PDA

View Full Version : How do you guys check temps?


Clevor
01-13-02, 05:30 PM
I just put together a KK266-R with a pretested XP1600, unlocked by the vendor. I had the foresight to install a digital temp probe touching the CPU.

If I look at the CPU temp in the BIOS, it says 32C idle, room temp of 22 C. Great, right? But when I check the digital probe, it says 42C. I kinda tend to believe the probe.

On my Asus A7V board, it's the opposite, the digital probe reads 10 C under what Asus Probe says.

I'm using a pretty decent cooler, with a copper core, aluminum fins, and a 5000 rpm fan (you can't get it in the U.S.)

HugoHenry
01-13-02, 06:54 PM
Most MOBO now come with a sensor on the socket and software to read the temps thru OS. I had 2 Ystech 6000 RPM fans on mine so what is this special fan you have? i just ordered the MCX462 supposedly the best cooler on the market it has a 80mm 68CFM fan on it, running at 4900 RPM. You can get a Delta also at 6800 RPM. Loud as all get out, but its fast.

Thelemac
01-13-02, 07:04 PM
It kinda depends on how you installed the probe.

I would tend to believe it as well, though. :)

pHaestus
01-13-02, 08:20 PM
From testing with a compunurse epoxied under the center of the core on a TBird as well as with an internal diode reader on an XP, I have found that the probe on the side of the CPU reads lower than the actual core temp (which isn't surprising since it has to read from the side of the chip like that). The temp results are usually reasonably consistent in a qualitative sense though. I mean that you can tell if one cooling method is better than another one, but you can't tell which of two systems set up by different people is better just from their temps.

Dissolved
01-13-02, 08:44 PM
Originally posted by pHaestus
From testing with a compunurse epoxied under the center of the core on a TBird as well as with an internal diode reader on an XP, I have found that the probe on the side of the CPU reads lower than the actual core temp (which isn't surprising since it has to read from the side of the chip like that). The temp results are usually reasonably consistent in a qualitative sense though. I mean that you can tell if one cooling method is better than another one, but you can't tell which of two systems set up by different people is better just from their temps.

very ture.. i wish my mobo would read the internal temp prob.. but its a lil older..
but im gonna get a dd5 soon i hope ;)

pHaestus
01-13-02, 09:00 PM
No mobos (except Fujitsu Siemans) properly read the diode. You can make your own reader though :)

Dissolved
01-13-02, 09:02 PM
Originally posted by pHaestus
No mobos (except Fujitsu Siemans) properly read the diode. You can make your own reader though :)

i wonder if theres a way to make a mobo to be able to read it? cuz my mobo has the amd761 chipset so its a bit outdated, but im gonna get a 2200+ and id like to use the diode

Yodums
01-13-02, 09:14 PM
Usually now and days people use the Digidoc to get the temperature as no one trusts the the diode under the socket no more. To your setup it looks like everything is correct but you shouldn't worry about idle temperature you should be concerned about full load.



Yodums

pHaestus
01-13-02, 09:18 PM
If your motherboard has a SMBus header and you can solder you can make a reader, or else you can tap into the SMBus signals from a DIMM slot. People are proposing that you can just rewire the traces under the socket to point to the diode in the XP, but that as of now doesn't properly work and IMO will never work as reliably as a separate IC with every trace and wire done to spec. Here's my internal diode reader: http://www.voidyourwarranty.net/review/review.php3?r1=xpdiode/index.php3

Joe has a kit from maxim-ic.com that was $150 that does the same exact thing but with better temp resolution.

Dissolved
01-13-02, 09:21 PM
Originally posted by pHaestus
If your motherboard has a SMBus header and you can solder you can make a reader, or else you can tap into the SMBus signals from a DIMM slot. People are proposing that you can just rewire the traces under the socket to point to the diode in the XP, but that as of now doesn't properly work and IMO will never work as reliably as a separate IC with every trace and wire done to spec. Here's my internal diode reader: http://www.voidyourwarranty.net/review/review.php3?r1=xpdiode/index.php3

Joe has a kit from maxim-ic.com that was $150 that does the same exact thing but with better temp resolution.
hmmm... id rather pay 50 bucks for a dd5 ;(

pHaestus
01-13-02, 09:46 PM
I have a DD5. It can't tell me the internal temp of my AMD CPUs.

Dissolved
01-13-02, 11:05 PM
Originally posted by pHaestus
I have a DD5. It can't tell me the internal temp of my AMD CPUs.

i didnt mean it that way, just to get the temps..

ButcherUK
01-13-02, 11:17 PM
Originally posted by pHaestus

Joe has a kit from maxim-ic.com that was $150 that does the same exact thing but with better temp resolution.

What sort of accuracy was the spec on Joe's kit? The maxim chip you used is +/- 3C iirc phaestus, that's not stunning but not bad either (a DD probe is likely to be +/- 1-2C and it's less accurate is it's not built in anyway).

pHaestus
01-13-02, 11:27 PM
No I used the 6657 it is +/- 1C at 60C and so for most CPU temps should be around that. The 1669 is +/-3C. Both of those numbers are with no calibration; it is possible to calibrate either one. I don't really know the details of the kit Joe is using; I think it is discussed in AMD's latest white paper about measuring CPU temps as well.

Dissolved
01-13-02, 11:38 PM
you guys are so neardy..
im jealous ;)