Sign in with
Sign up | Sign in
Your question

real real real real real max cpu temps

Tags:
Last response: in CPUs
Share
October 22, 2008 5:18:38 AM

So I want to know from the experts out there what the real real real real real max cpu temps out there are for the new cpus.

Most new cpu tjmax are rated at between 80-100 degs C. Hit that and THERMTRIP kicks in. eg. the very popular q6600 GO @ 95-100C tjmax.

But... is Intel or AMD holding back the real real real real real max cpu temps? Surely they burn a few in every batch to test the absolute threshold. Is 80-100 still a safeguard set by the manufacturers and distributors to protect themselves from returns and complaints? Can the core, silicon transistors, cache etc. actually hit 120c's+ before it really really gets damaged? What internal parts are most sensitive to the heat??

Isn't 100degs C a very convenient round number. If it was the real real real temp before critical overheat wouldn't it be like 102 degs C.

A few questions to be answered by whoever thinks they know the real answer because there is so much $h*t on the web plus all the impractically useless calculations. They all just derive from the back of the cpu box or Intel spec sheets, but those are just out for everyone to view. That doesn't tell us what's really going on in the labs.

Who can testify to their cpu running at tjunction throttle for like 5 months and then finally realized their fan was dead. Who's actually sacrificed a cpu without any cooling just to see if they could damage it? Maybe Tom's will do some definitive burnout testing? After THERMTRIP is the damage already done? I sure hope not. If so then why doesn't THERMTRIP of a q6600 kick in at like 80degs, 20 less than tjmax? That would surely save the cpu from damage and keep an accident from turning into someones bad day. If CPU's should be running below 60 and commonly well below 50 that still gives you 20-30 degs before "my safer tjmax" at 80degs.

When your fan stops, your pump dies, or your heat sink clip breaks and you go for a sandwich what really happens to your cpu before it throttles then shuts off? How does a damaged CPU run or does it run at all? What we need to know are the cold hard facts. What just happened to my machine? Is there any damage done to the CPU, PCB, socket etc.? Should I be worried about errors, buy a new CPU, mobo etc?

For example if my q6700 was running for 20 minutes at true core temp of 85 degrees does that matter at all?

Let's get the definitive expert answers here, and weed out all the hearsay.

Can I please have some piece of mind.
Any and all comments appreciated.

Thanks all, J

btw... I gave this topic heading this name to catch attention and replies not just to be cute ;) 
October 22, 2008 6:27:52 AM

Ive had my cpy fan stopping while my computer was running, nothing happened except it throttled down.

I played games for some days after that, before my new fan arrived, and i just had lower fps, thats it.

Although on my amd computer, it will shut down when the fan isnt working.
October 22, 2008 6:33:34 AM

Edit: Eventually on my old intel computer, the motherboard failed. That was after 2 years of heavy overclocking, without enough cooling on the motherboard. My experience is that the cpu wont be damaged unless you really want to hurt it.
Related resources
October 22, 2008 6:34:21 AM

(WHY CANT I EDIT MY POSTS?)

But why would you want it to run that hot anyway?
a b à CPUs
October 22, 2008 6:42:46 AM

Silicon melts at 1420C. It's everything else that can't get that hot ;) 

THERMTRIP activates at temperatures below where you could do damage in a short period of time. Perhaps if you ran your CPU just under Tj Max 24/7 you might do some damage, but just pulling off the heatsink on a Core 2 CPU won't damage it. I don't know how AMD's thermal protection works, as far as I know it is still tied to the motherboard, but I could be entirely wrong.
October 22, 2008 6:47:38 AM

No one wants to run that hot. That's the point.

Surely no one intentionally disconnects their fan or drains their coolant while the machine is still running but sometimes accidents happen.

The real question that seems to be disputed is a t what point is damage done. Of course every CPU is different. There are too many varied opinions out there some even say the CPU will be damaged at 65C some say that after tcase max at 71C for example the CPU will sustain damage, some after tjunctionmax damage while some disagree entirely explaining that there is no proof that is the damage threshhold only the shut down temp to "save" it from possible damage, and then some testify to having their CPU's fanless for months running at 90+C and they're still going strong.

So.... where's the definitive answer? All I'm trying to do is collect as much feedback to compile a worthy analysis and then make a conclusion unless we get a direct quote from an Intel chip engineer explaining the inside story perhaps we can only conclude from evidence.

My thanks to everyone who contributes to this thread.
Cheers, J
October 22, 2008 6:49:41 AM

i think, your theory could be quite right.
the only problem is the sense and purpose for the quest.
the most important requirement over speed and power consumption is reliability.
the question is how much reliability and speed for a given cpu temperature most people want. if a modern processor is indeed running at 100 to 120 Celsius, then it would run really slow or either unreliable, always crashing the apps or OS. Also, when an app crashes, the processor loses the load, thereby drop the temperature.
when a fan or pump fails, nothing will happen as long the heatsink is still in contact to the processor and the person using the computer will wonder why every few minutes of running time crashes the whole system. or, can get to bios but freezes.
as long you can use your computer,the processor is in good temps.
so, it is not true that one can destroy a processor through overclocking and prolong use. what overclocking kills are the capacitors in boards.
i have overclocked pentium mmx 166 and athlon xp 2000 that are overclocked since they were bought, imagine how old they are today.

to destroy a processor, one would not install a heatsink at all. in seconds, the processor will pop or explode at during loading the bios which is almost idle and uses a couple of watts.
testing could be done, but not worth it. heatsink mounts are very reliable and a box in operation is always stationary and processors are cheap to replace.
October 22, 2008 7:10:07 AM

Thanks for that reply. A few things I have to comment on. (btw I don't want to come across as aggressive ;)  The issue here has nothing to do with OCing. We are talking about temps and damage and the absolute truth. Firstly there have been tests on the new thermal sensors in the newest cpu's as of 2007 and I've read that they tested them without heat sinks and they throttled at tjunction for some time until THERMTRIP and they booted just fine after. Kudos to those engineers. Sometimes the heatsink comes lose or falls off altogether. As well a waterblock without flow only acts as an insulator creating more heat so... we still need to know the facts. CPU's with PROCHOT and THERMTRIP save cpus but do they sustain damage? or do they just save your house from catching on fire?

and CPU's ARE expensive to replace especially if you are an enthusiast who just picked up the newest QX9775 at 1800 bucks only to wake up the next morning to find your equally new and expensive cryo cooler lost pressure and the cpu has been throttling all night. You might want to know if you should be calling the pump manufacturer looking for a warranty or some compensation.

Thanks for keeping up the dialog.
Cheers, J
October 22, 2008 7:11:19 AM

oh, thought i was the first to comment. quick replies here huh. :) 
a simple answer would be, your cpu won't be damage at 70 90 or 100 degrees. if you're running the cpu at those temps for years, it won't damage it all.
that's the misconception of many people and many article editors, that a processor's lifetime is affected by overclocking.
October 22, 2008 7:16:08 AM

Cool thanks man. Thats your input. Hope its solid. We'll see what others say to that.

What we want to eliminate here is the conjecture that is found scattered amongst cyberspace concerning CPU's and heat damage.

We need either expert opinions or testimony to make a conclusion.

I'm just playing moderator here as I really have no true idea. If I did I wouldn't have started this thread. hehe ;) 
October 22, 2008 8:34:17 AM

As a practical matter, what a modern CPU can run max temp will vary from batch to batch. The MFG will set a limit, usually two, when the CPU starts throttling or shuts down. Why do they throttle? To protect their warranty. This question is unrealistic, impractical, and probably also confidential information from the MFG's point of view.
October 22, 2008 9:19:24 AM

as far as i know, military equipments can work without any error or problem up to 120 degrees C or even a little further. any other personal things up to 110 max. if you put them to more heat for a longer time , the transistors will be heavily messed up.
October 22, 2008 9:31:45 PM

Reply to Croc. Thanks for the reply. We've established that every cpu will be different. Yes they can take a wide variance of heat from mfg to mfg and from model to model. (please take no offense, I'm just playing moderator)

I don't believe the question to be unrealistic or impractical as if it were there wouldn't be the throng of posts scattered about the web on the issue. The question is however, prying and a little tedious and will require some amount of input to come to a conclusion.

Interesting you mentioned warranty. My theory is that the chips can run much hotter and for the distributors and MFG's to protect themselves from having to dispute or replace so many fried chips (because every noob F's with OCing these days) they've purposely under regulated the THERMTRIP temps.... But, that's just a theory.

I'm saying there was a time not long ago when chips would cook themselves before full bootup if there was a critical cooling problem and the MFG's likely got tired of all the complaints and implemented the internal safeguard. Some of those max temp settings posted were truly MAX settings before you found yourself with a useless wafer of fried transistors. I think alot of the info posted online are explaining predated figures. How do you get people claiming to be the authority on the matter explaining that the CPU will be damaged at over 70 degs while some claim 120+ even discussing the same models! That's really unsatisfactory info.

Still doesn't answer the question of "safety shutdown" as in 'your cpu was saved of absolutely any and all damage', or "safety shutdown" as in 'sorry you had a cooling accident we managed to save your cpu from destroying your whole system and smoking up your room, but we'll be looking forward to seeing you back here to buy a new cpu soon since your's now includes erratic behavior, displays visual artifacts, possible data corruption and is frankly just about on the way out.'

I've read much about the stress heat put's on the cpu. It can be calculated. For example a cpu that is properly cooled operating under the MFG's cool temp could actually last 20+ years. (Mom ran an 8086 for word procs until just recently) Whereas a brand new cpu that has just withstood a throttling until THERMTRIP may now only last 2-3 years since it's been severely fatigued.

Perhaps the question is confidential, thats why I'd like to research it. Intel didn't answer it directly when I asked them. They just reitterated what the spec sheets explain about PROCHOT and THERMTRIP. Not much help if one can simply find this info by googling their cpu model.

I have not been able to find the definitive answer to the question of whether the MFG's set their tjmax and THERMTRIP temps right at damage point (like some people profess) or well clear of any damage. That's why the post is titled real real real real real max cpu temps.
It's a collective research project seeking an answer to settle the minds of so many people posting "Help! Cooling failure... Did I just F my CPU"

Thanks again all
Cheers, J
October 22, 2008 10:27:17 PM

short term really is not an issue
long terms is a problem with degradation of epoxy and plastics in construction
the is migration a degradation of silicone increase with temperature and time
finally the resistance decrease with temperature that is why when you push systems they run better at lower temperature as temps go up electrons leak at an increasing rate

that is why the rule is a low as possible, 75c is a good limit for long term based on my experience not scientific data
a b à CPUs
October 22, 2008 11:55:33 PM

I think the real issue is not what damage will be done to the processor, but what will be done to the socket if you pull of your heatsink when disabling thermal protection mechanisms. Preventing the processor from reaching stupidly high temperatures prevents damage to other components as well. I'm fairly sure THERMTRIP would be set quite a bit lower than the real real real (etc, etc) max temp.
October 23, 2008 12:58:46 AM

Right good point. The socket although a part of the Mobo is directly linked to the CPU and thus related to this thread.

That's a big issue since were talking about damage here. How tough are sockets and the surrounding PCB and components on Mobo's. Can they withstand 100+ temps and not be damaged? You would think that they should be able to.

I'm running an Asus Blitz Extreme. Extreme for OCing and supposedly high heat threshold, but a chip at 100C sitting on the socket; that's like putting a hot plate on it. Anyone know what sockets are made from these days? I'm guessing some kind of high temp withstanding fiberglass. Anyone know?

Again what I'm trying to seek out is if any damage can be done to the cpu and surrounding area i the event of a cooling failure.

If the MFG's have engineered everything correctly than wo shouldn't have anything to worry about as the computer will have saved itself long before any damage has been done, but only possible inconveniencing one with having to reapply and reset a heatsink.

Dragonsprayer has contributed some very convincing and concluding evidence to argue that no damage will be sustained to the CPU at throttle temps.

So what about the socket and mobo?

j
June 4, 2009 2:21:58 PM

Hello, I've been reading your post. and the little I know about semiconductors is that they are very unstable with the temperature. Semiconductors are the holly grial of electronics, I mean that without their particular properties we couldn't have things like computers, flats TV, broadband access to internet, MP3 players...

The thing is that processors work thanks to the electrons running between transistors. Transistors are made from silicon, but not only silicon; the silicon have to have impurities to be a semiconductors, those impurities make that electrons can be managed as the designers wishes, but those impurities rise with a temperature. When a semiconductor can't manage the amount of current running in their surface. Someone says before this post that silicon gets burn at 1414 ºC, that's true, but it's wrong. What damage the semiconductors in fact is the energy released for the electrons when they get free of their prision in the atoms links.

In fact it's amazing that something so little that a processor could disipate a lot of heat. a processor managing 44 Amps its really awsome. When you increase the core speed, you aso increase the current running inside the processor. Speed means higher current!

I hope this very little (and maybe wrong) explanation could help you to solve this problem.
!