Validity of Posted CPU Idle/Load Temperatures

[DNA]

As with most, I‘ve viewed dozens of posts that note CPU idle and load temperatures and I question if they have any real comparison value. Hopefully we have some real subject experts out there who would be willing to share their knowledge on the topic with us.

[nikhsub1]

No, no real value. Especially boards that read with insocket thermistors, those are the worst and are good for nothing. Boards that read ondie, are OK at determining relative differences between blocks, but in no way are 2 systems able to compare temps to any accurate degree. Citing temps is like pissing upwind, trying to compare temps with others is like pissing up a rope, you get the idea.

[|kbn|]

I thought ondie temps were accurate enough for real comparasons?

So is it CPU specific or board specific or both? What if one used a maxim 6557 chip for the on die temps and same cpu?

[bigben2k]

It's just plain wrong.

Bill outlines why in his article, here: http://overclockers.com/articles638/

In short, you may not be able to reproduce the same power level, AND there may be a variation on the temp reading by as much as 10 deg C.

[pHaestus]

Don't listen to these clowns; the numbers generated from mbm are perfectly valid and objective and reproducible.

[Cathar]

Originally Posted by pHaestus

Don't listen to these clowns; the numbers generated from mbm are perfectly valid and objective and reproducible.

Step away from the hydrogen peroxide Derek.

[bigben2k]

Originally Posted by pHaestus

Don't listen to these clowns; the numbers generated from mbm are perfectly valid and objective and reproducible.

LOL! The Star Wars action figure shot backs up your statement with elegance...

[redleader]

Originally Posted by DNA

As with most, I‘ve viewed dozens of posts that note CPU idle and load temperatures and I question if they have any real comparison value. Hopefully we have some real subject experts out there who would be willing to share their knowledge on the topic with us.

They're completely worthless, and always have been. There are about a dozen people in the whole world who's numbers are worth much of anything for benchmarking.

Obviously if someone says "my CPU overheated with X" or "temps went down with Y" there is some value to those statements, however they're actual numbers are virtually useless.

[Butcher]

My CPU temp as reported by MBM used to vary by up to 10C just by rebooting. And that with an on-chip diode...

[BillA]

well Butcher, that's 'cuz yu dinna read pH's mobo calibration guide
get hip

[Butcher]

I don't really care for mbm temps per se bill. I only use them to gauge whether something has died - if my CPU temp hits 70+C in mbm I can pretty much say something in the cooling system is toast. Other than that stability is my deciding factor on whether something is cool enough - if it don't crash, it ain't too hot.

[BillA]

"if it don't crash, it ain't too hot."
wish more understood this
(I was j/king btw)

[Butcher]

Heh, hard to tell at times - humour often doesn't convey well in text.

[TaTs]

Ok guys you lost me here.
Just finished my first block and got some temperatures using MBM, ambient 25,6ºC idle 30ºC full 37ºC, ASUS A7N8X-X, are those numbers completely wrong? I'm considering to buy a new multimeter with a temp probe, is it worth the investment to get some more accurate temps?

Thx.
_______
TaTs

[Butcher]

Do you need accurate temps? As an end user I don't. A simple "not too hot" / "too hot" display does me fine and MBM can do that. Unless you have a need for accurate temps, I wouldn't bother.

[Bugsmasher]

Quick question....

Even though you cant trust the absolute value returned by a motherboard can you perhaps use the delta between idle and load as a tool to guage changes? By load I mean using a program designed to put as near of a 100% load on the system as possible. If you can fairly accurately determine any changes in ambient surroundings (temps primarily) is it a valid option to use the difference between idle and load as a litmus test for changes in a system?

[BillA]

if you ever do actually measure the air temps in the 4 quadrants of the fan inlet (per the Intel specs), you will be shocked at the range of temps, and the resultant average value

so if your 'assumed' (or half-assedly measured) air temps are crap, where can you even hope to go
the other temps are worse in terms of their accuracy
GIGO

the Overclocker's Mantra

[DNA]

Originally Posted by unregistered

if you ever do actually measure the air temps in the 4 quadrants of the fan inlet (per the Intel specs), you will be shocked at the range of temps, and the resultant average value

so if your 'assumed' (or half-assedly measured) air temps are crap, where can you even hope to go
the other temps are worse in terms of their accuracy
GIGO

the Overclocker's Mantra

What method would you suggest one use to calculate the approximate efficiency (C/W) of a water cooled system? I do have an in-line water temperature sensor on the output of the pump and others to monitor air.

[BillA]

temps as you can (reasonably) measure them are for your enjoyment
as a basis for numerically your system to someone elses ?
forget it, its the e-penis thing