The confusion that is an idle temperature

[Thelemac]

I was reading through many posts here on the cooling board and i noticed that many people are worried about thier idle temps. I have a question for you. Why? Why would you worry about getting your idle temp lower? What good does that do you? Why would you even want to run your cpu idle? (Ok, I have more than one)

First, the reason that I see not to worry about your idle temps is that those don't effect your stability...since you aren't doing anything, what can be unstable? Getting it lower will just, well, lower the activity (assuming constant load temp) of the cpu at idle. Which means it's doing less. Which is fine, except it doesn't really do anything helpful for you. It just increases the amount of temperature fluctuation when you load a program. Temperature fluctuation is bad. I believe that chips like to be at a constant temperature more than they like to be at a low temperature (within reason, of course). So in my view, lowering the idle temp could actually harm the cpu.

Of course, I feel that idle temps are completely irrelevant for more reasons that just those. There is also the fact that an idle temp will probably never be the same, as there are varying loads on the processor when there is an OS running (eg system calls, sentient programs, etc, etc). Then there is the fact that i don't even understand why you would want your cpu idle. Wouldn't you rather have it active? Isn't that why you paid good money for it? So that you can use it? Or did you just want to be able to say that your processor is at ambient temperature when it isn't doing anything PH34R |V|3!

In conclusion, just say No! to idling your cpu.

 

[Slake]

Very good point Thelemac. I also wonder why we post our temps in Celcius. It seems to me that Farenheit gives a more accurate guage of temps since there are more increments to reference.

 

[Thelemac]

Well, F (no, I'm NOT going to type that out, much too long. I would much rather type a rant than type out that word...) does have more increments...as long as you just go by integers. Currently (for AMDs at least) the temps aren't really accurate to that degree due to the insocket thermistor. So C is used more than F for a couple of reasons. First is the fact that only in the USofA is F used on a daily basis...mainly due to stuborness than anything else. Second, C is one of those nice metric measurements that goes by hundreds and such. It is also used in most scientific data (usually converted to K, but that's really the same thing.) So that's why I use C. That and the fact that I think that's what my MBM defaulted to.

 

[diehrd]

Well my processor idles at system temps,27c ,Amd I personally prefer it to be as cool as it can be at idle,Because I AM NOT USING IT.

It may be of no importance to idle at 27c over 40c but heat is a killer and if your system can idle at board temp I would make sure it does.It makes no sense to me to have my processor as warm at idle as it is when running full out.

If you have a via chipset there is a regestry mod that will set you up to idle the processor at board temps.

 

[bdf24]

You are right, load temps are more important. But for the most part. The lower your idle temps are normally the lower your load temps are going to be.

 

[vandersl]

Well, lets see - my computer sits in the 3rd floor loft of a townhouse in NJ. Its probably 28C/83F in the room ATM. If I didn't have my CPU (TB 1200/266) idling properly, it'd be sitting at 53C/127F all the time. So, for the summer anyway, I won't be running SETI or doing any folding. I really don't need a 70W heater running in my loft ATM, thanks.

But thats more a reason why to let the CPU idle, not measure temps at idle. I think the reasons for that are varied. Idle temps seem pretty useless to me, other than knowing what temp my CPU spends 90% of its life at.

As for C vs F - technically, F has more resolution, not more accuracy.

 

[Thelemac]

diehrd...
Yes, heat is a killer, but you're really stressing the chip when (if) you constantly change the temp that it is running at. When the chip heats up, it expands, which causes stress on the bonds of the chip...and when it cools off, it contracts, causing more stress. This stress, to me, at least, would seem to kill the processor off more quickly than having it warm (within reason, of course).
Incidentally, if you're not using it, why do you have it running at all?

bdf24...
Lower idle temps actually really give you no indication of what your load temps will be. I could idle my chip at board all the time (using some idleing program or something like that) but change my cooling, which doesn't affect idle (still at board) but will lower or raise my load temps. Or my idle could change for an unrelated reason, such as a new sentient program or something like that, which also will not affect the load temps.

vandersl...
Ok, that's a good reason not to load your cpu...though I think (could be wrong) that the small amount of heat generated by your processor (in relation to the size of you room) isn't really going to affect it's temperature all that much. Also, 53C is generally considered to be a bit high for an AMD cpu to be running at...though if you aren't unstable, it can't be that bad for it.

That's another thing (now this is really just what I think, I have no idea if I'm possibly renting space in an alternate universe on this one)...if your cpu is stable at a certain temperature, can it really be bad for it? Having it stable means that the processor isn't having problems running it's processes, so the heat can't really be bad for it. General heat guidelines really aren't terribly helpful cause every single chips made is different from every other chip made. This is proven by the varying successes at OCing a specific make and model. So each one will have it's own heat tolerances. I guess there is a little bit of good in a general guideline...saying that every chip will at least be able to get to a certain temperature without dieing. Though really, that's more for a manufacturing guideline than for us.

 

[vandersl]

Um, dude, look at the room temp that I mentioned, and you'll realize that 53C isn't too bad a temp for a 1.2GHz TB. Sure, if I have my room at 22C, I'll be at 47C, but alas, 'tis not to be.

And no, 70W isn't gonna hurt the room temp much. But it won't help, either. And when you've been sitting in a 28C room for a couple hours, nothing seems insignificant.

 

[Thelemac]

vandersl (May 05, 2001 05:50 p.m.):
Um, dude, look at the room temp that I mentioned, and you'll realize that 53C isn't too bad a temp for a 1.2GHz TB. Sure, if I have my room at 22C, I'll be at 47C, but alas, 'tis not to be.

And no, 70W isn't gonna hurt the room temp much. But it won't help, either. And when you've been sitting in a 28C room for a couple hours, nothing seems insignificant.

The room temp doesn't really affect what temperature the cpu can handle...only what it's at. Having a high room temp doesn't mean that it's going to be ok for the processor to run at a high temp...it just means that it's going to. I do feel you on 28C being rather warm...I had to work a couple days at 35, and my room is getting to be about 30 +humidity...so I know what you're feeling there.