does temp effect performance

I have always thought that since a CPU is set at a certain clock speed, so that it will always run at the same speed no matter what the temperature is.

one of my friends says he is going to get a new CPU fan so he can lower the temperature is increase performance, and he is not an overclocker. He thinks that having his CPU colder will make it go faster..


I told him that it won't effect performance, and that the only reason we have fans is to prevent over heating but he doesn't believe me..


what do you guys think?

thanx in advanced,
Byan

In a sense, cooler cpu's perform better. But not the way that was explained to you by your friend. No fan or heatsink or heatsink grease will increase the speed of a cpu, these things combined will as I said cool the cpu and therefore make it perform better so it doesn't overheat, thus causing problems.

Photolady is correct in that the CPU will not run faster.

Don't know about AMD's but the Intel's have had thermal management that will slow down the CPU if it is running too hot.

HERE IS INTEL's Link A bit heavy but explains for example the old PIII method.

just as I thought.., and his temp is at 36 C, and he is unhappy cause it used to be a 30 C

I told him that his temp is well under what a ussual temp is and he shouldn't touch it.., unless he wants to overclock it..



this is the same friends who I am unable to convince that switching a 20 ft LAN cable to a 15 ft cable isn't going to be able to effect speeds in any noticable way
epecially if the weakest link is already the DSL connection

oh well...

thanx again,
Byan

Byan, some friends just don't listen because they think they know everything.....;) AMD's don't have a built in shutdown to lesser speed to save them as Intel does.

How many internal fans does this friend have in his case. He should have at least one in front of the case blowing air in, and one high on the back of the case for exhausting. Tell him to check for airflow if he's worried about his temps are too high.

oh, I think he's got that covered.., he has 2 fans in the back and 2 in the front, and one on the side...

and he does have an AMD..

Putting aside the P4's thermal management stuff (which is a whole different ballgame)... As long as a cpu is operating within its design specifications as far as operating temperature goes there is no change in performance at different temperatures within the range. The processor runs at the same speed, processes the same number of instructions per second. Once the operating temperature starts to reach and exceed the high end of the its tolerable range what happens is that the integrity and timing of the electrical signals and operation of the logic gates degrades, and this produces internal errors. In the absolute luckiest of cases these errors might be survivable if the temperature spike was only momentary. But this would be unusual. Generally things go from a point of "ok" to "not ok" very fast (nanoseconds), and the upward trend is in almost all cases not reversible quick enough for the cpu to continue processing (too many errors). So, really the only point at which one might say that temperture causes a change in performance is at this "break point". But again, if this is reached and when this is reached things are pretty much at a point of no return.

So, accept it or not, your friend is wrong if he things he is gaining any performance by lowering his temperature from 36 to 30C (well within specs). In fact, he is not really even having a substantial or meaningfull effect on the lifetime of his processor. But if it gives him a warm fuzzy feeling to get things back to 30...