getting accurate cpu temperature readings on KT7...

[eyor]

In another thread someone was going on and on about how inaccurate the temperature readings for the cpu were on the Abit KT7. Is there a more accurate way of getting readings?
Thanks

 

[Mikewarrior2]

The most accurate low-cost way to to hook up a thermsitor to the side of the cpu core, and attemp to isolate the thermistor so that all it touches is cpu core. This will be much more accurate than the internal in-socket thermistor, but still won't be perfect.

And yes, it was me going "on and on" about it being inaccurate. And for being inaccurate is why I think a comparison like the Anandtech heatsink comparison should not be done.

Mike

 

[Maverick2002]

hey mike i got a question for ya:
which is the best thermosistor: the digital doctor or the compunurse? (size doesnt matter too much cause if i buy one ill use a 5.25 bay anyway)
thx

 

[Maverick2002]

sorry double post

 

[eyor]

mike: thanks for the info. as for the thing about you "going on and on," lol, I actually hadn't seen your most recent thread. I had a pretty good time reading your heatsink test bashings, lol. Although you do make a very good point!

 

[Mikewarrior2]

Maverick2002,

Either one should do fine. I don't have any personal experiences with either.

Eyor,

Thanks. The readings are "accurate" for where the thermistor is reading, but it is not necessarily accurate for change in core temp.


Mike

 

[Maverick2002]

yo mike so do you put the thermosistor between the cpu and the heatsink? (on die, NOT the core, of course) and should it be touching the core of just being real close to it? thx

 

[Maverick2002]

oh one more thing.....does the digidoc also function like a fanbus? cause it says you can control 4-8 fans.....hmmmm

 

[Mikewarrior2]

Put it up against the edge of the cpu core. You can't put it between the cpu and heatsink(that'll hurt cooling).

From Technonut.

I'm not sure how the Digital Doc controls fans.


Mike

 

[eyor]

Thanks for the picture, I have been wondering about that forever, lol. How much more accurate do you feel a setup like that is?
I think that you should do your own socket A cooler roundup, hehe. Seriously, it would be interesting to see your results.

I did wonder about the one posted on andandtech, because it basically said that everything tied, or came pretty close to it. And in other tests that I have seen that monstrous Taisol crushed anything else.

 

[Mikewarrior2]

The problem is, even with an external thermistor like that, you aren't getting true core temp.. The closest you'll get is probably -10%, so you're looking at getting a reading of 45, wth an actual temp of around 50C(for example). But an external thermistor is quite a bit more accurate than the internal thermistor, despite what tillman says. I believe he compared the thermistor readings versus a single heatsink, rather than all the heatsinks, because I believe the results would have been different between external thermsitor/in-socket thermistor with some of the poorer performing heatsinks.

I personally wouldn't do a review unless I/anyone found a way to get within 5% of theoretical core temp. Unless you can do that, a heatsink comparison isn't really accurate.


Mike

 

[Maverick2002]

ok mike, so here is the next question: it looks like the probe IS attached to the core....is that right? also, what kind of tape/adhesive is that, and what kind should you use (i assume regular tape would melt)?

 

[Mikewarrior2]

thermal tape usually works well, i've also seen people use high-temp electrical tape.


Mike