CPU temperature VS. system temperature

G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

First of all, thanks to everyone (on a previous thread) who suggested
a few changes on my PC's fans to improve the CPU temperature. I ended
up redirecting the fans on the side panel from "exhaust" to
"intake"... and I moved the extra exhaust fan at the back of my PC
further down, away from the power supply. The idle CPU temperature
went from 45 degrees all the way down to 37! And the peak temperature
under heavy load went from 64 degrees to 58 degrees. I was amazed at
the drastic temperature drop from simply moving the fans around.

But a strange thing happened: My system temperature went from 32, all
the way up to 42. I had thought that a better air circulation, which
lowered CPU temperature, would also lower the system temperature.
Instead, system temperature went up by a full 10 degrees.

I don't get it at all. I find it hard to fathom that the system
temperature would be five degrees higher than the CPU temperature. I
would expect it to be the other way around. In fact, I'm worried now
that the motherboard's temperature sensors are either broken, or
unreliable. The system temperature reading remains fairly constant
(+/- 1 degree) even when the Radeon 9800 Pro card and the CPU are
working full speed. Is 42 degrees too hot for a system temperature
during idle time?

Anyway, am I better off with the new setting (CPU 37-58 degrees,
system 42 degrees) or the old setting (CPU 45-64 degrees, system 32
degrees)?
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

The CPU temperature sensor is on the CPU. do you know where
the "system temp sensor" is located? Your new air-flow
pattern has probably created an eddy in the air-flow and the
sensor is in that eddy (dead air).
If you can make a side panel with a window and use some test
smoke so you can see the air-flow, you might be able to use
baffles or ducts to control the eddies. Your mobo manual
should have a diagram that shows the location of sensors.
Also, support from the maker of your case, should be able to
advise on the best location and orientation of the fans.

Remember, the temperature on the RAM and the graphics, even
the hard drive, is also important, so you need air-flow over
the entire system, not just a "system temp sensor."


--
The people think the Constitution protects their rights;
But government sees it as an obstacle to be overcome.


"Opticreep" <opticreep@yahoo.com> wrote in message
news:9e1f277e.0410040602.19e9a551@posting.google.com...
| First of all, thanks to everyone (on a previous thread)
who suggested
| a few changes on my PC's fans to improve the CPU
temperature. I ended
| up redirecting the fans on the side panel from "exhaust"
to
| "intake"... and I moved the extra exhaust fan at the back
of my PC
| further down, away from the power supply. The idle CPU
temperature
| went from 45 degrees all the way down to 37! And the peak
temperature
| under heavy load went from 64 degrees to 58 degrees. I
was amazed at
| the drastic temperature drop from simply moving the fans
around.
|
| But a strange thing happened: My system temperature went
from 32, all
| the way up to 42. I had thought that a better air
circulation, which
| lowered CPU temperature, would also lower the system
temperature.
| Instead, system temperature went up by a full 10 degrees.
|
| I don't get it at all. I find it hard to fathom that the
system
| temperature would be five degrees higher than the CPU
temperature. I
| would expect it to be the other way around. In fact, I'm
worried now
| that the motherboard's temperature sensors are either
broken, or
| unreliable. The system temperature reading remains fairly
constant
| (+/- 1 degree) even when the Radeon 9800 Pro card and the
CPU are
| working full speed. Is 42 degrees too hot for a system
temperature
| during idle time?
|
| Anyway, am I better off with the new setting (CPU 37-58
degrees,
| system 42 degrees) or the old setting (CPU 45-64 degrees,
system 32
| degrees)?
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

Opticreep wrote:

> First of all, thanks to everyone (on a previous thread) who suggested
> a few changes on my PC's fans to improve the CPU temperature. I ended
> up redirecting the fans on the side panel from "exhaust" to
> "intake"... and I moved the extra exhaust fan at the back of my PC
> further down, away from the power supply. The idle CPU temperature
> went from 45 degrees all the way down to 37! And the peak temperature
> under heavy load went from 64 degrees to 58 degrees. I was amazed at
> the drastic temperature drop from simply moving the fans around.
>
> But a strange thing happened: My system temperature went from 32, all
> the way up to 42. I had thought that a better air circulation, which
> lowered CPU temperature, would also lower the system temperature.
> Instead, system temperature went up by a full 10 degrees.

The airflow pattern changed. I can't give you a definitive answer without
being able to look at the board and box but consider that with the side fan
blowing inward it's now 'aiding' the CPU heatsink fan: I.E. moving air in
the 'same' direction. Now consider if the resulting CPU heatsink air is now
blowing more directly down onto the motherboard in a particular area that,
before, it was being pulled away from. And what if that area includes the
'system temp' sensor. It'll be blowing it's warmer exhaust air over that
sensor for less efficient cooling than before, of that spot.

There might be an additional/alternate factor. If that sensor is near the
rear exhaust fan port it's now got warm CPU air going over it to then get
OUT the case rather than outside air coming IN that same hole. Same kind of
effect, the altered airflow is putting warmer air over 'that spot'.

On the other hand, it could be the airflow changed in a completely
different manner, where ever that temp sensor is.

>
> I don't get it at all. I find it hard to fathom that the system
> temperature would be five degrees higher than the CPU temperature.

Because 'the CPU' is where 'the heat' (power) is? Yes, and no.

People often confuse heat FLOW [and power] with temperature and just say
'heat' and 'hot', as if interchangeable, for both but they're not the same
thing. Heat flux comes from the power consumed whereas 'temperature' is the
result of heat flowing through a thermal resistance, or vice versa, just as
voltage is the result of current flowing through an electrical resistance,
or vice versa.

So 'temperature' is the result of power consumed and the thermal resistance
involved in dissipating it, not the absolute amount of power, and, as a
result, a 'low power' device can get a heck of a lot hotter than a high
powered one if it isn't dissipating that 'small' amount of power well.

For example, it isn't uncommon for the power regulator FETs to be hot.

Also, consider that folks worry about how well the northbridge is cooled
but it is certainly not consuming nearly as much power as the CPU. It gets
hot because it has a smaller heatsink and the package itself isn't designed
to dissipate as much power. Less power in but the thermal resistance is a
lot higher.

The difference can be quite dramatic. Put 1 watt through a 1/4 watt
resistor and it'll melt the insides, the case, create smoke and all sorts
of disaster, I.E. get *real* dern hot, while a Prescott CPU, with it's
massive heatsink, is quite happy consuming over a hundred times more power.

temperature != power
(or else improving airflow and using better heatsinks wouldn't help)

So, just because the CPU consumes a hefty amount of power doesn't mean he's
necessarily going to be the hot boy on the block.


> I
> would expect it to be the other way around. In fact, I'm worried now
> that the motherboard's temperature sensors are either broken, or
> unreliable. The system temperature reading remains fairly constant
> (+/- 1 degree) even when the Radeon 9800 Pro card and the CPU are
> working full speed. Is 42 degrees too hot for a system temperature
> during idle time?

Your 'system temp' is probably more like 'motherboard near the regulators'
temp or maybe it's 'motherboard near the northbridge' temp.

>
> Anyway, am I better off with the new setting (CPU 37-58 degrees,
> system 42 degrees) or the old setting (CPU 45-64 degrees, system 32
> degrees)?

Well, I'd go for the lower CPU temp but I'd also want to look around and
find out just exactly where and what that 'system temp' really is.
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

> Anyway, am I better off with the new setting (CPU 37-58 degrees,
> system 42 degrees) or the old setting (CPU 45-64 degrees, system 32
> degrees)?

I think you are better off with the lower CPU peak temperature,
since the motherboard temperature, though higher, is not even
close to high enough to cause any problems. But 64 for the CPU was
getting up there.
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

I agree. My system runs at 28C for the CPU and 40C for the motherboard.
This is because I am running at 1.2 GHz on the FSB which really builds
up the heat. You can always get a chipset cooler too if you are worried
about it. Passive coolers should be fine, but you could even get an
active one (with a fan).

-----
Nathan McNulty

Al Smith wrote:
>> Anyway, am I better off with the new setting (CPU 37-58 degrees,
>> system 42 degrees) or the old setting (CPU 45-64 degrees, system 32
>> degrees)?
>
>
> I think you are better off with the lower CPU peak temperature, since
> the motherboard temperature, though higher, is not even close to high
> enough to cause any problems. But 64 for the CPU was getting up there.
 

GTS

Distinguished
Aug 24, 2003
520
0
18,980
0
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

"Opticreep" <opticreep@yahoo.com> wrote in message
news:9e1f277e.0410040602.19e9a551@posting.google.com...
> First of all, thanks to everyone (on a previous thread) who suggested
> a few changes on my PC's fans to improve the CPU temperature. I ended
> up redirecting the fans on the side panel from "exhaust" to
> "intake"... and I moved the extra exhaust fan at the back of my PC
> further down, away from the power supply. The idle CPU temperature
> went from 45 degrees all the way down to 37! And the peak temperature
> under heavy load went from 64 degrees to 58 degrees. I was amazed at
> the drastic temperature drop from simply moving the fans around.
> SNIP
Bit suprised with your statement "I moved the exhaust fan lower down, away
from the PSU", as the exhaust fans are typically placed as high as possible,
as heat rises - some people even put a top blow hole coming out of the top
of the case. Anyway, when I tried using a side input fan, because the air
flow was no longer being pulled through the front of the case, my hard
drives got a lot hotter, though the GPU was cooler. So following advice from
some websites I moved the side fan to the top rear - So I have PSU, 8cm Fan,
8cm Fan all butted together exhausting from the rear. Experimentation is the
only way to get the best set-up, but I do recommend you try all fans
exhausting from the rear too see what happens...
 

Clyde

Distinguished
Apr 10, 2004
136
0
18,680
0
Archived from groups: alt.comp.hardware.pc-homebuilt,microsoft.public.windowsxp.hardware (More info?)

GTS wrote:
> "Opticreep" <opticreep@yahoo.com> wrote in message
> news:9e1f277e.0410040602.19e9a551@posting.google.com...
>
>>First of all, thanks to everyone (on a previous thread) who suggested
>>a few changes on my PC's fans to improve the CPU temperature. I ended
>>up redirecting the fans on the side panel from "exhaust" to
>>"intake"... and I moved the extra exhaust fan at the back of my PC
>>further down, away from the power supply. The idle CPU temperature
>>went from 45 degrees all the way down to 37! And the peak temperature
>>under heavy load went from 64 degrees to 58 degrees. I was amazed at
>>the drastic temperature drop from simply moving the fans around.
>>SNIP
>
> Bit suprised with your statement "I moved the exhaust fan lower down, away
> from the PSU", as the exhaust fans are typically placed as high as possible,
> as heat rises - some people even put a top blow hole coming out of the top
> of the case. Anyway, when I tried using a side input fan, because the air
> flow was no longer being pulled through the front of the case, my hard
> drives got a lot hotter, though the GPU was cooler. So following advice from
> some websites I moved the side fan to the top rear - So I have PSU, 8cm Fan,
> 8cm Fan all butted together exhausting from the rear. Experimentation is the
> only way to get the best set-up, but I do recommend you try all fans
> exhausting from the rear too see what happens...
>
>

The case I use was a gift from my son. It has 2 fans on the back, 1 on
the top, and 1 on the side. They were all blowing out. I started running
into heat problems with my P4 3.0. When I moved to the Prescott 3.2, I
turned the side fan around to blow in. Now it blows just to the side of
the CPU fan/heatsink.

That seems to work very well for my situation.

Clyde