My E6600 is over 80 degrees and driving me crazy...

Xschtar

Weaksauce
Joined
Jan 9, 2007
Messages
103
Hey,

I've written here before but things don't get better and at full load my CPU begins throttling because it's like over 80-82 degrees with TAT @ 3,15 GHz. I have a E6600 with a Scythe Ninja.

* I've remounted 27 times
* I've applied AS5 in 10 different ways
* I've grinded down parts of the Ninja pressing against kondensors.
* I've throuroughly lapped the IHS and the Ninja.
* I've tried 4 different temp programs.
* I've tried with the stock cooler, basically same temps.
* I've tried pressing down on the Ninja, very minor differences.
* Is it not an airflow issue. I have 7 fans and have also tried with open case.
* The cooler does not become hot at full load.
* No matter how many fans are blowing on the cooler (none, 1, 2 or 4) the temperatures are unaffected.
* At stock speeds it's stable with BIOS defaults, but overclocked it needs very high voltage to be stable.
* To check the flatness of the IHS and the cooler, I tried the razorblade/flashlight test and they are both flat.



And finally I reached idles of 52 and loads of 77 @ 3,15 GHz and I was very happy... Then I thought I could get even better temps so I lapped the Ninja some more and now I can barely get under 60 degrees idle...

Also, my E6600 requires crazy voltage to be stable. I need like 1,45v @ FSB 350 to be stable and this is way too high.
 
A "concave chip" is a chip with a warped IHS, though by the sound of it, you've pretty much nixed any possibility of it being such a chip.

1. Have you tried running this chip at a STOCK configuration, so that you know it's not a dud in the first place?

2. If so, does it require higher-than-nominal voltage to be stable at stock speeds?

3. The chip is pretty obviously running hot, but have you actually checked the heatsink by hand while it's running? If the heatsink is cold/lukewarm while the chip is blazing hot, that means you're not getting near enough heat transfer between the two.
 
You got a bad chip.

I have an e6600 running at 3.2 Ghz on standard voltage, idle 29C full load 39C using an ArcticCooling Freezer Pro 7 with a smidge of ArcticSilver.
 
Test your temps with you case door off. If there is a drastic decrease in temps, it could be your case's airflow (I had this problem with a crappy case).
 
That vCore, although is pretty high, wouldn't cause those temperatures alone. I'm guessing it's a combination of different things: bad chip, since it required more voltage than usual and a concave IHS, explaining the high temperatures.

Although, i would have thought that the dip would have been flattened out after lapping. Have you tried holding the processor against a flat surface (like a razor blade, for example) and checking if light passes through?
 
As I said, the IHS is lapped and no longer concave. The cooler does not become hot, not even at full load. And of course, I also understand that there is something wrong with the heat transfer. That knowledge hasn't helped my temperatures. So, to add to the list:

* Is it not an airflow issue. I have 7 fans and have also tried with open case.
* The cooler does not become hot at full load.
* No matter how many fans are blowing on the cooler (none, 1, 2 or 4) the temperatures are unaffected.
* At stock speeds it's stable with BIOS defaults, but overclocked it needs very high voltage to be stable.
* To check the flatness of the IHS and the cooler, I tried the razorblade/flashlight test and they are both flat.
 
is the mainboard temp high as well?

I am curious as it could just be the temp sensors in the CPU itself are off, it was known to happen on earlier weeks but could happen later on as well.

Also, try going back to stock like said earlier.
 
May actaully be that the chip heatspreader isn't making contact with the die. Maybe, though if it runs at ok temps at stock, I doubt it.

Another thing I'm wondering: How OLD is your AS5? If it's pretty old, it might've settled.

I would've said that the mainboard sensors are going crazy if it weren't for the fact that you need high voltages to keep the CPU stable.
 
The mainbord is around 38 degrees constant. The MB temp sensors don't seem to be off, as I've seen the CPU throttle at around 84 which is kind of normal. And I have reapplied AS5, brand new, around 20 times. Even tried a different tube of AS5. And as I said, I already tried going back to the stock cooler, to no avail.
 
I would've said that the mainboard sensors are going crazy if it weren't for the fact that you need high voltages to keep the CPU stable.

I need at least 1.4875v to be stable at anything over 3GHZ...yet my chip will idle@38C and load@60C when set to that (this is with a Zalman 9500 w/AS5).


I was meaning go back to stock settings for the processor (no OC).

Also, it seems like the MB is far from the CPU temp. My mainboard gets up to the 50s when my chip is loaded.
 
The cooler does not become hot at full load.

You seem to have been very diligent and knowledgable in making sure your mounting is correct. Assuming this as true

The cooler does not become hot at full load.

Its a defective cooler, Heatpipes are not working properly.

To verify can you get a temp probe or your finger on it at the very bottom on the plate that contacts the cpu ? If thats hot, and the rest of the HS or pipes are not, it would confirm. If necessary turn off a cpu HS fan. Anyway you see what I am getting at, think of a methodology in which you can verify the HS is working properly with heat being transferred from the CPU plate and up the pipes like it is susposed to.

I have seen another post like yours, LOL maybe it was an old one of yours but anyway, eliminate as a cause or confirm a defective HS, you have covered most everything else.
 
1. A defective cooler;

2. A defective cpu (perhaps the IHS not mounted right);

RMA both back to supplier, if you can.


And, I think observing/measuring the heat directly, as you did, is the best evidence of either of these problems. 80c is no good.
 
BillParish, that may be something... the heatpipes malfunctioning on the Ninja. That's the only thing I haven't tested, but it's not impossible. When I feel the bottom plate it's kind of hot, though the big part of the Ninja, with the fins and all, is very cool. To maybe the copper pipes haven't been filled in properly. It's possible, right?

Well, the only problem is that with the stock cooler my temps are also high. What are normal idle/load temperatures with stock cooler @ 3,15GHz, 1,45v, FSB 350?

I would hate to replace the cooler only to get the same problem again!

With stock cooler and default everything, nothing overclocked, just regular E6600 @ 2.4 GHz, the temps are 56 idle and 65 full load. Slightly less on the second core. This is way too hot and this is with stock cooler... so this would suggest there is nothing wrong with the heatpipes, right?
 
With stock cooler and default everything, nothing overclocked, just regular E6600 @ 2.4 GHz, the temps are 56 idle and 65 full load. Slightly less on the second core. This is way too hot and this is with stock cooler... so this would suggest there is nothing wrong with the heatpipes, right?

This suggests that something's wrong with the CPU not the HSF, then. What a pickle. I really don't know what to suggest, apart from getting a new CPU, maybe. Has the theory, that the sensors are just... off, been ruled out yet?
 
To maybe the copper pipes haven't been filled in properly. It's possible, right?

Thats exactly what I was thinking.

Well, the only problem is that with the stock cooler my temps are also high. What are normal idle/load temperatures with stock cooler @ 3,15GHz, 1,45v, FSB 350?

Pretty damn high, you got big cannonballs running that Vcore on stock :D I wouldnt even try it and cant guess other than real damn high.


With stock cooler and default everything, nothing overclocked, just regular E6600 @ 2.4 GHz, the temps are 56 idle and 65 full load. Slightly less on the second core. This is way too hot and this is with stock cooler...

That does seem a bit higher than I would have expected. But!!! if you have EIST enabled (or some other fan speed control in the bios) and the fan not forced onto full speed, you will notice that those temps conforms EXACTLY to intels recommended thermal performance curve. In other words, to help with noise, the fan runs just fast enough to stay on the performance heat curve intel has determined is optimal. So the temps will seem high if compared to other OCers that force the stock fan on Full Speed. (I am having to speculate here but you did say "default everything" and I think EIST on is the default but I dont have that exact board")


Here is my problem, being an engineer, even if an old over the hill one, when you have carefully eliminated all other possbile causes whatever is left must be the source of the issue. Works pretty good, of course not %100, but pretty darn good. You cant return the cpu and besides you lapped it. You did use a pane of glass as a flat surface when you lapped it ?

In short I think you have eliminated about everything BUT defective heat pipes in that heatsink.

That said

When I feel the bottom plate it's kind of hot, though the big part of the Ninja, with the fins and all, is very coo

This now worries me, that plate should be pretty hot at full load, but still it follows my theory, shame you cant get to the heatpipes with the temp probe of an indoor outdoor thermo.

Rats I see you had to grind on the HS.

Ok final answer, yes, my final answer :p

Unless you did the worst lap job in history, and I dont think you did, that HS is defective or (and again I dont think so) there is some mounting issue that is so simple you are constantly overlooking it, backplate shorting something, yada yada, again, I dont think so but thats the fun of trying to figure stuff out without being there. At worst, and money dont grow on trees for me either, get a new sythe or turning tower and you will know for sure.

Good luck man.
 
Is it possible that the motherboard is sending more volts than you've asked it to? I have an E6600 that needs some voltage to overclock and my temps are nothing like that. at 1.55v my cpu gets into the mid 60's at load. At stock it's more like mid 40's. This is with a Scythe Ninja w/AS5. Unlapped and, I'm pretty sure, not flat=/ When I first installed it I noticed that it appeared that the contacte between the chip and HS was slightly less than flat (I'll lap soon).

Anyway, it sounds like you've been thorough. I agree w/others that you can rule out a defective cooler since you've tried two, you can rule out a misshapen CPU since you've lapped it and tested w/razor blade, rule out AS5 since you've used multiple tubes, rule out case airflow since you've run it w/the side off (and mine doesn't run that hot even w/the ninja's fan OFF). It must be the motherboard or the processor itself.

As I hinted at above, my chip just hits 60 at load w/stock voltage if I have the fan turned off! When it's on, load temps are more like 40-45C.
 
After reading the whole thread, my money is on bad IHS contact. Did you ever run the chip unlapped, on the stock HSF?

Also, is it possible that while lapping, you dislodged the IHS? I've lapped chips before, but never thought about the possibility of dislodging the IHS.

Since you can't return the chip now because you lapped it, have you thought about going lidless? If you can't find anything else wrong, peel off the IHS and see what happens.
 
Thanks for all replies, and extra thanks to BillParish! I really appriciate the help. I know it's a tough problem but I've been posting on countless boards and the replies are mostly the same:

"You have airflow issues, you are not mounting properly, the IHS is concave, you are applying thermal paste incorrectly, etc etc" - all the classical stuff. So thanks for bringing a couple of new ideas in! I am gonna run a new round of temp tests since I've done some stuff since the last round.

Ok, with the Ninja:

@2.4 GHz

44 idle, 60 full load (not so bad? or?)

@3,15 GHz

59 idle, 81 full load (bad bad bad)


Just gonna pop in the stock cooler and make sure all fans are on full speed and give you the results! Brb!
 
With the stock cooler:

@2,4 GHz

40 idle, 53 full load (cooler than the ninja)

@3,15 Ghz

61 idle, 82 full loaf

Now the big question - are these normal temperatures?
 
Definitely not. I'm still going to have to go with bad IHS or bad IHS contact with the core. If the Ninja was bad, I think you would have seen lower temps with the Stock HSF and not very similar temps.

Again I ask. Had you ever run the stock un lapped HSF with un lapped CPU? As in, complete stock untouched configuration?

I'm just wondering if you dislodged the IHS during lapping or if it came like that.
 
Ok, hoping I catch ya before you get the ninja back on, and this is a bit (OK a LOT) crazy so use your own judgement. I was thinking that if you had a large cast iron frying pan (a big thick one), set the HS (take the fans off) in it and put the heat (hopefully gas) as low as it would go and see if the top of the middle of the HS got hot. perhaps put water in the skillet up to just below the bottom of the heat plate that the cpu makes direct contact with and place the HS to the side so its not directly under the flame. If the water starts to boil, its too hot. ?warm it up and turn off the heat. Actually put the water in, get it to boil, turn off the heat, then put the HS in as see if you can figure anything out.

this is probally the stupidest thing I have suggested someone other than myself to do, you dont want to cook the thing, just get the bottom warm enough to check the heat transfer of thing where you can kinda get at it and you dont want heat from the flame going while you are feeling around. You will probally burn yourself and anyone else in the house will now have proof we are nuts.

I am now wondering about a sub-par chip too, (think I was just ignoring that rare possibiliy) the IHS is held on by a slug of solder that is melted between the chip and the IHS. So you are not going to get the IHS off in any fairly safe manner although there are reports of it being done, mostly resulting in dead chips.

anywho, gonna go play my game and check back in later.
 
Ok, hoping I catch ya before you get the ninja back on, and this is a bit (OK a LOT) crazy so use your own judgement. I was thinking that if you had a large cast iron frying pan (a big thick one), set the HS (take the fans off) in it and put the heat (hopefully gas) as low as it would go and see if the top of the middle of the HS got hot. perhaps put water in the skillet up to just below the bottom of the heat plate that the cpu makes direct contact with and place the HS to the side so its not directly under the flame. If the water starts to boil, its too hot. ?warm it up and turn off the heat. Actually put the water in, get it to boil, turn off the heat, then put the HS in as see if you can figure anything out.

this is probally the stupidest thing I have suggested someone other than myself to do, you dont want to cook the thing, just get the bottom warm enough to check the heat transfer of thing where you can kinda get at it and you dont want heat from the flame going while you are feeling around. You will probally burn yourself.

I am now wondering about a sub-par chip too, (think I was just ignoring that rare possibiliy) the IHS is held on by a slug of solder that is melted between the chip and the IHS. So you are not going to get the IHS off in any fairly safe manner although there are reports of it being done, mostly resulting in dead chips.

anywho, gonna go play my game and check back in later.

That should work, only the high temps aren't typical to the scythe HSF, they are high with the stock HSF, too. Therefore, it must be something other than the HSF causing the problems; well, that's what i think anyway.
 
With the stock cooler:

@2,4 GHz

40 idle, 53 full load (cooler than the ninja)

@3,15 Ghz

61 idle, 82 full loaf


Ok, with the Ninja:

@2.4 GHz

44 idle, 60 full load (not so bad? or?)

@3,15 GHz

59 idle, 81 full load (bad bad bad)

Your stock cooler is BETTER at idle and almost keeping up overclocked!? There is no way your ninja should ever take a back seat to your stock heatsink. Sounds like getting a new HS is the next step to try. Even if your temps are high for both stock and Ninja, it sounds like the Ninja isn't working. It much harder to solve a problem when its actually two problems. Maybe its a combination of your Ninja not working correctly and something else... like your IHS is loose, your chip is runs way hotter than normal, or something else.

But whatever the other problem is (if there even is one) it sounds like the ninja is toast.
 
But whatever the other problem is (if there even is one) it sounds like the ninja is toast.

Surely, isn't the stock HSF then toast, too? Which backs up my point that the problem lies not on the HSF, but on the chip itself; or indeed something related to it such as the temp. sensors.
 
Lots of good feedback! Yes, yes! Sorry if I was unclear; DeadlyAura: yes I have run unlapped IHS on unlapped Ninja, default configuration, same problem. I have run basically all configurations imaginable and still the same problem. Maybe you think I messed up the C2D or the Ninja while lapping - I am pretty sure I didn't!

So these temps I am having now, 40 idle, are not normal? I think these are great temperatures. Never had an idle of 40 before. Its @2.4GHz of course, but still.

BillParrish: god damn that's a good plan! Why didn't I think of that? I am gonna go test it right now, we have ceramic electronic hi-tech plates in the kitchen... I am gonna put it on low heat and then test... then middle heat... hope the copper doesn't melt ;) Be right back with the results!
 
Dammit! Just put the Ninja on the stove and it heated up on a few seconds! The entire metal construction was hot (and I've never felt it getting so hot before, especially not the metal fins) and I now realize this is how the Ninja is supposed to feel when my CPU is 80 degrees.

So unfortunately, I think I just ruled out the heatsink as the problem. Dammit, I was just gonna order a new Zalman. So the CPU is broke... probably something I could never fix. And a new CPU is 300 dollars... or more.

Dammit.

So probably the problems are with the cores themselves, right? It can't possible be something else since I've made sure everything else works.
 
Yeah I know but I was kind of hoping for a malfunctioning Ninja... coolers are cheap, CPU:s are expensive ;)
 
Hmm, i think either your CPU is kaput, probably not getting a good contact with the IHS. Or, it's all just a non-issue and the temp. sensors are reading it wrong, but you did say you've tried many different programs.. urgh. Did you actually notice any throttling when 80 decrees C was being reported?
 
Ah, looks like that leaves you with the expensive problem, then. If only you could return it, since it was obviously broken when you got it- If it is indeed the CPU that's broken.
 
Actually I was in contact with my retailer and they refused to take it back. They said 70 degrees and more was perfectly OK and that I should call back if the CPU started malfunctioning at stock speeds.
 
the temp. sensors are reading it wrong,

That can be tested with a thermometer (I think). You'll need a fast response thermometer, not something that takes a minute to stabilize.

Get the CPU up to speed and "hot". Turn off the machine and quickly open up and remove the CPU from it's socket (good luck doing that fast!). You don't want the thing to cool down.

Apply the thermometer to the top side and check the temp. Then check the bottom side. See if the two sides are the same and are or are not hot.

Assuming the top side is not getting heat from the CPU chip to the top cover, then a cool top is expected. If the bottom side is hotter, then the chip is actually hot and you've demonstrated that the heat transfer is kaput. That should allow you to return the CPU to Intel as a manufacturing defect.

However - I don't know how well this test will go. I've never actually compared the top and bottom side temps and am only assuming the bottom will be hot (reflecting the actual chip temperature) if the top is cool due to a broken heat transfer. Anyone have experience with this???
 
Don't CPUs cool down in a matter of seconds? Very good test, theoretically, only i fear it will not be achievable. An alternative would be to get an aftermarket temp. monitor, and use that to check temperatures during load; although, it may be difficult to mount it close enough to the CPU.
 
Back
Top