get replacement screws. And why is that one core always so much cooler than the rest...
If my bios would let me I would drop the voltage some more and underclock to about 2ghz from 2.4 to try and keep temps lower.

[edit3]
Underclocked the FSB to 222mhz and left the multipier at 9 and dropped the voltage down too. Idle is lower by a bit. I think it's safe to leave it for now.

[Sam__]

That is the problem with today HSF makers even the high ends, they sort of "lazy" to put this very basic & crucial technical specification. Tuniq does have it and most high end HSF makers did "in the old days". :D

Intel LGA775 stock HSF officially stated they're at 0.2 C/W mark while Tuniq Tower 120 is at 0.12 C/W when on full fan speed.

Please remember this number means nothing if the HSF is not mounted properly with good TIM.

Most high end are around 0.10 to 0.15 mark, while on water cooling the good one can be reach down to < 0.1 C/W.

Selling HSF wihout these number its like selling an iron without mentioning it's wattage but put tons of claim sthat it can be hotter than a sun or better ironning certain fabrics or cloths, strange isn't it ? :D

Just done a bit of maths to find the rough estimate of the thermal resistance.
this is what i did:

Ambient : 24
Cpu temp with fan on full : 36
Wattage of my cpu : 95
let Y = thermal resistance


24 + ( Y x 95 ) = 36
Y x 95 = 36 + 24
Y x 95 = 12
Y = 0.1263

So the thermal resistance of my titan amanda is 0.1263 C/W aparently.

[blackjackel]

I ordered a q6600 (B3, bugger) from newegg last week and got everything assembled this monday. Since coretemp 95 and 94 both reboot my computer I need to use speedfan with the +15 offset trick to guess at temps. I'm on all stock except for lowering the voltage to 1.2125V in the bios.

Running prime95 for just a minute or two, even with my tuniq tower's fan manually cranked up to max, gives me some pretty high temps for stock settings.

CPU: 49C
Core0: 58C
Core1: 57C
Core2: 56C
Core3: 58C

Idle temps are around 40-45C across the 4 cores. So do I need to redo the thermalpaste and reseat my heatsink or is this about what I should expect from a b3?

how do you get temps from all the different cores? is this special motherboard support?

EDIT!! SOLVED!!! Coretemp does the trick :D

[Karadhas]

Ok, ignoring the continuing threadjack my latest feat of possible stupidity is underclocking to 2.0ghz (222x9) and undervolting to just above 1.100v. So far 4 instances of prime95 running torture tests for max heat generation have gotten load temps down below the 62C limit. Hopefully this will also still be stable for gaming.

[edit]

So I've had prime95 running 4 instances (one for each core) for about 30 minutes now and I'm 59-59-54-57. Even before the thumbscrews broke on the tuniq I was getting temps that were ridiculously high. Every other temp thread for b3 stepping q6600's I've seen has people getting temps like this only after overclocking to nearly 3ghz. I'm starting to think that I got a bad chip from newegg or something. I just don't see any reasonable explanation for how absurdly hot this thing has been since I installed it, especially since it's been like this from when the heatsink was still fine.

[edit2]

This can't be right. underclocked to 2ghz and down to almost minimum voltage and only one core is showing temps around what people get when OCed to 3ghz? It's been like this since I got it and that cooler core keeps sending back errors in every test i've done (even on stock everything). I've only had this for 3 days, I can RMA to newegg for replacement or get an E6850.

So... give quads another chance or what?

http://img120.imageshack.us/img120/7175/wtftempsnu0.th.png (http://img120.imageshack.us/my.php?image=wtftempsnu0.png)

[SuperDave1685]

I strongly urge you to remount your HSF SEVERAL times... have a read over here at XS. Its definitely worth the time to help get your temps down:

http://www.xtremesystems.org/forums/showthread.php?p=2348359

http://www.xtremesystems.org/forums/showthread.php?t=131481

Great reads. I hope it helps ya out man...

[Karadhas]

Thanks, I'll give that a try after getting some replacement bolts for the tuniq tower tomorrow. I'm worried about breaking the processor though. About how tight should I have the screws for the heatsink if I hardmount like they suggest? I don't have any of the foam on the backplate since it's my second use of it. Needing to take the power supply out to tighten one of the screws makes the constant-reboot method kinda useless for me...

I'm still going to return the processor for another quadcore though, I started out by breaking 66C under load and even underclocking and undervolting by this much only brought it down to 60C or so, except for that one blasted core that's at times 10C below the rest.