Temperatures on MSI k7t

[iamfried]

I have a MSI K7T Pro2 mobo with a duron 800 o/c to 896 w/ the retail HSF thing. I am currently running MBM5.0 to check out my temperatures and was wondering if these were realistic and nothing to worry about. PLS let me know. (also, what are these different sense points and what significance do they have, ie; where are they???)

Side of case on:
Via686A-2 = 42C Via686A-1 = 25C Via686A-3 = 30C

Side of case off:
40C // 25C // 26C

Temperature differential seems somewhat high seeing what others have had. (maybe, I don't know) Due to some posts, I enabled +3C compensation for sensor#2 and sensor #1. I am also thinking about getting the HSF combo that this link re's to and was wondering what any of you thought about it, if anything? (I seriosly feel like burning $25)

Text

 

[Mikewarrior2]

hmm... the duron retail heatsink, is it a taisol unit?

You might want to try the forged taisol, the cgk742092, or a fop heatsink. that might drop your temp. Bear in mind that the MSI boards usually under-read by a good margin(more than 3) and that any new heatsink's "percieved temp drops" will be less than actual temp drop.



Mike

 

[Jalapeno]

Let me contribute some of my observations on the same mainboard with a 1.1 TBird running at 1.25Ghz:

- I never trusted the "FuzzyLogic" utility MSI provides, so I went ahead and bought me a CompuNurse sensor unit. I was able to place one probe between the chip and the heatsink, touching the core with its tip. Temperature reading as we speak: FuzziLogic - 38 degrees, CompuNurse - 44 degrees (with SETI running constantly). BIG difference!

- I also switched from the forged Taisol HS to a FOP32 which brought my temps to the current level; before it was about 3 degrees higher. I used generic thermal paste (the white stuff) with both heatsinks.

Stay away from the ORB crap!

 

[Poof]

Jalapeno - I will 2nd your assessment and verify exactly what you've found with the MSI K7T Pro2-A. And note that I have done the EXACT same thing as you - I have a CompuNurse probe attached to my CPU (a Duron 700@954 now) next to the core, and am consistently reading 6-7C higher with the probe than the on-board sensor reports.

Granted, I'm sure the probe is picking up some air temperature too but that's a significant difference of temps in my mind...

 

[Mikewarrior2]

Especially considering that the thermistor touching core-edge is a bit lower than core temp... (probably around 10-15%, so if you're running 25C over ambient, tack on another 3C or so).

THe MSI uses one of the in-socket SMD resistors to measure temp. Probably the least accurate Socket-a board. It doesn't even touch cpu back-substrate.


Mike

 

[Jalapeno]

Finally I know HOW the K7T "measures temps"! At least how it tries to do so...

Thanks Mike!

 

[Mikewarrior2]

Don't thank me... SOmeone posted it in CPus/processors forum, and he got the info from the tomshardware guy who did the review of the pro2a. He got the info from MSI(which is funny, since many of us have e-mailed MSI and never got a response).

I personally think its a pretty crappy method...


Mike

 

[Jalapeno]

It IS crappy, considering the M7K is a pretty decent board in all other respects.

Oh well, nothing is perfect...

 

[iamfried]

Great, thanks for the information. Where can I pick up one of those probes and is there a site that explains how and where to hook them up??

 

[Poof]

I got mine at coolerguys.com. It's on that link's page but you need to scroll down a bit.

I just attached mine with some tape, making sure the tip was just by the core (you have to be careful not to have it actually full on the core and then smashed by the HS - which would mean a dead probe and loss of HS/core contact).

And MikeWarrior - whereever that thermistor might be located, it certainly is hidden because I sure couldn't see it in the socket... geez.

 

[Jalapeno]

POOF,

it's not really a thermistor, just one of those SMD resistors USED as a thermistor. Actually, it's a good idea - just doesn't work too well...

 

[Mikewarrior2]

certain issues like socket-temp(temp from the other resistors) and other stuff effect the kt7 series readings. The least they could have done was at least raised the SMD resistor that reads temp. When I was looking at a board yesterday, they all are the same height.


Mike

 

[Hanky]

<< Probably the least accurate Socket-a board. >>



Maybe, but it's not alone: I think AOpen uses the same method on their SoA boards.

 

[Poof]

As far as I'm concerned, it's measuring the air temp near the backside of the CPU...