Reading Athlon XP thermal diode?

[Leo V]

Does anyone know of a program that reads the Athlon XP's on-die temperature sensor? I'm using Asus A7V133 and I believe it's only showing the Socket sensor's results.
Leo

PS: Just upgraded CPU from Athlon 1.1ghz to Athlon XP 1700+ (1.466ghz), and I'm very happy. It actually runs cooler on idle (37C with default mhz/voltage instead of 46C on vanilla Athlon). And software compilation under Linux is that much faster

 

[Mikewarrior2]

no current mb save the Fujitsu-Siemens KT266 board reads the diode off the shelf...

the people currently reading the diode are modifying their motherboards to do so. reading the diode requires hardware modification onboard the motherboard to enable this feature.

The A7V133 may be able to be modifed for it, but no software program alone will be enough to "force" diode reading.



Mike

 

[Rand]

<< no current mb save the Fujitsu-Siemens KT266 board reads the diode off the shelf...
>>


And Asus dually Athlon board.

 

[rickn]

it really isnt THAT important. You can basically figure on a 5-7c degree difference between a diode reading and a thermistor. So, if a thermistor is showing 44c under full load, you can figure the internal diode temp would probably be in the ballpark of 48-52c.

Manufacturers have algorithms they have programmed into the bios that try and 'approximate' as close as possible the actual temp. ASUS is one company that always seems to be tweaking those algorithms.

 

[Mikewarrior2]

Rand,

Damn, I keep forgetting the asus dual board .

RickN,

Where do you get your info? 5-7C Off? Are you claiming the socket-thermistor is accurate? Overclockers.com is saying 7-15C off.

Well, here's some info on a diode versus a7v133 socket-thermsitor:

VoidYourWarranty Internal DIode Heatsink Roundup I: Pay close attention to the results. The Inaccuracy of the socket-thermistor isn't constant. sometimes it over-reports temps, sometimes under. It compresses results as seen on the Millenium Thermal heatsinks page. 7C diode difference, yet the socket-thermistor only shows 2C? Another huge issue with socket-thermistors? Exaggerated airflow gains, as exhibited by the volcano 7. High speed fan shows a socket-thermistor 4C improvement, whereas the diode only shows a 2C improvement.

No matter how much tweaking by the manufacturer, there is no way in hell that a socket-thermistor will ever be an accurate type of CPU temp measurement. Compression problems are the most obvious, heatsink ariflow problems will be a problem, and they aren't even consistently inaccurate.

Also, check out this void your warranty diode page: Initial Diode Findings. 7-5C diode changes aren't even picked up by the asus socket-thermistor.


Mike

P.S. If you trust the socket-thermsitor measurements, heatsink ranking changes drastically... the Antec will beat the glaciator II, etc. hell, the Volcano 7 with maxed fan will beat the Glaciator II.

 

[thermite88]

As seen from A7vTroubleshooitng Internal Diode review, internal core temps are probably higher than people would like to admit. I suspect that no MB manufacturer wants to be the bad news messager. Being the first MB to report 80 degrees C CPU core temperature is not good PR.

<< it really isnt THAT important. You can basically figure on a 5-7c degree difference between a diode reading and a thermistor. >>

The MB manufacturer gets away with it (not measuring thermal diode) because most AMD users are not very demanding in thermal protection.

 

[BD231]

<<I suspect that no MB manufacturer wants to be the bad news messager. Being the first MB to report 80 degrees C CPU core temperature is not good PR.>>

Why would anyone care if there computer has been running fine and dandy?, to me this means theres more room for improvement on heat management which is a good thing.

 

[Mikewarrior2]

because the first mb to fully support the diode will be showing 60+C with retail heatsinks while socket-thermistor equipped mb's will show the same heatsinks in the 40C range.... it isn't good PR. Asus gets enough flak already for having "higher than actual" temp readings even though they raen't really higher than actual.



Mike