2 CPUs - 8 different temps!!??!?

[gas-man]

Thermal probes say CPU1: 32.9 CPU2: 36.3
MBM5 + Hmonitor say either 23 + 25, 49 + 57, or 96+ 96
Bios says whatever HMonitor and MBM say....

What the hang is going on?? I get the different software temps by either using the thermometer, transistor, or diode..

Anyone explain?

[Liss]

2nd cpu has greater work load..

how it always been like tha..

[Freeloader]

Artic Silver's website has an article about how to calibrate your temps. The Intel version is close enough. See if this can help you separate fact from fiction. You may also want to make sure that the second block isn't hung up on the socket cam.

[rogerdugans]

I've had a couple of duallie's, and worked on a few more- temps are never quite the same:) As long as they are within a few degrees it is normal- if its more than 5 or 6c TOPS than something is wrong!

My P3 1000 had about 2c difference between the two cpus at all times.

[gas-man]

I undestand that, my point is that the probes will say one is 34C but the software would say it is (same one) 25C, or 57C, or 96C....

[safemode]

thermal probes are just thermistors or thermal resistors or also known as thermal diodes. They're all the same thing.

put your system on max load for a couple hours ...touch the heatsink or water block and if it's cool then things are good. If it's hot then things aren't so good, but they're not bad. If your computer crashes during this test things are quite bad. That's the only temperature testing you can trust compared to the cheap crap they put in computers.

[Since87]

Originally posted by safemode
thermal probes are just thermistors or thermal resistors or also known as thermal diodes. They're all the same thing.

put your system on max load for a couple hours ...touch the heatsink or water block and if it's cool then things are good. If it's hot then things aren't so good, but they're not bad. If your computer crashes during this test things are quite bad. That's the only temperature testing you can trust compared to the cheap crap they put in computers.

Repeating to myself, "This is a child friendly forum, no flames allowed."

safemode you are wrong.

Thermistors are not the same thing as thermal diodes.

Touching the heatsink/waterblock and feeling that it is cool doesn't prove things are good. It may indicate that the HS/WB is not making good contact with the CPU.

If a waterblock is "hot" then things are most definitely bad.

[Since87]

Originally posted by gas-man
I undestand that, my point is that the probes will say one is 34C but the software would say it is (same one) 25C, or 57C, or 96C....

Can you tell us what device MBM says is measuring these temperatures? That info might help come up with an explanation.

Based on what you've said so far, I'd guess that the problem is bad communication between the processor and the temperature measuring IC('s). This may be caused by bad components, but an even more likely cause is a bad solder joint between a component and the MOBO PCB. Another possibility is just bad MOBO design that is coupling noise into the comm lines between the temp sensor IC and whatever it connects to.

[safemode]

Originally posted by Since87

Repeating to myself, "This is a child friendly forum, no flames allowed."


what flames?


safemode you are wrong.

Thermistors are not the same thing as thermal diodes.


Thermistors and thermal diodes are the same. If you think they're not than show me the proof.



Touching the heatsink/waterblock and feeling that it is cool doesn't prove things are good. It may indicate that the HS/WB is not making good contact with the CPU.


This is why i also stipulated that if it crashed during the tests you had a problem. If the heatsink wasn't making good contact it would overheat and thus crash. In which case you would know the cause. If it doesn't overheat and crash then you know that's not the case and touching the heatsink is perfectly valid. If the computer runs stable at full load for hours then things are perfectly fine. The exact number means nothing because our measuring tools suck and essentialy render themselves meaningless.

You might argue that the heatsink could still not be making full contact but the system doesn't become unstable so touching the heatsink still doesn't accurately tell you the temperature of the cpu. In which I would reply it doesn't matter. If the cpu runs stable at the temperature it's running at then it doesn't matter. Apparently in that case the cpu is making "sufficient" contact with the hsf. All that matters is that the cpu is stable at full load for many hours or days if you need it to, not if it's running at 35C or 40C etc. Numbers mean nothing when the tools you're using are as unreliable as making your own rulers from memory of what an inch is about.

[safemode]

Bah, i confused myself with the word thermistor and two possible names to come from it: thermal resistor (real one) and thermal transistor which would be a thermal diode. A thermal diode creates energy from heat. A thermistor changes resistance of a electric current based on heat.

[FunkDaMonkMan]

BAH! ALL OF YOU!!!



gas-man, what motherboard are you using. I've experienced the same problem with my tyan s2460. You will probably have to find a temp monitor made for your motherboard.

[gas-man]

I have an MSI K7D Master-L (known to have these problems)

I was thinking of a Hardcano 2 - see my topic in W/C

[FunkDaMonkMan]

so you have received

23 + 25 AND 49 + 57

from your bios????


if this is so.. then I have no idea what the hey is goin on, but usually your bios should give you the correct temp.

MBM can be screwy on dualies, I have to use "TyanTEMP", a program designed for tyan dualies to monitor my temps.

Another note, in the bios if it doesn't automatically refresh, try refreshing the temp because my bios will first tell me 76C, but after I refresh, It will tell me the correct temp.

:edit: by saying "refresh" i mean hitting enter or something to get another reading, not clearing your CMOS.

[Since87]

Originally posted by safemode
Bah, i confused myself with the word thermistor and two possible names to come from it: thermal resistor (real one) and thermal transistor which would be a thermal diode. A thermal diode creates energy from heat. A thermistor changes resistance of a electric current based on heat.

Wrong again.

First your statement, "A thermal diode creates energy from heat."

This is a fairly meaningless statement. Heat IS a form of energy. Maybe you meant to say, "A thermal diode converts heat to electricity." In that case you would still be wrong. Thermal diodes are used by measuring their forward biased voltage drop with known amounts of current flowing through them. You can see details here. (http://www.maxim-ic.com/appnotes.cfm/appnote_number/1057)

Maybe you were thinking of a thermocouple. A thermocouple does produce electricity as a function of temperture differentials. A thermocouple is yet another type of temperature sensor which is distinct from thermistors and thermal diodes.

Second your statement, "A thermistor changes resistance of a electric current based on heat."

An electric current does not "have a resistance". A material through which an electrical current is flowing has a resistance. You could accurately say, "A thermistor's resistance to an electric current is a function of its temperature." This may seem nitpicky, but if you ever want people to take your opinion seriously in a scientific or technical realm, you'd be doing yourself a favor by practicising writing precisely and accurately.

You seem to be suggesting that, because motherboard temperature sensors are not very accurate, there is no point in being concerned with whether the mobo temp sensor is functional. And, that the processor temperature doesn't matter, the only thing that matters is whether the cpu is stable.

This is the General Cooling section of the Overclocking section of Overclockers Forums. If someone only cares that their cpu is stable why would they be reading and writing here?

I think it's safe to presume that the majority of the people reading and writing here understand that the cooler their processor is, the higher they are likely to be able to overclock stablely.

A large number of the people here are also going to realize, that a temperature sensor that is able to tell them that their CPU is running roughly 5 degrees cooler with setup B than setup A, is of more benefit than a temperature sensor that is producing random numbers.

[Freeloader]

Gas-Man, did you check your thermal probes for accuracy before you installed them. If not, you could stick them in a warm glass of water with a thermometer and compare the results.