Core i7: Different core temperatures / overall temperature okay?

[binarykhaos]

Hello everyone.

Yesterday I installed a Core i7 860 w/ a Noctua D14 (w/ both fans installed) for the first time, so in this case I lack some experienced data and I'd would like to ask for some advice if the following is okay (good, average, bad?) or something to worry about. I read the core temperatures w/ coretemp by the way.

1. Core 1 is always about 6-7 C hotter than the other cores which pretty much stay together (within +/- 1 C). My old C2D E8400 had both cores always at the same temperature while at idle or full load.
2. I stressed w/ LinX (linpack) and core 1 hit 55-56 C while the other cores never crossed 49 C. The ambient room temperature was 25 C.

Thanks for any information in advance.

So long,
matthew

 

[Seero]

Hello everyone.

Yesterday I installed a Core i7 860 w/ a Noctua D14 (w/ both fans installed) for the first time, so in this case I lack some experienced data and I'd would like to ask for some advice if the following is okay (good, average, bad?) or something to worry about. I read the core temperatures w/ coretemp by the way.

1. Core 1 is always about 6-7 C hotter than the other cores which pretty much stay together (within +/- 1 C). My old C2D E8400 had both cores always at the same temperature while at idle or full load.
2. I stressed w/ LinX (linpack) and core 1 hit 55-56 C while the other cores never crossed 49 C. The ambient room temperature was 25 C.

Thanks for any information in advance.

So long,
matthew

The thermal pasting was not applied evenly. There is instruction on AS5 web page that visually show you where the Core is on the die, so you can apply them on the spot, evenly at least.

 

[Markfw]

I disagree with the above poster. Most likely its what people call a stuck sensor. At any rate the temps are not that much different, and way within norms IMO.

 

[Gillbot]

I have seen disparity in cores. Sometimes a TIM re-application can cure it, other times it's inconsistent pressure from the heatsink attachment mechanism or it can also be a stuck or laggy sensor. Rarely is it any cause for real concern.

 

[lsv]

What ^ he said, my i7 is a little hotter on the cores where there's less pressure because my HSF is fucking massive and drags down a bit. Temps are within 5c so I let it go and am happy with my stability.

 

[bob5568]

For the record, my cores 1 and 4 are the hottest. I doubt re-applying tim will change that. Its been a consistent result over the use of three different hsf/tim combos. As others, the difference is about 4 or 5 degrees. I've set the 24/7 config so that 100% loads remain in the 60s, so I'm happy.

 

[Tsavo]

My E3200's temps are off, too. 42 on one core at idle and 33 on the other.

They even out under load though.

 

[aigomorla]

I have seen disparity in cores. Sometimes a TIM re-application can cure it, other times it's inconsistent pressure from the heatsink attachment mechanism or it can also be a stuck or laggy sensor. Rarely is it any cause for real concern.

+1

however i think it could be a stuck sensor.

You can tell really fast when you pull that sink off and look at the tim spread on the cpu.

 

[Tsavo]

How does a sensor become stuck?

 

[aigomorla]

How does a sensor become stuck?

factory..

There is no way that i know of that would unstick it.

 

[binarykhaos]

Thanks a lot for all the replies.

The general problem with core temperatures is that it is hard to find any data to compare to. Some people measure their temps while the case is open or even with the board out in the open which is not really representative for every day usage. Usually you don't even know those details when you read the temps.

My problem currently is figuring out if those temps are good or not... rank them to see where I stand with regard to the summer and the overall functional efficiency of my cooler. I have seen people who overclock their i7 860 and have the same or slightly lower temps with worse CPU coolers which would hint that my temps are at the upper end of the ok scale or they simply "cheated" by having the board out in the open. I mean the D14 is one of the or even the best cooler you can currently get and if the temps are just slightly ok there has to be something wrong.

Is there a way I could test sensor 1 to see if it is really a stuck sensor? I would really like to avoid taking a look at the thermal paste for now because it's always a mess to clean things up and I am pretty sure it was about the right amount I put on it (pea method)... never had any problems with it so far.

One thing I could check is if I have tightened the two heatsink screws enough... "until they stop" leaves some room for interpretation. It could either mean, tighten it as good as possible or simply tighten it until it gets harder. Maybe the pressure is not (equal) enough. I was afraid I would break the socket or cpu if I tightened it until no further tightening by hand was possible.

Thanks again for everyone's patience. I really appreciate it...

So long,
matthew

 

[qazwsxokmijn]

From what I've been told Core 0 is almost always the hottest core because it's the core that is utilised first or the most, hence the heat. My i7's Core 0 idles up to 8 degrees hotter than core 3, although it seems to be stuck at anything lower than 40, and under full P95 load up to 11 degrees hotter than core 3. Cores 1 and 2 and always around the same.

 

[binarykhaos]

So, I tightened the heatsink screws a bit more until they could not be moved any more by hand. The result: my temperatures for the other cores increased a bit (approx. +2C), core 1 stayed the same and thus the temperature difference is down to a max. of 5C. Not exactly what I was aiming for but nevertheless an interesting result. I guess I can just live with it...

I have done some research and I guess my overall temperatures are in the upper range of excellent if I get things right. I have seen several reviews for this chip and usually they ended up w/ temperatures above 60C w/ Prime95 and a good 3rd party cooler while at around the same ambient temperature. So I get a max. of 56C (Core 1) respective 50-51C on the other cores w/ LinX which loads the CPU even more. So I guess I can be satisfied.

Thanks for all the helpful and kind replies by the way!

So long,
matthew

 

[Makaveli]

I have also noticed this here is my idle.



I was thinking maybe I bolted down the TRUE to much to one side of the core and will try to remount at a later date. Not really a big deal as my temps are fine.

 

[tweakboy]

This is why people lap their CPU ,

Its called lapping the CPU, which means they take a grit 5 sand paper, and keep sanding the CPU surface until lapping is achieved and its smooth and even. Now temps will be the same.

I have not lapped my CPU and I wish I did. Its 8c more then the other ..

If you google in, lapping cpu youll know. But I believe people use grit 5 sand paper.. not sure.. anyhow. Its normal man and I know it sux! , but it is what it is. the processers are not made perferctly even. This is where thermal paste comes in to fill the gap. Your temps are normal and Intel says in 60's is optimal. For idle in the 40's are optimal. Your CPU will live until you kill it hehe. your temps are normal for your setup.

If your load temps are in 60's or below then your optimal and safe. gl

 

[Makaveli]

I was under the impression most would lap the base of the HSF not the cpu itself.
Then again I can see the point of doing both.

 

[Gillbot]

I was under the impression most would lap the base of the HSF not the cpu itself.
Then again I can see the point of doing both.

IMHO, if you are going to the trouble of lapping, you NEED to do both.