YOUR safe operating temperature?

[Razorfist]

I just wanted to see what everyone thinks about temps that are safe for their CPU

Either make a post with your CPU type (any other info you want too, such as HSF, mobo etc) but mostly it is which CPU that is important.

I am running an e8400 at 60C full LOAD.

(but im not really sure, coretemp says 60C with tjmax of 105, realtemp says 50C with tjmax of 95. speed fan appears to have a tjmax of 100 as it gives me around 55C. asus motherboard says it runs at 30C idle, usually i witness 10C-20C increase going from idle to load, so 50C would make the most sense. Im afraid to push my *guess* of 50C to 70C because if its wrong and coretemp is right it is actually 80C.

For the poll, please answer at FULL LOAD. as in 100% of all cores

Also, if you have anything unique to say about temps and CPUs please do

does anyone have any links to official documentation for safe operation of e8400s? they probably give out a real low number to cover their asses

 

[Tempered81]

75-77C max [coretemp] 65-67C [RT] for me!

 

[Drsignguy]

55c - 59c max with my Q6600 @ load but at normal use is around 30c - 45c. My E7200 is a bit lower.

 

[rge]

There is no specs on core temperature (diodes on cores) for desktop cpus. There is a Tcase max which is the temp at geometric center IHS, and is 72.4C max for E8400. Although no sensor reads Tcase, the cpu temperature (diode between cores read by speedfan, etc) is the closest thing. And if cpu temp is less than that max, intel considers it to be within specs, which gives an additional safety margin as Tcase (located across tim1 and IHS from cpu temp) will always be lower than cpu temp. I tried to get intel to say that clearly, but for amusement, here is my emails with them. My apologies to Razorfist for long email string. Though personally, I am more with Jaredpace on what temps I use.

To intel customer support:

The thermal limit of my E8400 is 72C Tcase (from spec sheet).
Tcase is measured at the geometric center of the IHS, where no diode
sensor is located (from data sheet).

CPU temp is from a diode between the cores. So although this is not
really Tcase (geometric center of IHS), is this the temp we are supposed
to monitor to be within spec, since cpu temp will be higher than Tcase??

But in the data sheet it states..."PROCHOT# is designed to assert at or a
few degrees higher than maximum TCASE". So does that mean if thermal
margin or delta to TCC is not reached we are still within spec? ...but I
guess that would be providing we have not exceeded Tcase at that point,
but is it improbable to exceed Tcase without reaching TCC??

It is exceedingly difficult to determine what is and what is not "in
thermal specs" from reading spec sheets, any help would be appreciated.

thanks,
bob

________________________________

Hello Bob,

Thank you for contacting Intel(R) Customer Support.

I understand that you are concerned about the operating temperature of the
Intel(R) Core(TM) 2 Duo processor E8400.

The maximum operating temperature of the Intel(R) Core(TM) 2 Duo processor
E8400 is 72.4 degrees Celsius. As long as the processor is operating under
this temperature it is operating within specifications. We do not have a
normal operating temperature for the processor as this temperature will
vary depending on the chassis and other hardware installed on the system
as well as the actual load the software is placing on the processor.

To verify the thermal information for this processor please visit the
following website:
http://processorfinder.intel.c...0&sSpec=slapl&OrdCode=

There is always going to be a 5-to-10-degree difference between cores
because there is a main core, which is going to have the constant load of
the system (even when this is at idle conditions) and a secondary core
that the motherboard and the operating system would determine when to
function.
I would recommend updating the system BIOS to the latest revision
available.

Sincerely,

Adolfo S.
Intel(R) Customer Support

______________________________________

Thanks for the quick reply. Unfortunately it did not pertain to my
question. My bios is updated and supports the E8400.

My question is what does the 72.4C degrees maximum pertain to...is that the
max for the cpu temperature (diode between the cores) or Tcase (measured at
geometric center of the IHS). According to intel spec sheets it is for
Tcase, where there is no sensor. If it is for Tcase, is it that Tcase
should always be lower than cpu temperature since it is closer to cooling
sink, and thus cpu temp is OK to use as an approximation? Or should the cpu
temp itself not exceed 72.4, even though Tcase would likely be lower than
72.4 on load if cpu diode was 72.4.

Thanks,
bob

<===========>

Hello Bob,

Thank you for contacting Intel(R) Customer Support.

Since we do not have Tcase thermal information, to obtain the
information that you are asking for, you will need to contact one of our authorized
distributors. You will need to work with a Field Application Engineer
(FAE) at one of our distributors. A list of our distributors can be
found at the following web site:
http://www.intel.com/cd/channe.../asmo-na/eng/46168.htm

Once you get to this site, select your location, then select from a list
of our distributors to find one located in your area. If you are outside
the United States, please access the following link and select a
location,
to find the closest authorized distributor:

Sincerely,

Adolfo S.
Intel(R) Customer Support
<===========>

Adolfo,

Thank you again for the quick reply. Unfortunately I did not purchase
anything from any of those companies, and I have no chance of getting to
speak with an FAE. Can you not tell me what the thermal specs of 72.4C
max of E8400 pertain to? Is 72.4C the max temp of "cpu temp" read from
diode between the cores? Or does 72.4C refer to max temp of Tcase...but cpu
temp is ok approximation? or other?

thanks,

bob

<===========>


Hello Bob,

Thank you for contacting Intel(R) Customer Support.

72.4 degrees Celsius is the maximum recommended temperature for the CPU
read from the diode.

Sincerely,

Adolfo S.
Intel(R) Customer Support

<===========>

<===== ======>

Adolfo,

Thanks again for the quick reply. I think what confuses me and most
people is the definition for thermal specs on the intel site.
If you go to the link for specs,
http://processorfinder.intel.c...ails.aspx?sSpec=SLAPL# and then click
on thermal specification for example E8400 72.4C, it reads...
"Thermal Specification: The thermal specification shown is the maximum
case temperature at the maximum Thermal Design Power (TDP) value for that
processor. It is measured at the geometric center on the topside of the
processor integrated heat spreader."

So according to the spec sheet 72.4C is for Tcase, but others have said
that intel verbally states it is the cpu diode temp, which is why I wanted to
ask myself, who knows the accuracy from what you hear second hand.

Any enlightenment as to the discrepancy though?

thanks,

bob

<===== ======>

<===== ======>

Hello Bob,

Thank you for contacting Intel(R) Customer Support.

The only way for you to have the accurate data is by contacting a Field
Application Engineer (FAE) as mentioned on previous emails:

Please contact one of our authorized distributors. You will need to work
with a Field Application Engineer (FAE) at one of our distributors. A list
of our distributors can be found at the following web site:

http://www.intel.com/cd/channe.../asmo-na/eng/46168.htm

Sincerely,

Adolfo S.
Intel(R) Customer Support

<===== ======>

 

[n7]

rofl :laugh:

What a nice little loop they have going there.

 

[TC91]

Ideally for me 60c or less is good, but 65c would be absolute max for me (coretemp + small fft's). Right now I'm at 57c dual small fft's.

 

[Duvie]

I am hoping for 4ghz and under 50c.....That is with my modified water cooler setup....Should now by this friday when all of my stuff will be here...

When my water cooler (not modified was working correctly it held my qx6700@3.2ghz at 49-51c load......That was with a 130watt cpu though, so I am hoping for much better eventhough I will need some voltage to obtain it....

 

[myocardia]

Originally posted by: TC91
Ideally for me 60c or less is good, but 65c would be absolute max for me (coretemp + small fft's).

Yeah, that's exactly how I feel, also. But then again, I hear that I'm anal about temps, so higher is most likely fine.

edit: BTW, Razorfist, you need to clarify in your OP whether your poll pertains to CPU temp, or core temps.

 

[Tempered81]

nice post rge. I think that TCase of 73C [72.4C] on the E8400 would be the equivalent of about a 73C to 74C core temperature. The gradient from core heat to the die center under-side of the IHS is minimal correct? Uncleweb's measurements with a digital thermometer to the top-side of the IHS would be pretty close to TCase, possibly 1 degree less. So according to [Realtemp?], you would not want your cores running hotter than 73-74C for the long term (DC or folding as an example)... or 83-84C coretemp. Now, on the other hand: Assuming there is still a huge 10-15C gradient between Tjunction & Tcase, you would want your cores running somewhere around 60C Realtemp / 70C Coretemp maximum.

Keep on pressing! I choose a delta to TJmaximum of > 28 to 30C for E8400's.

My guess is the safe # lies somewhere between 60C & 80C

 

[n7]

< 80C max.

Ideally < 70C.

 

[rge]

The gradient from core to tcase is dependent on many factors, but intel publishes a tested formula, which they state is very accurate.

The gradient from core temp to Tcase (IHS temp) at FULL TDP LOAD is roughly 23C, if all parameters of testing is exactly same as intel including intel stock cooler, and if tjunctionmax is 95C. This can be worked out from the formula,
Tjmax - Tcasemax = theta(coretocase)*TDP (intel published formula), putting in numbers 95C - 72.4C= gradient 23C or can be worked out from .35c/w * 65W = gradient 23C.
(note if tjunctionmax is 105, then this full load core to case gradient is 33C, and thermal resistance is .5c/w)
This is confirmed by intel stating prochot is set to assert (core junction max reached) when tcase max is reached.

The gradient from core temp to Tcase at idle, undervolted underclocked state is .35c/w * 4W = 1.2C. This is why all testing must be done under idle underclocked, undervolted state, where intels formula demonstrates a 1-2C max gradient from core to case. But clearly a highly variable, and max ~23C gradient exists under intel set conditions at full TDP load. However as both core temps and cpu temps are both on die substrate, one on core, one between cores, where the conductance is very high, you will never see this gradient, unless you place a sensor on IHS, which is across tim1, which has a much higher resistance and is where the gradient primarily occurs.

Though this gradient depends on many factors, one of the biggest variables is TDP. While this may lead one think they can run their cpu at 94C core temp, which might be 71C tcase, that will be in error, as intel states this gradient is highly variable on type of load, cooling, etc. Since this gradient is highly variable and unpredictable, and it is not practical or accurate for intel to put a sensor on the casing, we are stuck with cpu temps/core temps and assuming worst case scenario, that the gradient may be small at the particular load, hence use cpu and core temps as IHS spec max temps.

But as jaredpace points out, if you keep delta to tjmax at 25-30C, you know you are within specs as your safety margin exceeds any possible gradient, you know you are accurate, and you can ignore all theories and guesses.

I went through math and references here:
http://www.xtremesystems.org/f...=2962586&postcount=916

 

[Drsignguy]

Nice work "rge" :thumbsup:

 

[Tempered81]

Well we have VirtualLarry here folding with his E2140 Core2duos and their temps are reaching the 80-90 range. I think he just recently posted he was running strong 24/7 in the summer heat with a 9C delta to TjMax. Crazy if you ask me... but would this mean that hes still lower than his rated Tcase Maximum since the processors aren't initiating the Processor-Hot throttling modes?

IMO, someone should author a program that shows a visual thermometer graphic. The prog could recognize cpu types/revisions & designate the concurrent Tjmax/DTS/Delta/TCase etc then relay that information in an easy to see visual.

Kind of like a coretemp/realtemp that shows cpu temp/throttling in the form of an internet speed bandwidth "speedometer" display.

I made some of these suggestions at XS in the past:
http://www.xtremesystems.org/f...=2819496&postcount=252

http://www.xtremesystems.org/f...=2867708&postcount=505

Also rge clearly knows a lot about this, his guess on "Safest #" is as good as anyone's.

Btw, if you run an E8400 too hot and PROCHOT initiates at 92.4C, does that mean the gradient was 20C at that point, and that PROCHOT is dependent on TCase maximum = 72.4C, and also that Tjmaximum cannot be determined as an actual value (it just happened to be 92.4C) because it fluctuates depending on TDP at any given temp/vcore/cooling scenario? Maybe this is the reason intel doesn't tell us Tjmax, only Tcase max?

 

[StinkyPinky]

The other thing is how realistic is Prime 95 anyway? I certainly never put my CPU under the stress that Prime does. so if you get 75c in CoreTemp with Prime, does it mean you're going to get that playing a game for example? Not in my experience since for gaming the GPU takes a chunk of the load.

 

[rge]

Some E2140's at idle, undervolted have IR reading of 85C suggesting 85tjmax , coretemp may have that one wrong as it uses 100tjmax, so when he is at 9C from tjmax he may be at 76C core temp according to real temp. Though E2140 have two separate Tcase max values, one ~60C one ~70C, depending on the stepping, and the waters are very muddy on that one.

TCC as intel calls it (we call it tjunctionmax or tjmax) is a set temperature for a given cpu. There is some defined, set in stone, absolute temp that each cpu throttles. In the paper, http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf, intel states some loading programs have a 0C gradient from core temp to cpu temp, others have max 5C gradient. The gradient from core to IHS is widely variable as well, but who knows from ? to 23??? or 30?? The gradient will fluctuate based on TDP, and load, and many other factors, but actual TCC or Tjmax is a set number. While TCC could vary a few degrees C from cpu to cpu, even within same family from calibration, what argues against that is people getting the exact same IHS temps at DTS=0 when testing the exact same stepping cpu. Suggesting that tjmax or TCC is same for example for all E8400 of same stepping.

When you run my, unclewebb's or many others that have tested E8400 to DTS=0, which is prochot assertion, the IHS measure 95C on all that have tested. Suggesting TCC is constant on all. Only question is small gradient error at this undervolted, underclocked state, where temp is high only because no cooling.

When you use a high load and cooling, and you hit 92C, you might have a 20C gradient and be within specs, or you might have a 10C gradient and be way out of specs. You only know the gradient at two times. One, at idle, undervolted state. And two, (actually only intel knows this one), whatever punishing TAT like program is used to fully load cpu, full TDP, under worst testing specs.

But the fluctuating, undefined gradient as you state, I am sure is one reason intel does not want us using core temps, and hence does not publish tjmax. Another is at idle, core temps are almost random number generators due to non-linear, ie no calibration, at low end.

Regarding you ideas on using delta to tjmax, you would make intel engineers happy, they have been suggesting that to coretemps author, everytime he inquires about temps. I think the problem is the masses wont go that route, just like USA can not adopt the metric system, even though it makes more sense.

 

[VirtualLarry]

Originally posted by: jaredpace
Well we have VirtualLarry here folding with his E2140 Core2duos and their temps are reaching the 80-90 range. I think he just recently posted he was running strong 24/7 in the summer heat with a 9C delta to TjMax. Crazy if you ask me... but would this mean that hes still lower than his rated Tcase Maximum since the processors aren't initiating the Processor-Hot throttling modes?

As far as I know, I'm not throttling.

Originally posted by: jaredpace
IMO, someone should author a program that shows a visual thermometer graphic. The prog could recognize cpu types/revisions & designate the concurrent Tjmax/DTS/Delta/TCase etc then relay that information in an easy to see visual.

Kind of like a coretemp/realtemp that shows cpu temp/throttling in the form of an internet speed bandwidth "speedometer" display.

Tried RMClock? It has graphs showing temp, cpu load, and throttling.

 

[rge]

Throttling occurs at DTS=0, so if virtualLarry cpu is 9C delta to tjmax, he is 9C from throttling. He may or may not be exceeding Tcase max, depending on what program he is using to load. cooling used, etc, and no way of anyone knowing for sure. But as StinkyPinky accurately points out, he is only rarely exceeding tcase max, even if he is doing so, as he is probably only at that high temp when stress testing, unless he folds, etc.

Tcase max actually has nothing to do with throttling, intel simply means that under their max load testing program using their testing parameters/cooling, when DTS=0 and throttling occurs, tcase max is designed to be reached simultaneously. But they go on to say this may or may not happen based on cooling, loading program, etc, because this gradient is too highly variable to predict this will always occur, and in fact it many times will not occur.

 

[Razorfist]

Thanks for all the replies, it seems my own feelings about a 70C limit are about average

Nice email post rge, lots of official intel info (as well as shifted responsibility) in there

Myocardia, I'm sorry i didnt specify about whether it is core temp or CPU temp, i believe someone else clarified that nicely, the CPU diode temp or perhaps Tcase would be what I am looking for (even though the email between rge and intel shows that perhaps those are not the same, i am still trying to assimilate all the knowledge on the net into my database, so im not expert yet) however as JaredPace said, the core temps are not far off from that (and the core temps might be easier to find/readout).

Stinkypinky, I thought the same as you, that even though prime95 or orthos takes me to 100% 100% (2 core), I would never see that temp in real world applications that I use. But then just yesterday I converted an .avi file to DVD and that program was using on average 75% of one core and 50% of the other. Then i didnt want to just wait around watching this thing run so I started playing a game, which used up the rest of the CPU I had! (also running winamp in background). So I want to be sure that my max temperature is a safe number.

I believe im going to use 75C as my ultimate maximum, is I ever hit 76C even for a second, im toning something down. aiming for 70C, so I hope i dont get over 75C.

I should probably upgrade to watercooling and eliminate heat as my limiting factor. I hear that watercooling is no longer as messy as it used to be. Very newbie friendly now.

 

[Tempered81]

Information here:
http://forum.lowyat.net/topic/666785/+0

excerpt:
"The readings are directly from the DTS, so it is the most accurate reading you can get.

So, what is an acceptable value for "Delta to Tjunction". Well, nobody knows!!! From observations of "safe" temperature and recommendations by certains enthusiast, the acceptable value seems to be ABOVE 20-25'c (Note the word ABOVE, NOT below 20-25'c)

So, if you look at my Q6600 "Delta to Tjunction", I'm quite far from my maximum core temperature, so I'm quite "safe".

Note : Alternatively, you can use realtemp, which shows you the "Distance to TJMax" which has the same functions.


--------- 5) Is there such a thing as safe/optimum temperature? ----------

Note : If you're not an overclocker, then this question is not really important.

Most people will ask "What is a safe temperature to run my overclocked CPU?"

Well, there is actually no straight forward answer to this.

The "answer" depends on various factors such as:
1) Type of CPU
2) Maximum temperature
3) Duration of high temperature @ How you run your rig/CPU e.g. a few hours a day @ never switched off.....

Type of CPU
This is obvious as different CPU would have different thermal specification.

For example, take a Q6600's thermal specification (TcaseMax):
Q6600 B3 stepping: 62.2'c
Q6600 GO stepping: 71'c

Obviously, if both processors were running at a TcaseMax of 60'c, definitely the B3 will be the more "tortured" of the two CPUs!

And I guess it would be also true for mobile & server CPUs which are more reliable and can take higher temperatures....

Maximum temperature
Unfortunately, there is no absolute value for a safe temperature.
--- some would recommend 60-65'c
--- some would say 50-55'c
--- and some would run their procs at 70'c without any short-term problems

Just remember this : Heat is the number one enemy for your overclocked CPU (voltages comes in 2nd.....)

The higher the core temperature:
- the higher the risk of damage
- the shorter the life expectancy

Note : Remember that the lifespan of a CPU can be very long..... usually >10 years? >50 years?

In general, for current Intel processors, the recommendation is around 60-65'c

Very good in-depth explanation here :
The truth about processor degradation (Anandtech)
Overclocking's impact of CPU's life - by Joe Citarella

Duration of high temperature
Running your CPU at a high temperature for a few hours would definitely be better than running it 24/7

------------------ 6) How to reduce my CPU temperature? -------------------

Can be divided into different components:
1) Casing
2) Cooling (e.g. air, water, more extreme)
3) Air cooling - the fan setup
4) Thermal paste
5) CPU
6) Ambient temperature
7) Power saving methods
8) Overclocking/overvolting"



http://forum.lowyat.net/topic/666785/+0

@rge get with the author of this article who goes by the name, "kmarc". You guys could probably come to a clear conclusion on the temps...

The only thing Kmarc states that I tend to side differently on is the "heat is more of a threat than voltage" idea. I thought using greater than 1.40 Actual Vcore on E7xxx/8xxx posed a greater threat since aftermarket air cooling easily keeps the temps below 70 (if the overclocker has any idea what he/she is doing) even at this much voltage. I guess if you are careless it is possible to hit 80C with less than 1.3v.

 

[Tempered81]

Originally posted by: VirtualLarry

Tried RMClock? It has graphs showing temp, cpu load, and throttling.

Yah I did briefly when trying to undervolt my t7100. Never paid much attention to the graph though, I'll have to check it out again. thanks

 

[unr3al]

Once my temp goes past 60'C I turn down the clocks as fast as I can reboot lol Don't like getting too close to the thermal limit. Which is 65'C in the case of the Athlon64.

 

[Foxery]

Good read. I take Intel's official spec of ~72C as a general guideline - the temperature reading we get at home isn't exactly what they measure with their own tools, but it's a fair estimate, and they know this is all we have to go by.

rge's analysis of temperature gradients seems to back up the empirical evidence of safe Tcase temps for overclockers: TJ of 95C - 23C gradient = 72C, Intel's Tcase max. How convenient

Originally posted by: StinkyPinky
The other thing is how realistic is Prime 95 anyway? I certainly never put my CPU under the stress that Prime does. so if you get 75c in CoreTemp with Prime, does it mean you're going to get that playing a game for example? Not in my experience since for gaming the GPU takes a chunk of the load.

Programs like Prime95 and OCCT are considered "Stress Tests" because, um, they're stressful. They aren't supposed to be realistic, or normal in any way; just max out components as best they can in order to highlight faults. My CPU is as much as 10C cooler running even Folding@Home.

 

[VirtualLarry]

Originally posted by: unr3al
Once my temp goes past 60'C I turn down the clocks as fast as I can reboot lol Don't like getting too close to the thermal limit. Which is 65'C in the case of the Athlon64.

CoreTemp displays the TcaseMax, apparently it's encoded into the CPU for the A64. Mine says 59C on my 3800+, but my X2 3800+ said 70C or 72C.

My core temps for A64 3800+ got as high as 65C as measured by the internal thermal diode (CoreTemp), still no apparent problems for me.

 

[aigomorla]

the best answer is the LOWEST POSSIBLE YOU CAN ATTAIN!

however as for thermal max, 70C is always a good number.
Gives you that room so your chip lasts longer then its upgrade cycle.

If your asking me? My quads minus my B3 none of them load past 50C.

 

[unr3al]

Originally posted by: VirtualLarry

Originally posted by: unr3al
Once my temp goes past 60'C I turn down the clocks as fast as I can reboot lol Don't like getting too close to the thermal limit. Which is 65'C in the case of the Athlon64.

CoreTemp displays the TcaseMax, apparently it's encoded into the CPU for the A64. Mine says 59C on my 3800+, but my X2 3800+ said 70C or 72C.

My core temps for A64 3800+ got as high as 65C as measured by the internal thermal diode (CoreTemp), still no apparent problems for me.

Well I guess you won't see any apparent problems until it pops eh? lol