Printable View
I am writing a script that will shut down my computer if the CPU goes above a certain temperature. I would like advice on how high I can safely let it go before I shut it down. I don't mind being conservative, but on the other hand I don't want it shutting down all the time either.
I think normal temperatures may hover around 40-50 degrees with the bulk fan.
Of course, undervolt your CPU and perhaps use a better fan and you do not have to worry at all.
Well... I'd say 80 degrees C. According to the tech documents ~95 celsius is the maximum design rating.
I have never seen a processor get even close to that.
Anything around the 60C mark is fine IMO... considering humans can withstand such temperatures :-)
Silicon begins the process of melting at about 170F - 175F which I think is 70 to 75 celsius. At this temperature the outside of chip begins to get a bit soft and gooey. My processor has gotten pretty high but after a cleaning it's back down to the 55 - 60 range. Good ranges depend on the CPU in question. AMD's usually have a much different range. I do know for a fact that AMDs are super resistant to heat as I've seen one that hit 200F when the fan died and yet still runs today.
My current CPU has gotten a bit hot in the past and I can see areas on the chip where the surface texture is extremely smooth. Also see this around the various chips on the surface and it is definitely from heat. However no issues wth the CPU and I've had it now for going on 6 years.
My GPU has hit 90C before and it still runs just fine. NVidia claims they can hit 125C before having critical failure. I thought my recent blue screens were due to this but it is definitely drivers since 168.21 drivers work perfect on my card. All 178's and up where digital vibrance is broken cause infinite loop driver errors.
The question is really not how hot but how hot for how long. Long durations of high heat will certainly lead to failure. Spikes on the other hand probably will not.
Sorry, but silicon doesn't melt for several hundred degrees above 175'F. Having worked at the site that handles "faulty processor returns" at AMD in Europe, I have seen a few examples of what happens when the processor gets FAR too hot - the common problem with these processors is shorts between pins - this is caused by the solder that is used to connect the actual chip to the substrate melts - this happens at around 200'C - roughly 400'F.
Athlon 64 processors all have an internal shutdown temperature sensor that shuts the processor off before it takes any damage. This will probably happen around 100-120'C.
--
Mats
My CPU can take heat upto 70C.
A slight increase and...
*KABOOM* it restarts. ;)
[sigh]
Because of this i am never able to play high-end graphics games
[/sigh]
Undervolt your CPU and get a better cooling fan.
Or underclock it.
Definitely undervolt it, though, if you haven't already. Everyone should do that.
Thanks. :)
I'll try that.
I was able to undervolt my CPU from 1.35V to 1.1V. Quite the leap. Temperatures that was in the 40-50 now dropped to around 25-30 or so.
Not to mention it draws much less power now!
Also, check if there is a BIOS setting for "shut down processor above X degrees", because it may be the BIOS messing about rather than the CPU actually failing.Quote:
Originally Posted by stevesmithx
Getting a better heatsink should allow you to keep the temperature down at high load.
--
Mats
Thanks Mats for the tip.
I remember seeing such a setting.
I'll check this next time i switch on my computer.
Maybe you can, but I get a fever from the 24C in my office -- 60C and I'd probably be dead in no time.Quote:
Originally Posted by zacs7
60C is pretty lethal I would say. Anything above 30C is tough to handle for folks who are not used to high temps (like me)
Interesting. I got the information from one of our hardware folks and he said 175F was the beginning stages of silicon degradation. Perhaps melting was a bit extreme. My chip does have signs of degradation though and it has only ever reached 90C or so. I have small halos in the brown material surrounding the entire CPU and it's various chips on my AMD 3200 XP+. My XP 3200+ has about 6 chips arranged 3x3 around the central chip that is raised up a bit. When you look parallel to a light source you can definitely see some surface alterations in areas where heat has built up.Quote:
Sorry, but silicon doesn't melt for several hundred degrees above 175'F. Having worked at the site that handles "faulty processor returns" at AMD in Europe, I have seen a few examples of what happens when the processor gets FAR too hot - the common problem with these processors is shorts between pins - this is caused by the solder that is used to connect the actual chip to the substrate melts - this happens at around 200'C - roughly 400'F.
Athlon 64 processors all have an internal shutdown temperature sensor that shuts the processor off before it takes any damage. This will probably happen around 100-120'C.
Good information though. Now I will have to take to task the fella that misinformed me. I apologize for the misinformation.
30C?! Wow. I could never get that low unless I put water cooling in or liquid nitrogen. My fan setups and CPU fan keep the CPU around 50C to 55C and 60C to 65C under load. Of course I'm sure Athlon 64's run cooler than the 32-bit. I'll find out soon enough when I upgrade. :DQuote:
Anything above 30C is tough to handle for folks who are not used to high temps (like me)
Wouldn't by chance be a GigaByte board would it? My father had a similar problem with his previous GigaByte board. The folks at GigaByte said there was some type of video temp sensor that was tripping and resetting the system. After trying everything they said and more to fix it we never were able to resolve it and went back to an ASUS mobo. It only happened when playing games. Of course NVidia techs blamed it on the mobo as well.Quote:
My CPU can take heat upto 70C.
A slight increase and...
*KABOOM* it restarts.
We were referring to Human temperatures there, not CPU temperatures. Although my CPU temperature has never reached 30C yet with my Ultra Chilltec cooler. Right now it's at 23C. ;)Quote:
Originally Posted by Bubba
Oooh icey. My 3800+ ran at 30C even while playing games for hours, my 5200+ runs at 25-30 under the same sort of loads. And that's with the standard fan and the stock massively fat thermal pad.
For cooling, nothing beats 8 or so 120mm case fans to create a tunnel effect in the case -- seriously :D. Granted my PC sounds like it's about to fly away, but I can live with it.
Bubba, I think you are referring to the (slow) process of atom migration. That happens at all temperatures above 0'K, but the lifetime of a processor is calculated based on an average working temperature of 55'C - roughly 150'F - anything above that will shorten the life-expectancy of the chip. However, the chips that I know the life-expectancy of aren't likely to die within before the PC world has moved on many generations and the chip is no longer "useful".
--
Mats
And a reply to "temperature the processor tolerates": The max case temperature for MOST models of 3500+ are 70'C.
This is the maximum temperature that the processor is GUARANTEED TO WORK - it doesn't mean that it won't work at higher temperature.
--
Mats
http://products.amd.com/en-us/Deskto...88&id=89&id=90
You need to find the actual model number of your processor, as the max temp varies. If you dotn feel comfortable pullign the heatsink off, just use the lowest temp, 65 C. That is the absolute max before damage occurs, so I would use the lower rating of 55 C
No, it's not "the max before damage" - it is the maximum that the processor is GUARANTEED TO work - it's like the red-line on a car's revcounter: It tells you that "it's no good going beyond here" - it doesn't mean that if you manage to get beyond 6500 rpm [or whatever it is] that the engine will immediately be damaged - but it's obviously a bad idea to try to go much further, and don't change down a gear instead of up at that point!Quote:
Originally Posted by abachler
Likewise, the temperature where the chip takes permanent damage is MUCH higher - probably in the range of 150-200'C. It will malfunction (produce incorrect results or read/write bad data from/to memory), but this gives no permanent damage. Storing the chip in an oven at 100'C will be perfectly fine for months.
Also bear in mind that there are versions of the same chip that are OK to run at higher temperatures, up to 83'C - just not sold as Athlon64, but Opteron. These are the same design and process - just happens to be selected for their ability to run well at higher temperature. There will certainly be no permanent damage until well beyond the 83'C mark (but my experience says that chips keep "working" until the solder under the chip melts at some 200'C - they just do not work CORRECTLY any more).
--
Mats
prolonged operation of semiconductor devices at temperatures above their recommended maximum will cause diffusion of the doping ions across the NP junction at accelerated rates. As the percentage fo ions with atomic velocities above the design limit increases, the NP junction shows an increases rate of diffusion. Diffusion of doping ions leads to a decrease in NP junction stratification, a gradual degredation in the reverse bias potential, increased leakage current, increased switchign time, decreased V/S slope rate, decreased maximum useable frequency, increased temperature due to power consumption and eventual failure of the device. While ions naturally cross the NP junction even at temperatures close to absolute zero, the rate is greatly increased as the mean molecular velocity of the sample is increased.
but you guys already stated that earlier, so ill let it go at that.
Yes, and even at 100'C, it's not a INCREDIBLY fast process - devices will last many years at those temperatures. The mean time to failure of these type of devices is calculated at a 55'C temperature, and at that temperature, they will last MANY years - so many years that I can guarantee that you will not want to use it by the time it gives up it's ghost from natural old age. [Of course, the usual bath-tub shape of failure frequency applies - some devices will die very early, then very few in the middle of the age-range, and then increased failures ate the end of the expected life-time].Quote:
Originally Posted by abachler
--
Mats