E6600 at default = >70 degrees stressed (!)

[Breeze]

This is bizarre: I installed my retail E6600 with stock fan, having removed the original paste and replaced with Arctic Silver 5, and since everything was working, I did a safe overclock: 333 FSB * 9 = 2.97 GHz, Mem 2:3 Corsair 8500C5 5-5-5-15 and it boots stable at the stock CPU voltage.

I then tried the usual tests (Intel TAT, Prime, etc...) and was getting 37-40 degrees on idle and over 70 degrees on Prime/ortho stress using CoreTemp, SnM, and TAT for temp measurements. I tried replacing the Arctic Silver twice following their recommended procedure, but this had no effect, and from the spread it looks like the paste was well distributed. However, while at worst the heatsink was slightly warm to touch, it was never very warm or even close to hot.

So I went back to the BIOS, set EVERYTHING back to default settings, rebooted and found that I was getting exactly the same thermal behavior even at default settings.

I've read many discussions on the Conroe thermals, and even refered to the Intel technical docs and the max temp at the max 65 Watts should top off at 60.1 degrees C.

When I opened the retail box, I did notice that the top of the CPU has some signs of use: slight glossy areas and one clear glossy area you can clearly see in the picture. I assumed this may have been from testing at the factory. I carefully inspected the box (I always do), and there seems to be no evidence of tampering. If there was, it's definitely a pro job.

On the other hand, maybe there's something wrong with the thermal reporting. If three different programs are reporting the same temps, if there is a problem, it could only be at the sensor, or its supporting circuitry. Other than trying the CPU on another mobo, or an older BIOS, I don't see how I could test for that.

So either I have a bum original chip, the thermal reporting is off, or this chip was abused and sold to me as new. I'll go back to the retailer and what he has to say.

Anyone have any ideas on this?

Breeze

 

[LuminoZ]

Feel Better, I got my e6600 up to 75C with the stock cooler, running good old Age of Empires 2. My motherboard's BiOS must have had a classical influence, that thing plays a creschendo of alarms! The best part is that its UNDERCLOCKED!

Same story, intel heatsink feels cool & I have a ton of air moving over it...
what gives?

I ordered a Sythe Infinity, it will get here wednesday... we will see if that helps, until now then... i'm frying eggs if anyone wants one.

 

[Breeze]

I took the time to reseat the HSF one more time, this time adding a dab of AS5 on either side of the recommended midline application, and since then it seems to have gotten better: abouit 34-36 idle and it maxes out at 63 with TAT; oddly it maxes out at 58 with orthos. Is TAT a better stresser than orthos?

The Intel spec for max temp specifies the top dead center of the heatsink as a temp reference. I've read that there's a difference between the on-chip diode temp and this temp, but until someone actually imbeds an accurate thermal sensor into the base of a HSF, we won't know what this difference really is.

But from what I've read, even chips with the same designation don't necessarily have the same power profile or Vcore. The Intel table on temps has this footnote:

"Thermal Design Power (TDP) should be used for processor thermal solution design targets. The TDP is not the maximum power that the processor can dissipate."

Which means that the specified 65 watts @ 60.1 C is not the maximum for this family; but I'm sure a lot of us would like to know how conservative this estimate is.

 

[Breeze]

Same story, intel heatsink feels cool & I have a ton of air moving over it...what gives?

I ordered a Sythe Infinity, it will get here wednesday... we will see if that helps, until now then... i'm frying eggs if anyone wants one.

Yeah, I think the AS5 is definitely making a difference, but a better HSF is good insurance. I'm thinking of same, but my video card is the one with no fan and a heatsink you can position over the CPU for cooling, so really tall and/or noisy cooling solutions are out. My Sonata II case doesn't allow for clearance over the motherboard "north" of the CPU, so I figure any of the big roudn Zalman's probably wouldn't fit.

Anyone have a good recommendation that would fit my bill?

TIA

 

[deathman20]

TAT max load temps (using 100%) is way way overkill on the max temp that you'll actually see. Orthos even is a slight overkill on max temp depending on programs that you use, but if you want a more accurate temp use Orthos over TAT loading.

Thats hot either way default or not.

 

[enz660]

TAT max load temps (using 100%) is way way overkill on the max temp that you'll actually see. Orthos even is a slight overkill on max temp depending on programs that you use, but if you want a more accurate temp use Orthos over TAT loading.


Thats hot either way default or not.

yeah, I stopped using TAT as my loader becuase it would never EVER get that hot by using any other program. I Dont understand though, do you or do you not have the bump on the cpu? If so, I would either RMA or lap it off ASAP.

 

[Breeze]

Do you or do you not have the bump on the cpu? If so, I would either RMA or lap it off ASAP.

It's not a bump, just a glossy area on the surface of the CPU, like through friction or polishing. Like I mentioned, it could be from the factory, and the package really didn't seem tampered with. Since other people here and elsewhere have confirmed these thermals, I've stopped worrying. Could be that they polish off irregularities of the cover at the factory.

Thanks.

 

[LuminoZ]

So im rather confused. The AS tech and a PDX lan awhile back told us to smear, and spread the AS all over the HS with a credit card. Yet, if you look at the directions on the AS website, they say to just use the line technique. Odd. Well im going to try the line technique with my infinity on wednesday.

 

[deathman20]

so im confused as to how your applying your AS5, breeze. I've been applying mine using the "as5 tech' method; Where you take run the as5 down the CPU in an S pattern, in about 7 horizontal lines. Then you use a credit card to evenly spread the AS5 over the cpu, excess goes in the middle.

I thinking about re-doing my CPU too... I want cold temps!

I've done AS5 many ways, best way I've done it is an cross in the center and 4 dots on the outer edges. I use to spread it but found that it increased my tempatures doing so.

 

[leojharris]

It's not a bump, just a glossy area on the surface of the CPU, like through friction or polishing. Like I mentioned, it could be from the factory, and the package really didn't seem tampered with. Since other people here and elsewhere have confirmed these thermals, I've stopped worrying. Could be that they polish off irregularities of the cover at the factory.

Thanks.

your cpu is most likely pretty concave. i'd recommend lapping both the cpu and heatsink and you should see a rather nice drop in temps. that 'polished' area is due to the fact that the edges of your cpu are higher than the center and therefore the edges recieve most of the contact/pressure from your heatsink block.

i lapped my 6400 a while back and the thing was like a radar dish ... *so* concave. i sanded for 20 minutes before the grit ever hit the center of the chip.

i've read a lot of lapping logs for 6400 and 6600's and almost across the board everyone had a rather concave surface.

if both your cpu and heatsink surfaces are completely flat ... then AS5 need only be applied in a rice sized kernal in the very middle; the pressure after seating your heatsink will absolutely disperse the AS5 to the extents of your block.

search around the forums for some good lapping guides.

 

[Mr. $T$]

Take your processor and a small piece of glass, put a dab of ketchup on the center of the IHS and press it against the glass. Looking through the glass on the other side look for a uniform color across the entire IHS. If you see a darker area in the center and lighter on the outside you have a concave IHS. If it's darker on the edges and light in the center you have a convex IHS.

I heard this somewhere on these forums so credit is due there.

 

[FIZZ3]

TAT max load temps (using 100%) is way way overkill on the max temp that you'll actually see. Orthos even is a slight overkill on max temp depending on programs that you use, but if you want a more accurate temp use Orthos over TAT loading.

Thats hot either way default or not.

There's something to be said for testing under the toughest conditions. Also remember that although for example a game will not immediately get the CPU very hot, the heat accumulating from all the components working together (videocard, ram, mainboard, even sound etc.) in your case may result in a steady rise in temperature.

TAT stabilizes at 60C after hours of load on my 6700. That's with the bone stock cooler, with the Intel thermal grease.

 

[hUMANbEATbOX]

TAT gives a theoretical "100%" workload. its not actually doing any work, its just firing up every part of the cpu to 100%. SO, th test is not really indicative of what a program that is actually doing work will load your cpu at. there isn't a program in the world that will use literally 100% of the transistors in the cpu like TAT does. so, IMO, TAT's "100% workload" is overkill, and is not a real world number. i would ignore it, and go by temps generated by orthos (but still use TAT to monitor the temps).

 

[LuminoZ]

i like the trick with the katchup! wheres the best place to get orthos? is motherboard monitor a good thing to have or is that an AMD thing still?

 

[leojharris]

i like the trick with the katchup! wheres the best place to get orthos? is motherboard monitor a good thing to have or is that an AMD thing still?

i like motherboard monitor ... lots of time-based detail. it shows the same temps as asus pc probe and to be honest, i usually go with my mb makers temp readings even considering the fact that TAT and Coretemp use a different approach.

i also like mb monitor cause it gives me constant temps in my system tray.

considering that your cpu should throttle past it's built in temp threshold ... any of the so far listed measuring apps will most likely be fine.

 

[LuminoZ]

I've seen mobo mon before and really liked the way it worked. The question now is if I should put my UGuru software back on... I re-formatted, thanks to creative drivers.

 

[FIZZ3]

TAT gives a theoretical "100%" workload. its not actually doing any work, its just firing up every part of the cpu to 100%. SO, th test is not really indicative of what a program that is actually doing work will load your cpu at. there isn't a program in the world that will use literally 100% of the transistors in the cpu like TAT does. so, IMO, TAT's "100% workload" is overkill, and is not a real world number. i would ignore it, and go by temps generated by orthos (but still use TAT to monitor the temps).

Let me try to explain the reasons for preferring the exhaustive test, even if its actual test is overkill for real-life situations.

Suppose TAT stresses CPU components A, B and C at the same time to 100% each. If this works fine, any program that stresses either A, or B or C will have no problem. Even a heavier program stressing A and C will still be fine. Or B and C... etc. Peace of mind.

Orthos may stress components A and B at the same time, and thereby present a real-life working load. However, component C may still be problematic, even if that's the only one being stressed by another program.

Ask yourself whether you want to run a PC that cracks under TAT load (overheats), but is fine in Prime, 3Dmark etc. I'd say that it's running too close to the limit to be comfortable. Now imagine dust buildup or summer coming in, and it may start overheating even sooner.

 

[deathman20]

Let me try to explain the reasons for preferring the exhaustive test, even if its actual test is overkill for real-life situations.

Suppose TAT stresses CPU components A, B and C at the same time to 100% each. If this works fine, any program that stresses either A, or B or C will have no problem. Even a heavier program stressing A and C will still be fine. Or B and C... etc. Peace of mind.

Orthos may stress components A and B at the same time, and thereby present a real-life working load. However, component C may still be problematic, even if that's the only one being stressed by another program.

Ask yourself whether you want to run a PC that cracks under TAT load (overheats), but is fine in Prime, 3Dmark etc. I'd say that it's running too close to the limit to be comfortable. Now imagine dust buildup or summer coming in, and it may start overheating even sooner.

While it might make the most heat, which we assume its stressing the CPU the most, does it truely give any feedback that its unstable or stable? My experiances, extremely unstable it locks up the system, semi unstable it continues all fine and dandy. I wouldn't call that an acceptable stress tester.

 

[FIZZ3]

While it might make the most heat, which we assume its stressing the CPU the most, does it truely give any feedback that its unstable or stable? My experiances, extremely unstable it locks up the system, semi unstable it continues all fine and dandy. I wouldn't call that an acceptable stress tester.

Yes, I would only use it for temperature-related concerns, not for troubleshooting errors. For that I recommend Memtest86+, Orthos prime and 3Dmark05. Not to say that there is no link between overheating and errors though.

 

[nikhsub1]

I've read many discussions on the Conroe thermals, and even refered to the Intel technical docs and the max temp at the max 65 Watts should top off at 60.1 degrees C.

60.1C spec is for Tcase, which is the max temp for the center top of the IHS, NOT THE MAX TEMP FOR THE CORE. Tjunction (which is core temp) is somewhere around 85C. Tcase can not be measured without milling the IHS and placing a thermal probe in the absolute center of the IHS.