unanswered temperature questions

[not_this_punk]

x2's maximum allowed temperature is 65 degrees. ok. what temperature should that be? the temperature of the IHS as meassured with a probe, the temp reported by the thermal diode or the core's actual temp (meassured with coretemp for example)? the answer seems simple but i`m asking this because i don't see any mb manufacturers putting the core temperature as the cpu temperature in bios...instead the cpu temperature is the one reported by the diode but that should be way off...

i kind of babbled around but i`ll make it short: if no motherboard producers incorporate that sensor in the temp monitoring, i figure that the temperature meassured by the thermal diode is the one that should not exceed 65 deg. am i right? the autoshutdown at x degrees feature would be pretty useless then. and that would also mean that all the mb manufacterers would lead the clients in error, consciencely assuming the risk of fried CPUs because of inaccurate temperature reporting.

ok.

another dilemma of mine is why does temperature matter that much? i mean a physical, electronic explanation. does the temperature affect the electrons in such a way that they could interact with the actual "circuitry", and cause possible damage?

[Praz]

The motherboard manufacturers don't use the thermal sensors for temperature reporting because it is an undocumented feature of 939 processors. Except for maybe some industry insiders, no one knew this form of temperature measurement even existed until The Coolest released CoreTemp.

 

Unless otherwise stated, when temperature values are being discussed it is assumed it is the value of the thermal diode that is being quoted,

 

As for as one method of measurement being more accurate the the other it's anyone's guess. Since the thermal sensor technique of measurement is undocumented by AMD there is no published accuracy information available. AMD white papers for the 939 processor do have accuracy information for the thermal diode. It's accuracy can be off by as much as 14C.

 

AMD does outline a relatively simple testing method that will allow an accuracy of 2C excluding equipment error. But it's of little value as it will only apply to the processor being tested.

[ReelFiles]

It means the actual temperature of the core. (or so I thought)

 

The cooler a conductor is the better it conducts electricity.

The circuitry in your chip is so small that it doesn't take much to burn out, think of a light bulb burning out when if finally dies. Most light bulbs die when initially turned on because the voltage fluctuates or "spikes" when you first flick that switch, so the wire burns out just like a CPU would under too much heat/electricity.

[not_this_punk]

thanx

[Cuttlefish1212]

You would think that manufacturers would devise a more accurate and efficient way to report temperature values. Yet, most people aren't buying a computer to overclock the hell out of them, so those that do just have to deal with the lack of innovation.

 

Eventually though, they will need to come up with something better. The 65nm architecture is already proving to be significantly hotter than it's 90nm predecessor and if they hope to go smaller, they need to solve the heat problem.

[not_this_punk]

sorry for my eventual lack of information but i thought that the 65 nm cpu's were a little cooler and less power hungry...at least that's supposed to be the case with pentium d 8xx vs pentium d 9xx from what i've read.

[MalibuSS]

sorry for my eventual lack of information but i thought that the 65 nm cpu's were a little cooler and less power hungry...at least that's supposed to be the case with pentium d 8xx vs pentium d 9xx from what i've read.

 

For the most part, you are correct. Smaller die's generally take "smaller wattage", which means less heat. (for the most part) There are exceptions to every rule of course.

 

 

-MalibuSS