Which reported temps do I use?

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

Threaded View

hello I have just replaced my northbridge fan with a heatsink, and the
airflow around it is not so good. Kony said it should be OK, and I trust
him, but I want to keep and eye on the temps. I have installed PCWizard
2008 and this gives me a ton of info, but I'm not sure which temp is
which, and some seem to contradict!

This is my basic setup:
Quoted text here. Click to load it

I have CoolnQuiet running. I have one 8cm fan blowing out, the fan of
the Enermax modu82+ PSU blowing out, and 3 little fans blowing into the
case via a free 5.25" bay. There is no chassis fan blowing in to the case.
There is a blanking plate next to the graphics card missing.
My temps are (as reported by PCWizard):
Hardware Monitoring :    Winbond W83627THF
Voltage CPU :    1.07 V
+3.3V Voltage :    3.20 V
+5V Voltage :    5.00 V
+12V Voltage :    12.52 V
Processor Fan :    1155 rpm
Chassis Fan :    1896 rpm
Processor Temperature :    35 C
Mainboard Temperature :    28 C
Power/Aux Temperature :    18.5 C
ACPI Thermal Zone :    21.8 C
Processor :    Thermal Diode
AMD Athlon 64 (Diode) :    25 C
NVIDIA GeForce 8500 GT :    nVidia Driver
GPU Temperature :    59 C
Hard Disk Monitoring :    S.M.A.R.T
Hard Disk WDC WD2500YS-01SHB1         :    36 C
nVidia Monitoring :    NVSU Interface
Processor Temperature :    35 C
Mainboard Temperature :    18 C

HDDHealth reports the same temp for disk +/- 1 deg, so that is OK. MSI
CoreCenter reports all the voltages pretty much bang on (vcore 1.06v,
3.3v, 5.0v and 12.2v), unlike PCWizard. It reports CPU temp as 35 deg
and system temp as 28deg.

1) WHat is
Power/Aux Temperature :    18.5 C  this seems to jump 2-3deg sometimes, and
ACPI Thermal Zone :    21.8 C

2) The CPU temp and the CPU diode temp... which is more accurate? There
is 10deg difference!

3) Also the mainboard temp, it's reported as  18 deg by NVSU and 28deg
by Winbond?

4) Is the system temp just the temperature of the air in the case? or is
it a component that's being measured?

5) And which, if any, is the northbridge chip - this is the most
important question!

6) Oh and the GPU temp!! Is that normal??

Thanks as usual for any help :)


Re: Which reported temps do I use?

S.Boardman wrote:
Quoted text here. Click to load it

I downloaded a user manual, and it mentions the CoreCenter utility.
The temperature display there has room for three values

    CPU Temp   41C
    Sys Temp   34C
    NB Temp    N.A.

The "N.A." designation, suggests a valid third readout is not available.

To start, you can stick your finger on the heatsink of the Northbridge.
My assumption is, the heatsink is making good contact, and there is a
small amount of thermal paste or a thermal pad there, to conduct
heat well. Using too much paste, can make a mess. Or, depending
on the design of the chip (like if resistors are present on the top
surface of the chip), in extreme cases it is even possible to
compromise something electrically. It helps if a paste product
is present, but only enough to keep the hot surfaces in contact.
(I "calibrate" my paste usage. Apply less than you think will be
needed, "squish" the heatsink into it, and determine the percentage
of surface area covered. Adjust the final "dose" for a more
complete coverage.)

So, if the heatsink is properly applied, we can use a "common sense" test.

Stick your finger on the heatsink fins. Next, touch the base of the
heatsink, since the temperature there will give some indication of
how hot things are getting.

If you cannot hold your finger on there for very long, the heatsink
would be at 65C. If it is only luke warm, you're probably OK. If
you get burned, the chip is too hot.

A number of chips now, have a thermal diode inside. And that can be
connected to a monitor chip. As long as the monitor chip is set to
"diode" rather than "thermistor", then there is a chance it would
give a representative temperature from that diode. Some monitor chips
have three channels, so there is an opportunity to measure three things.
If the MSI utility says a channel is "not available", then perhaps MSI
knows something is not connected. It may not be possible for
some other third party utility, like Speedfan, to know something
is not connected.

On a channel which is not equipped with a sensor, sometimes a
manufacturer will connect a 10K ohm resistor. And that would
result in the readout indicating a stead "25C" value in any
software tool that attempts to read it. So even when you see
a number displayed on the screen, sometimes the channel is
"nullified" by having it measure something which is basically
a constant value. If the manufacturers chose to document
their choices, these issues would be much easier to deal with.

If you really had a Northbridge sensor, cupping your hand over
the Northbridge heatsink, such that it didn't have airflow for
a few seconds, should cause the monitor readout to change.
If it isn't changing, then chances are there is no real
measurement capability there.


Re: Which reported temps do I use?


Quoted text here. Click to load it

What I meant was, if the video card is running at stock
speed, it'll tend to stay stable at temperatures higher than
we'd like to see for other parts like the CPU.  That's no
guarantee it isn't overheating, but 73C isn't very high for
a video card, might be about what manufacturers target to
allow them to save a few dollars on the heatsink.

Re: Which reported temps do I use?

On Fri, 01 May 2009 12:04:25 +0100, "S.Boardman"

Quoted text here. Click to load it

Video cards, providing they're still at stock speed, do tend
to run quite a bit hotter than other parts.  It's common to
see them getting to 80C although some of the nVidia 8xxx
series chips had an underfill problem that could cause them
early demise if they get much above 70C, though that tends
to cause screen artifacts or total death not just a lockup.

Could be the video though, but the only indication so far is
a temperature that seems typical.

Quoted text here. Click to load it

There are other things you can do.  Take a couple L-brackets
from the hardware store, scew them to two holes on a 80mm
case fan, then mark on the case floor where you want to
mount it.  Next drill out holes in the case floor and
nut/bolt it down.  Just make sure the bolt head is shorter
than the feet on the case so moving the case around doesn't
cause the bolt head to scratch the desk or whatever it's
sitting on.

Re: Which reported temps do I use?

S.Boardman wrote:

Quoted text here. Click to load it

whats the temps after doing a stress test

Re: Which reported temps do I use?

S.Boardman wrote:
Quoted text here. Click to load it

Was playing cod4 and by grapics went funny and i had about 5 secs to see
what the temp was, before my system froze. I already has a 7" deskfan
right on the northbridge heatsink which was lukewarm, and it was also
going on video card. The temp which had been reported 73 deg at full
load now read only 60!
So the lastest problem can't be heat related. A friend suggested the
high temp either from the card or from the close hot heatsink may have
damaged the video memory. He did a remote onto my system, downloaded
some nVidia tool, made fan faster for cooling and lowered the clock speed.

With the same desktop fan running as before, i managed to play an entire
match of cod4 without the distorted graphics/crash.

What do you think the likelihood is the video memory is faulty? Should i
get a new card?


Re: Which reported temps do I use?

On Sat, 02 May 2009 00:35:03 +0100, "S.Boardman"

Quoted text here. Click to load it

Run ATITool, leave it in the foreground with the cube
spinning and after a half hour click "scan for artifacts"
and let it run another half hour.  If there are any errors
or lockups they should be evident, though it is possible
certain games could stress the video card more than this,
but generally it is a good indicator.

Site Timeline