I need some advice about TEMPS..Please Help??

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
I am bewildered as to why my computer temps have increased so much.
I am hoping someone here would be able to either explain or tell me
how to fix this issue.
I have provided links below so you can see the JUMP in temps since
February 2006.

In Feb 2006, Just after applying arctic silver:

June 2006, Arctic silver still being used
but look at the temp increases?

The ambient temp in the room has, at most, increased by 3 or 4C
Currently my ambient temp is 25.5C
Current ambient temp in other rooms 23.8C
Running Average CPU Temp 40C


Re: I need some advice about TEMPS..Please Help??

Dr.PedroSanchez@yahoo.com wrote:

Quoted text here. Click to load it

You used two different versions of Speedfan ?

Vcore in one is listed as 1.71V in one picture, and 1.49V in the other ?

I would say your methodology is a bit flawed. The experiment
should be repeated by using the same version of Speedfan when
making your pictures.

The "goodness" of a CPU cooler, is measured by taking the
difference between the CPU temperature and the case temperature.
I'll illustrate with an example.

Let us say I have a P4 2.8E Northwood processor. My room temperature
is 25C. My computer case temperature is 32C. My CPU temperature
is 52C. According to processorfinder.intel.com, my processor's
power dissipation rating is 89 watts.

First of all, a well cooled computer case has a difference of
7C between the room temperature, and the computer case air
temperature. You can have a higher difference, but don't let
the computer case get too hot, as that is bad for the hard
drive. You would need some good fans on the computer case, plus
plenty of open ventilation holes, to achieve 7C temperature

The CPU cooler cannot cool the CPU below the computer case
temperature. If the computer case temp is 32C, the processor
can never be cooler than 32C by just air cooling.

The "goodness" of your cooler, is measured by the difference
between the CPU temperature and the case temperature. In this
example, 52C minus 32C is a 20C degrees difference.

The other factor, is the amount of heat the processor puts out.
When the CPU runs at 100% loading (Prime95 or CPUburn), we
know the processor is rated to waste 89W of electricity. This is
only an estimate, as there will be unit to unit variation. My
measurement of my processor, proved the TDP number was
surprisingly close to the Intel stated value.

If we divided the temperature difference, by the power, we
get a number in degrees_C per watt. From our example, we take
20C / 89W = 0.225 C/W. If I go to the Zalman web site, and
look up the thermal resistance theta_R for my CNPS7000 AlCu
heatsink, it is rated at 0.22 C/W when the fan is running at
full speed. Therefore, my thermal grease is only adding
0.005 C/W to the theoretical 0.22C/W value.

To state it another way,

  CPU_temp = room_temp + case_delta + (theta_R * TDP_watts)

           = 25C + 7C + (0.22C/W * 89W)

If my room temperature changes, you can see that the result would
be a direct change in the CPU temperature. If I made my computer
case cooling worse, by accidently covering the vents, or slowing
some of my computer case fans down, my case_delta will rise above
7C, and again, my CPU_temp will rise in response.

If the thermal grease becomes worse for some reason, the thermal
resistance will rise above 0.22C/W. Both the resistance of the
thermal grease, and the resistance of the heatsink design (fan
speed is a factor), contribute to the 0.22C/W number.

If the CPU power dissipation increases, that can cause a temperature
increase. One way to accidently increase the CPU power, is to
increase the Vcore in the BIOS setup. Even changing BIOS versions
has been known to cause the Vcore value to change - if there is
a BIOS bug.

Now, having said all that, the temperature of the human body is
37C. Your CPU is at 40C. Your CPU is hardly in any trouble.

Improve your experimental methodology. Use the same measurement
tools and techniques for each measurement. Measure room temp,
computer case temp, CPU temp, while running a 100% CPU load. Use
the same testing program to load the CPU each time. Check your
Vcore and operating frequency, as CPU_power = F*C*V**2, where
F is the core frequency, and V squared is the Vcore voltage
value squared. Make sure you are using the same Vcore and
operating frequency during each measurement.

If the experimental conditions are not carefully controlled,
you'll get erroneous results.


Re: I need some advice about TEMPS..Please Help??

On Sun, 04 Jun 2006 23:45:30 GMT, nospam@needed.com (Paul) wrote:

Quoted text here. Click to load it

Dear Lord, that is a lot of information..I will use it but I am a
little slow so it might be a while before I reply..Please keep/watch
this thread if you can

Thanks paul!

Re: I need some advice about TEMPS..Please Help??

Quoted text here. Click to load it
Or to cut a long story short, there's nothing wrong with your CPU temps,
don't worry about it.


Re: I need some advice about TEMPS..Please Help??

Quoted text here. Click to load it

I gave you the information, because I want you to understand that
you have to be careful to set up the same test conditions. We are
only talking about small temperature differences, like maybe your
performance is worse by about 2C or so. I trust your Starcraft
temperature measurements, more than your idle measurements, because
your idle CPU may be doing more work in one case than the other.
I recommend comparing performance with a known 100% CPU load.

I would not put a lot of trust in the CPU diode temperature measurement.
The accuracy of the diode is around +/- 8C or so, and that is when the
correct measurement technique is used. The monitor that Asus uses
use the correct measurement technique (it uses only one sense current,
while AMD recommends alternating between two currents, to subtract out
some of the error.) I have also seen reports that the diode can be
by attempts to overclock.

At this point, if you want to continue experimenting, I'd suggest you
carefully make some measurements with your current setup. Then,
the CPU heatsink. Wipe off the excess material. Clean off the residue
propyl alcohol. Let the surfaces dry. You don't want any solvent left on
the surface, to dilute or alter the properties of the Arctic Silver.
Apply some fresh Arctic Silver and reassemble the heatsink.

Then, over the next couple of days, make some more measurements. Use
the same 100% CPU load, and allow enough time for the temperatures in
the case and the CPU to stabilize. That can take a while to do properly.
You should see the performance improve as the Arctic Silver "beds in".
In about a year's time, you may notice that the performance has
degraded. Thermal grease/paste is not permanent, and it will have to
be reapplied any time that you find the performance level has dropped
below what you find acceptable.

Don't forget to record: Room temp, case temp, CPU temp, with a 100%
CPU load. For a load, you could use one of these programs. (Scan them
with a virus scanner before using them, just in case. I haven't used
CPUBurn in quite a while.)

http://users.bigpond.net.au/cpuburn /
http://pages.sbcglobal.net/redelm/ (I think I may have used this one.)

Your Speedfan screen lists a bunch of temperatures. Have you figured out
what each one is ?

I have a machine here with AS3 on it, and the performance is worse than
when I originally applied it. But the machine isn't bad enough to worry
as the CPU is only running 43C at full load. If it hit 60C, I might
take it apart and fix it :-)


Re: I need some advice about TEMPS..Please Help??

Quoted text here. Click to load it

I'm going to fire up starcraft again for a few minutes and see what I

Re: I need some advice about TEMPS..Please Help??

Ok, I ran starcraft for 15 solid minutes. The CPU highest temp was
52C, motherboard was 40C, ambient room temp is 81F

So now what to do?

This sucks because I bought into that Arctic Silver hype..

But someone reply please. It's hot in here now..

Re: I need some advice about TEMPS..Please Help??

On Sun, 04 Jun 2006 23:49:13 -0400, Pedro Sanchez

Quoted text here. Click to load it

1)  Your idle temp is not excessively high.  We could ponder
why it is different but in the end it doesn't "really"
matter, it is low enough.

2)  Starcraft does seem more of a load than previously, but
is still a variable to resolve.  A system needs be under the
critical temp threshold when as near a sustained full load
condition as possible.  Prime95's Torture Test, large
in-place FFT's setting may come closer to this... try
running it for at least 30 minutes, and also watch to be
sure no errors occur which stop the calculations, as this
indicates instability which should be resolved.

3)  Arctic Silver isn't necessarily a magic bullet, but it
is still desirable to use because (at least in the latter
versions) it  is a synthetic base (oil) compound, meaning it
will hold up better over time.  Not having to change the
thermal compound even once over several years' time is
enough reason to buy some... and have it to use again and
again since even the smallest tube is enough compound for
dozens of applications.

If you are going to be continually troubled by this, take
off you CPU, clean off the compound from it and the 'sink,
then reapply it in a very thin coat in the middle only.   It
typically takes a few thermal cycles (days and power cycles)
for AS to reach the most effectiveness.  Again, don't be
concerned about idle or partial load temps, the peak the
system could ever attain is the main concern.  Under 50C in
everyday use is fine, and to some it would even be
preferrible to running at a lower temp but having louder
fans to do so.

Re: I need some advice about TEMPS..Please Help??

Quoted text here. Click to load it
Those temps are fine. Even in a room at 72f, they would not be extremely
high. For example, my EM64T Intel @ 3.6ghz with a room temp of 72f idles at
37c and under a load runs at 50c. This is well within specs for the
processor. Even the HS fan doesn't kick into high speed until about 57C, so
my system stays very quiet too. You have to keep in mind that in a room at
81f, you can expect about 3-5 degrees higher CPU temps than in a room at
room temp (72f).


Re: I need some advice about TEMPS..Please Help??

Dr.PedroSanchez@yahoo.com wrote:

Quoted text here. Click to load it

Your motherboard is 40C. Your room temp is 27.2C .
The case to room delta is 13C, while a well cooled case would be
7C to 10C.

Your CPU to case delta is 12C. Compare the June difference,
of 52C - 40C, to the Feb difference of X - Y (whatever
the Feb values were). If the difference is 12C in both
cases, it means your CPU cooling performance has not changed.

It is not the absolute temp that determines the effectiveness
of the cooling, it is the delta that rates the performance.
The absolute temp is important for establishing limits, like
defining a temp that is too hot for a disk drive, or too hot
for a processor. But to figure out whether the cooling ability
of either the case or the CPU has changed, that is measured
by taking deltas.

Looking at the picture of your computer case on Newegg, I see
there is room for fans near the bottom front of the computer



Site Timeline