Is My Cpu Temp Too High????

[TheWesson]

CCW,
Is that based on personal experience or just assumed "common sense"?
After extensive testing on my case, I found that having all of my fans act as intakes and leaving the exhaust up to passive venting worked the best for me.

I have to chime in and agree here.

I'm not totally sure, but I believe where there is considerable impedance (small vents or a cluttered case) you might be better off with an intake and an exhaust fan. Axial fans can't fight impedance (air resistance or back pressure) well.

However, if the impedance is low, you should get the best airflow by having all your fans to either intake or exhaust - if they are mounted where it is sensible to do this.

You'd want to plan your airflow and make your vents so the air goes where you want it of course ...

the wesson

 

[TheWesson]

MY cpu temp (according to PC PROBE) is 38°C at idle and at max load (using prime95 in torture mode or counterstrike) temps are reaching 56°C, Is this too high, also considering I just yesterday replaced the stock intel fan with a 92 mm zalman with an enormous heatsink????? could i have messed up when I put the new fan on???

from browsing reviews, I find that 18C between idle and load is unusual but not unheard-of.

56C load for the Zalman 7000 is unusual but not unheard-of.

38C idle makes me not suspect case ventilation as the culprit.

The thing that would raise a red flag is, the Zalman is doing no better than the stock Intel CPU heatsink. If that's so, then you need to check some things:

Heatsink properly seated?
Heatsink properly goo'ed with a thin layer?
Previous carp from Intel heatsink cleaned off the CPU?
Zalman fan on "quiet" mode somehow by accident?

also, if you have the Zalman heatsink which has the fan off to one side on a bracket, I wouldn't expect great cooling from this one. But it sounds like you have the 7000, the big bowl-shaped HS where the fan lives in the bowl. That one is supposed to be pretty good (if not the very very best.) Better than the stock Intel by at least 5-10C ... !

the wesson

PS 56C is no great cause for alarm tho I believe overclockers would prefer it lower if possible.

 

[aaroncat]

Hey guys I just went and bought 3 fans today and some arctic silver goo.

I put 2 80mm fans in the front of the case blowing in and a 60mm (all my case will allow) in the back blowing out and my 2 existing fans (80mm) are on the side also blowing out

I took the existing goo off the zalman 7000 and the cpu and applied the arctic silver.
I have seen dramatic temperatures drops and it has passed prime95 at 2.88GHZ (160 FSB, 1.55Vcore) where it did not before

Is it important to let prime 95 run for a specific time? I had it running for like 2hours and it passed every test then I stopped it.

I dont believe that the cooling could affect the resut of the torture test (prime95) but I will try and go higher now, maybe raise the FSB to 167.


Also how do you tell if you need to raise the Vcore???? I mean if I had the FSB clocked up and left the Vcore down at 1.5 how would i know if i needed to raise it (as the heat of the cpu and MB seems to be somewhat proportional to the increase in Voltage.
Would it fail prime95 if it had insufficient voltage???? I am scared of having higher voltage than necessary!!!!! please help

 

[aaroncat]

by the way the cpu is 33°(idle) and 52°(load)
the motherboard was the big difference where it idled at 32° and with load was only 5°hotter, before it raised up to 45°

 

[TheWesson]

the accepted wisdom as I understand is this:

Your computer will crash at a certain speed + temperature. For the stock speed that temperature is quite high (70C?). When overclocked, it will crash at a lower temperature. The reason is that transistors work better (can switch faster) at lower temperatures, and when you overclock, you're asking the transistors to switch faster.

Almost all CPU's can be overclocked some (10-15%?) because they are designed to keep working even when in a 40C case with the stock heatsink. So at lower temperatures (which you are providing) they have some headroom for extra speed.

At "minimum" temperature (-40C?) it's claimed that all CPU's can be run [at least] 33% faster.

Raising the voltage also helps prevent it from crashing at a higher clock speed. I don't know the details why.

Unfortunately, raising the clock and the voltage both produce linear increases in the heat generated, which must be dissipated somehow. (Raising both volt and clock produces a quadratic increase.)

The usual suggestion for raising voltage is to raise the clock until the CPU becomes unstable (running Prime95 or some other torture test), then raise the voltage in minimal increments (0.025 v, perhaps) until the CPU is stable again.

You know you need more voltage when
a) you raise the speed
b) the temps are under control
c) and you get errors (system or program crash) anyhow.

For my AMD CPU, the max "approved" (by AMD) overvoltage is 0.1 volt. Some people have gone much higher with much better cooling - .25 volts increase or more.

...

For a decently ventilated case, twenty minutes will show almost all of the temp rise you are going to see. Max temp might be 1-2C higher at most.

...

the wesson

 

[CrashOveride]

by the way the cpu is 33°(idle) and 52°(load)
the motherboard was the big difference where it idled at 32° and with load was only 5°hotter, before it raised up to 45°

That is an indication that you probably need better case cooling, the raise in tempurature of the air going into the HS is probably causing the higher temps. I would say anything over 50C is too hot.