Cpu Overclocking Temps

The Badger · 14 Jan 2008 at 15:54

Hello, I have an E2180 in a P35C-DS3R (f10 bios). I currently have it running at 3ghz by using 300 x 10 multi and 1.44V core! My temps are threw the roof at 75oC (underload). I have trouble overclocking because of these temps and i wish to overclock further. To cool my processor i use the ACF7. A decent fan but not the greatest.

Should I invest in a new fan to lower my temps? Can anyone help me with my bio's settings? Or suggest anything i can do to lower the temperatures?

Many Thanks,
Paul.

iKettles · 14 Jan 2008 at 16:01

If you're not happy with those temps then maybe lapping is an option. Other than that the only choice is to get a new CPU cooler. If you're not happy with those temps drop your clock down to a voltage which doesn't give you that high temperature. 200-300mhz isn't going to make a big change.

dovey · 14 Jan 2008 at 16:46

Would be worth reseating the heatsinc and fan to see if there is any difference. I'm using the intel stock cooler and ger 62oC at 3.3 ghz with 1.4 volts

maf718 · 14 Jan 2008 at 18:02

Assuming you're using the latest version of coretemp I wouldn't worry too much, that temp is still 25°c below the shut off point for that chip. Well within safe margins, especially considering your vcore is quite high.

To test whether your case airflow is up to scratch, take your side panel off and run orthos for a couple of hours to see if that makes a difference to your load temps. My E2140 drops 3°c when I do this, I'm considering getting another case fan.

The Badger · 14 Jan 2008 at 18:16

I have already reset it and no change. Dovey what settings have you got your chip on may i ask? 3x330? and what Power Settings?

dovey · 14 Jan 2008 at 18:34

iKettles · 14 Jan 2008 at 18:42

maf718 said:
Assuming you're using the latest version of coretemp I wouldn't worry too much, that temp is still 25°c below the shut off point for that chip. Well within safe margins, especially considering your vcore is quite high.

To test whether your case airflow is up to scratch, take your side panel off and run orthos for a couple of hours to see if that makes a difference to your load temps. My E2140 drops 3°c when I do this, I'm considering getting another case fan.

That suggests there's a lack of airflow in your case.

maf718 · 14 Jan 2008 at 19:03

iKettles said:
That suggests there's a lack of airflow in your case.

Yeah, the exhaust is pretty good with an akasa amber 120mm, the psu fan and my graphics card exhausts outside the case. Only one 92mm intake at front though, there's space for another which I might take advantage of if I can be bothered. Can't see it making more than a degree or two difference to cpu temps and the only other option is a beefier 120mm to replace the Akasa which will be too noisy for me - I overclock but I'm not extreme.

Richie P · 14 Jan 2008 at 19:15

maf718 said:
Assuming you're using the latest version of coretemp I wouldn't worry too much, that temp is still 25°c below the shut off point for that chip. Well within safe margins, especially considering your vcore is quite high.

To test whether your case airflow is up to scratch, take your side panel off and run orthos for a couple of hours to see if that makes a difference to your load temps. My E2140 drops 3°c when I do this, I'm considering getting another case fan.

100% Wrong. Please don't give advice if you don't know what you're talking about.

He's already over the chips specified max temperature, plus with an increased vcore, he's not gonna be doing the chip any favours. That being said, that's what overclocking is all about and we all know these chips have long lifespans that we're not gonna see out. 73C is a high core temperature, and you wont find anyone else recommending this as a day-to-day 'safe' temp, let alone 25C extra headroom like you're suggesting.

Pauliee, I would suggest trying to lower your vcore. Drop down to 1.4Ghz and see if it's Orthos stable for 6 hours at that, then start going down in smaller steps if it's still stable. Find the point at which Orthos errors and then increase the vcore one or two steps from there to be safe. vcore is by far the biggest contributor to core temperatures.

Info: Intel's Processor spec finder:
http://processorfinder.intel.com/details.aspx?sSpec=SLA8Y

dr. sarf · 14 Jan 2008 at 19:19

Do you really need that voltage for that clock?
You should be able to drop the voltage a bit and it still be stable at 3Ghz??

maf718 · 14 Jan 2008 at 23:42

Richie P said:
100% Wrong. Please don't give advice if you don't know what you're talking about.

He's already over the chips specified max temperature, plus with an increased vcore, he's not gonna be doing the chip any favours. That being said, that's what overclocking is all about and we all know these chips have long lifespans that we're not gonna see out. 73C is a high core temperature, and you wont find anyone else recommending this as a day-to-day 'safe' temp, let alone 25C extra headroom like you're suggesting.

Um 100% right. Coretemp only measures temperature as a delta from the tjunction max(ie the core temperature at which the processor will shut itself down to avoid damage)read from the diode in the processor core. 75° on a 2180 in coretemp 0.96 is 25° below the tj.max. I'm not suggesting Pauliee has 25° to play with, just that its ok running at that level. If he is using an earlier version of coretemp (or some other program) and is getting 75° as a core temperature then that would be worrying. If you have a M0 stepping 21xx processor try running coretemp versions 0.96 and 0.95.4 side by side and the newer one will give temps exactly 15° higher because it assumes the tjunction max. is 100°c rather than 85.

I regard 25°c below the maximum under full load to be entirely safe and I don't think you'll find many who disagree.

As to your advice to Pauliee to try and reduce his vcore levels that is entirely sensible and I agree with you.

Rroff · 15 Jan 2008 at 00:24

I personally dislike runnig the cores over 60C - maybe peaking at 65C or so during full on Orthos... Intel's specs on the Core 2 line are similiar IIRC above 71C with most core 2 CPUs intel will not guarantee "normal" operation.

maf718 · 15 Jan 2008 at 01:10

Intel's thermal spec of 73°c on the M0 stepping 21xx processors refers to the tcase, ie the temperature on the top of the processor's integrated heat spreader.

This is likely to be about 15°c below the actual core temperature, so for a coretemp 0.96 reading of 75°c its gonna be around 60.

Safe I'm sure.

Richie P · 15 Jan 2008 at 01:42

maf718 said:
Intel's thermal spec of 73°c on the M0 stepping 21xx processors refers to the tcase, ie the temperature on the top of the processor's integrated heat spreader.

This is likely to be about 15°c below the actual core temperature, so for a coretemp 0.96 reading of 75°c its gonna be around 60.

Safe I'm sure.

Where did he say he was using Coretemp, let alone what version?

You were completely wrong in giving the initial advice,

that temp is still 25°c below the shut off point for that chip. Well within safe margins, especially considering your vcore is quite high.

And you're still giving incorrect advice now. I suggest you start using facts to justify your opinion in order toback up the advice you're giving.

Without knowing what program he's using to monitor those temps, and if Coretemp then what version he's using, we can only say that his temps appear to be on the high side.

The Intel spec says not to go over 73.2C as a maximum operating temperature. No special +/- or anything, just 73.2C max temp. As long as the monitoring software is correct, he's already over than limit.

I wouldn't ever recommend a load temperature over 70C for 24/7 use, and my personal maximum operating temperature is 60-65C, especially at increased volts.

Please don't take my comments the wrong way, but bad advice helps no-one and suggesting [email protected] is "well within safe margins" is bad advice.

fornowagain · 15 Jan 2008 at 04:03

Richie P said:
The Intel spec says not to go over 73.2C as a maximum operating temperature. No special +/- or anything, just 73.2C max temp. As long as the monitoring software is correct, he's already over than limit.

The Intel design spec is referring to the case temp Tc, this is the external IHS temperture measured via thermistor mounted in the center of the heatspreader.

The junction temperatures reported by coretemp et al are taken from DTS values with a -ve towards the chips thermal control throttling point TCC. The value of Tjmax is an assumed value and not the actual TCC point as this is unknown.

Core Temperature = Tjunction - DTS

If he is reading 75C from a thermistor or a on die diode approximation of cpu temperature then that is very high. If however the delta values using coretemp says its 25C delta (till zero) it is within the factory set safety margin for the chip.

Assuming EIST etc hasn't been disabled the chip will modulate/step (throttle) to avoid damage using control signals TM1 & TM2. If temps still go to far there is an additional shutdown safeguard THERMTRIP# designed for fan failure etc.

http://download.intel.com/design/processor/datashts/31698104.pdf

Assuming the program you use has the correct value for Tjunctionmax (approximation of TCC), a 75C coretemp will have a lower offset casetemp of at least 10-15C. Placing it below the maximum working case temperature of 73.3C. However the ONLY value that is truly accurate at higher temps is the DTS delta value to the throttle point. A 25C delta is a little hot for an overclocked chip, i.e. poor stability and I wouldn't run it that hot, but no it won't damage the chip. Overvoltage on the other hand will in the long term degrade the chip.

maf718 · 15 Jan 2008 at 16:25

Well explained fornowagain, thats what I was trying to say all along.

The original poster is not going to fry his chip at a 25°c delta to Tjunction max. That's why I said that if (and only if) the readings were from the latest version of coretemp then they were safe.

Before christmas many people were posting temps of 60° under load for M0 stepping E21xx cpus as measured in the then current version of coretemp, and no-one was worrying about it. That is exactly the same temperature as 75° with coretemp 0.96, ie 25°c below the Tjunction max.

I personally run an overclocked 2140 at near stock volts and am perfectly happy with a delta to the Tjunction max of 51°c at idle and 28°c at full orthos load - its rock solid at those temps.

Richie P · 15 Jan 2008 at 18:30

Thanks fornowagain, I stand corrected on the max temp issue, thanks for providing a thorough explanation.

Maf, that was not what I understood you were implying, even if that was what you were intending.

As fornowagain highlighted...

If he is reading 75C from a thermistor or a on die diode approximation of cpu temperature then that is very high. If however the delta values using coretemp says its 25C delta (till zero) it is within the factory set safety margin for the chip.

That is the most important point. With Pauliee posting in this thread again, we have nothing left to discuss.

maf718 · 16 Jan 2008 at 02:50

Agreed, Pauliee seems to have done a runner from his own thread so its all a bit irrelevant. He didn't even let us know what app. he was using for his temps.......