OC temp question?

[maleficarus]

Ok I have a question about what is considered safe temps for my E8400 when overclocked. Using the temp program "Real Temp 3.4" when my chip is oc'd to 3.6Ghz I idle at around 52.0C give or take. It goes up a few and drops a few all the time. When I'm under full load for example using the "Intel Burn Test v.2.4" it goes up to 105.0C which in turn forces real temp 3.4 to go into alarm mode and throttles down my 3.6GHz OC to 2.3GHz. Is this too high? I would naturally assume it is at least under full load. I know 50.0C surfin and stuff is not hot for any Intel chip as I used to have a Prescott and that chip ran high 60's idle. I'm using stock Intel cooling fan in my Antec 4480B mini-tower case with one 120mm fan at the back sucking hot air out and one side intake fan drawing fresh air in.

Running stock speeds (3Ghz) I idle 40.0C and under full load jump to around 55.0C.

Can any of you [H] overclockers tell me what is considered ok and what is not?

 

[friend'scatdied]

Not ok! You should be at or under 65'C full load with a 45nm dual-core chip, and you want to stay there tops when overclocking. The stock Intel cooler should be doing a better job than that.. try re-seating it with some better thermal compound and/or check to make sure all the push pins are secure and locked.

 

[maleficarus]

Not ok! You should be at or under 65'C full load with a 45nm dual-core chip, and you want to stay there tops when overclocking. The stock Intel cooler should be doing a better job than that.. try re-seating it with some better thermal compound and/or check to make sure all the push pins are secure and locked.

Running stock speeds I am no problem. When I OC to 3.6 it jumps big time. Everything is set up ok, pins all in place fan spining at 1600ish RPM. Ok, just ran it again using stock speeds and under full load max of 55.0C but again this is stock 3Ghz speed.

The thing I like about my setup is you can't hear my case fans other then a slight hum. I can leave my system on all night and my desk is beside my bed and I won't lose a wink of sleep. I just don't want to have to run out and fill my case with loud ass fans just to overclock sigh...

 

[heatlesssun]

Not ok! You should be at or under 65'C full load with a 45nm dual-core chip, and you want to stay there tops when overclocking. The stock Intel cooler should be doing a better job than that.. try re-seating it with some better thermal compound and/or check to make sure all the push pins are secure and locked.

Yes, 105C is beyond the thermal limit. Now the question about what is "safe" below that is a bit debatable. 65C tops under full load is I think is going a little to far. The chip is designed to operate under 100C. Of course the lower the better and you probably want to keep it under 80C at full load. Much beyond that and yeah its time to get a bit concerned

Just remember, you're NEVER going to run a chip at full load 24x7 unless you're doing something way outside of normal beyond stress testing and I think people get a little paranoid about it.

Try to shoot for mid 30's at idle and 70 at full load. But 10C higher at both ends is fine really.

 

[heatlesssun]

Running stock speeds I am no problem. When I OC to 3.6 it jumps big time. Everything is set up ok, pins all in place fan spining at 1600ish RPM. Ok, just ran it again using stock speeds and under full load max of 55.0C but again this is stock 3Ghz speed.

The thing I like about my setup is you can't hear my case fans other then a slight hum. I can leave my system on all night and my desk is beside my bed and I won't lose a wink of sleep. I just don't want to have to run out and fill my case with loud ass fans just to overclock sigh...

Do you have the CPU voltage manually set?

 

[maleficarus]

Do you have the CPU voltage manually set?

Auto.

 

[maleficarus]

Yes, 105C is beyond the thermal limit. Now the question about what is "safe" below that is a bit debatable. 65C tops under full load is I think is going a little to far. The chip is designed to operate under 100C. Of course the lower the better and you probably want to keep it under 80C at full load. Much beyond that and yeah its time to get a bit concerned

Just remember, you're NEVER going to run a chip at full load 24x7 unless you're doing something way outside of normal beyond stress testing and I think people get a little paranoid about it.

Try to shoot for mid 30's at idle and 70 at full load. But 10C higher at both ends is fine really.

Yes, I understand full load is rare and not something that happends all that often. I was using Intel Burn Test and it says in the read-me that this is testing extreme. Right now just surfin the net I'm at 36.C mind you I'm bone stock 3GHz. My GPU is at 43.C which is always oc'd 24/7.

 

[BillParrish]

Some comments and facts, as heatlesssun stated, its all about what you consider safe to mean.

Target Tjmax for your processor is 100C. You need to make sure whatever reporting software you are using has this value for Tjmax. Most programs will allow you to change the Tjmax value.

When the Digital Temp Sensors inside the CPU core report 0, delta to Tjmax = 0, the DTS can be thought of as a "count down timer" in that as CPU temp goes up the DTS reports values that approach 0. The temp the software reports to you is a simple Tjmax - DTS value = core deg C. For example if the DTS is sending out a reading of 20 (actually -20) with a Tjmax of 100 and a DTS value of 20 the temp reported by Realtemp, Coretemp, Everest, Speedfan etc. will be 100 - 20 = 80C. Said another way the CPU is 20C below Tjmax or yet another way the Delta to TjMax is 20C.

At Tjmax the CPU will issue a PROCHOT# signal which will instruct the motherboard fan control and cooling system to spin the fans up to 100% and the CPU will start throttling to reduce heat. (if these options are enabled in the bios). At -1 (100 - 1 = 99C in our example) the CPU does not regard itself as hot and normal operation occurs.

Therefore one could regard 99C as the max safe temp your CPU can run 24x7x365 and be both under warranty and inside Intel's operating envelope.

When OCing more is involved than just temp. Increased CPU voltage is also normally used. It is a fact that electro-migration and transistor leakage currents increase as transistor junction temp increases. A combination of high operating temp and leakage current will likely reduce the ability of the transistor junction to operate properly when the junction is asked to perform faster than designed/speced for, in other words overclocked. Electro-migration will eventually reduce the performance of the transistor junction and if the added voltages are high enough the transistor junction can and will be damaged. This damage is an extreme case and knowledgeable OCers have provided a baseline of operation through experience and information provided by Intel data sheets that make damaging your CPU almost impossible if any common sense is used at all.

Several deg C above TjMax (Intel has not directly said but some Intel data on mobile processors indicate it is likely 10-20 deg C above TjMax, i.e. 110C -120C, {the previous is only my opinion from reading Intel documents and user testing reports, there is no Intel document for C2D desktops that state this exactly} there is a Catastrophic Thermal Trip point. When this temp, whatever it might be above and beyond TjMax, is reached the CPU will turn off its clocks and shutdown/crash to protect itself.

So that is the facts. You can run at about 99C forever if not overvolted or OCed. If OCed it is my opinion and one shared by many here that it is smart and prudent to provide a "cushion" to account for any extra voltage and increased frequency of operation to aid in long life and reliability.

I personally use a max reported core temp of 85C (assuming TjMax is correctly set) as a good compromise in achieving a good OC while still maintaining some thermal margin. OCers who like to go to the hairy edge will of course have there own higher number that they use as a personal "stopping point" and more conservative people may want a "full load" personal max operating temp lower.

But the fact remains that at 99C your CPU does not regard itself as hot. At 100C it does.

You are using the stock heatsink in a small enclosure so your alarming 105C at full load using IBT (which heats up a CPU like nothing else) does not surprise me. And it is cause for alarm as the program indicated and it is (I am pretty sure) not the program throttling the CPU but instead the CPU started to throttle and the program just reported that fact. All is working as intended.

So there you are.

OCing to high values with the stock Intel heatsink is not a very good idea. Pick your personal max operating temp, I agree with heatlesssun and recommend something between 70-90C depending on how hard you want to push it. An now that IBT has told you your cooling solution and OCing combination exceeds the operation envelope use something that more replicates your day to day max usage for temp testing. Do not forget your video card will add significant heat to the ambient case temp when in 3D mode so using programs that just do math problems might allow conditions that will result in freezing in a game as the extra heat in the case due to the video card on a warm day becomes an issue. Or continue to use IBT as a totally "worst case" scenario. It is your call. But clearly your cooling solution and OC as you reported are at the limits and worst case you are approaching conditions that are not good for the long term life of the system.

Improve your cooling, reduce your OC, or as mentioned above never run anything that stresses the CPU like IBT. You have to decide which combination, and what works for you. It could be as simple as cutting out a fan grill or disabling EIST in the bios to run the fan 100% all the time, if you can stand the noise. Everything is a trade off.

oh, Intel has also indicated that some 45nm CPU's like yours, the DTS is only accurate to -35 below TjMax so any reported temps below 100C - 35 = 65C are to be taken with a grain of salt due to limitations of the DTS. Still reported temps are useful as a guide and we all use them.

 

[maleficarus]

Thank you for that info. Here is a screen shot of my temps running stock and OC'd:

Oc'd:



Stock:

 

[maleficarus]

Finally to end my questions...

Can anyone recommend to me a CPU fan that is capable of cooling my E8400@3.6Ghz that dosen't sound like a small vacum?

I'm usually anti ocing simply for the fact that I cannot stand loud cases with several case fans. I would like to maintain 3.6Ghz 24/7 though. Thank you in advance!

 

[magoo]

If you don't mind a little work and a few dollars, buy this: http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=25881

very low noise.

 

[unclewebb]

Many Asus motherboards use WAY too much core voltage when you are overclocking and leave it set on AUTO in the bios. Post a screen shot of CPU-Z while overclocking. It will likely show way too much core voltage which is the main cause of too much heat.

It's not RealTemp slowing your CPU down. That's the CPU itself. Intel designs their CPUs to slow down and cool down when they get too hot. It should be easy to run this speed with less voltage than the AUTO setting is using. The huge jump in your temps at idle by overclocking 20% is an obvious sign of too much voltage.

 

[Battleneter2]

Many Asus motherboards use WAY too much core voltage when you are overclocking and leave it set on AUTO in the bios. Post a screen shot of CPU-Z while overclocking. It will likely show way too much core voltage which is the main cause of too much heat.

It's not RealTemp slowing your CPU down. That's the CPU itself. Intel designs their CPUs to slow down and cool down when they get too hot. It should be easy to run this speed with less voltage than the AUTO setting is using. The huge jump in your temps at idle by overclocking 20% is an obvious sign of too much voltage.

I can confirm Gigabyte also ramp the core and CPU voltages to high if you leave them on "Auto", agree manually set all esp vcore.

Throttling CPU speed is also a power saving option in the bios that should be disabled when overclocking.

 

[Zero82z]

Can anyone recommend to me a CPU fan that is capable of cooling my E8400@3.6Ghz that dosen't sound like a small vacum?

I'm usually anti ocing simply for the fact that I cannot stand loud cases with several case fans. I would like to maintain 3.6Ghz 24/7 though. Thank you in advance!

I thought you were against overclocking because it kills hardware. Or did you drop that particular aspect of your trolling act?

Anyway, this is the best bang for the buck cooler you can get right now, and would do a very good job of keeping your CPU cool without producing a lot of noise: http://www.newegg.com/Product/Product.aspx?Item=N82E16835103065