Thread: Max temp for Athlon 64 3500+ CPU

  1. #16
    Registered User VirtualAce's Avatar
    Join Date
    Aug 2001
    Posts
    9,607
    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.
    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.

    Good information though. Now I will have to take to task the fella that misinformed me. I apologize for the misinformation.

    Anything above 30C is tough to handle for folks who are not used to high temps (like me)
    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.

    My CPU can take heat upto 70C.
    A slight increase and...
    *KABOOM* it restarts.
    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.
    Last edited by VirtualAce; 01-04-2009 at 09:21 PM.

  2. #17
    and the hat of sweating
    Join Date
    Aug 2007
    Location
    Toronto, ON
    Posts
    3,545
    Quote Originally Posted by Bubba View Post
    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.
    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.
    "I am probably the laziest programmer on the planet, a fact with which anyone who has ever seen my code will agree." - esbo, 11/15/2008

    "the internet is a scary place to be thats why i dont use it much." - billet, 03/17/2010

  3. #18
    Woof, woof! zacs7's Avatar
    Join Date
    Mar 2007
    Location
    Australia
    Posts
    3,459
    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 . Granted my PC sounds like it's about to fly away, but I can live with it.
    Last edited by zacs7; 01-05-2009 at 12:03 AM.

  4. #19
    Kernel hacker
    Join Date
    Jul 2007
    Location
    Farncombe, Surrey, England
    Posts
    15,677
    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
    Compilers can produce warnings - make the compiler programmers happy: Use them!
    Please don't PM me for help - and no, I don't do help over instant messengers.

  5. #20
    Kernel hacker
    Join Date
    Jul 2007
    Location
    Farncombe, Surrey, England
    Posts
    15,677
    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
    Compilers can produce warnings - make the compiler programmers happy: Use them!
    Please don't PM me for help - and no, I don't do help over instant messengers.

  6. #21
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    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

  7. #22
    Kernel hacker
    Join Date
    Jul 2007
    Location
    Farncombe, Surrey, England
    Posts
    15,677
    Quote Originally Posted by abachler View Post
    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!

    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
    Compilers can produce warnings - make the compiler programmers happy: Use them!
    Please don't PM me for help - and no, I don't do help over instant messengers.

  8. #23
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    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.
    Last edited by abachler; 01-07-2009 at 11:18 AM.

  9. #24
    Kernel hacker
    Join Date
    Jul 2007
    Location
    Farncombe, Surrey, England
    Posts
    15,677
    Quote Originally Posted by abachler View Post
    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].

    --
    Mats
    Compilers can produce warnings - make the compiler programmers happy: Use them!
    Please don't PM me for help - and no, I don't do help over instant messengers.

Popular pages Recent additions subscribe to a feed

Similar Threads

  1. questions on multiple thread programming
    By lehe in forum C Programming
    Replies: 11
    Last Post: 03-27-2009, 07:44 AM
  2. Replies: 1
    Last Post: 10-27-2006, 01:21 PM
  3. Game of life
    By JoshR in forum C++ Programming
    Replies: 30
    Last Post: 04-03-2005, 02:17 PM
  4. Problem Putting INTs Into a CHAR Array
    By cram in forum C++ Programming
    Replies: 13
    Last Post: 10-13-2004, 07:53 AM
  5. If the RGB color is (64, 64, 255), change it to (64, 255, 64).
    By Grayson_Peddie in forum C# Programming
    Replies: 2
    Last Post: 06-14-2003, 04:26 PM