How hot is too hot?

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

Threaded View

How hot is too hot for an ASUS mobo, the CPU temp and the MOBO temp?

Also will adding another half-gig of RAM make the temp problem worse?
(I have them memory already but havent' been using it.)

I have an ASUS mobo, A7M266, with one gig of RAM, and an 800 MHz CPU,
not by intel, (I forget the company, but they rate the speed higher
iiuc. Amtrak? Amstel?)

My oil furnace malfunctioned for a while this past winter, and tv
screens attracted noticeable soot.  Some parts inside the computer
were probably also affected.  I've vacuumed it, but the story starts
before then.

When the hot days of summer started, I found that the computer was
overheating.   I exited winXP and it turned the computer off**.

When it was cool again, I dl'd and installed ASUSProbe.   The default
temps at which points it would flash and beep warnings seemed very
high.  I lowered them to about 4 degrees of the operating temps on
non-hot days.  So if it gets 4 degrees above that, it beeps and

But I have no idea what the temps should really be.  What would you
set your temp limits at?

For example, right now the MOBO temp is set for 41C/105F.  Some days
it's 102 degrees out and maybe 90 something in my house.  It's not a
long way from room temp until the mobo reaches 105, but is this really
an unacceptable temperature?

Similarly, the CPU is set for 156?F/70C, and it will reach that if I
have the newsreader, mailreader, and web browser open, especially if I
have a lot of tabs open, but is that really too hot.  

When it first overheated, I had no way to measure temp***. I knew
because it stopped working right.  That hasn't happened again.

**A couple weeks later, I also remnmoved the CPU fan and vacuuumed the
heat sink with a real vacuum, not one of those battery operated tiny
things.  At that point the computer temp went 10 to 20 degrees cooler,
and as long as the outside temps were in the 80's things were fine.  

I plan to vacuum again with a better attaachment (for lamp shades) but
uutil then, I'd like to know.

I"m in the middle of assembling a bigger faster computer

***The actual temp shows in the BIOS settings, but I can't look there,
because it's never that hot when I'm not running programs.  

Also FWIW it seems to run cooler when I use win98, another partition
on the same HD.

Also fwiw, it turns out the soot is not sticky and it's not oily. It
comes right off when sucked.  It even comes off when I rap on the heat
sink, although then it falls, mostly onto the bottom of the case.
Doesn't seem to leave a trace, so I must have missed some of it for
other reasons, like wind dynamics.

Any help is greatly appreciated.

Re: How hot is too hot?

mm wrote:
Quoted text here. Click to load it

There are at least two ways to measure CPU temperature. You can use
a socket measurement (thermistor located below the CPU body). The CPU
temperature limit with that as a measurement might be 65C.

Or, you can sense the diode on the silicon die, and the die gets hotter than
the socket would (under the same conditions). The limit in that case, is
85 or 90C. The 85C/90C limit has been enforced, on motherboards later than
yours, using chips such as Winbond W83L785TS-S. (As far as I know, the "offset"
feature in this one, is to make end users comfortable, with "socket-like"
temperature measurements, while continuing to enforce the silicon die temp
limit. I don't think the 3 bit offset actually affects the trip point. The design
engineer sets the resistors up, to give some idea of an offset to apply to the
value read via I2C bus, to make "socket-like" temperature readings.)

Your motherboard might be limited to estimating the temperature via
a socket based measurement. And not have any reliable automated computer
shutdown, in the event of the CPU getting too hot. (If the heatsink falls
off your CPU, it may not shut down in time, or shut down at all, leading to
a ruined CPU.) There was a generation of motherboards, that lacked adequate
CPU temperature protection, leading to the over-hyped Tomshardware video of
an AMD CPU burning up. Later motherboards came up with better methods
of providing protection. Modern AMD and Intel systems, both have
THERMTRIP, so this is a solved problem on your new build.

So it really depends on how the temperature is measured. Also, the socket
method is subject to the manufacturer applying the right correction
(offset) to the measured value, to give a sensible value. (Since fudge
factors are used all over the place, it's pretty hard to say what
is really going on. The temperature reading in the BIOS, may give you some
idea what fudge factor Asus liked, when they were designing the BIOS code.)

(Maximum die temperature 90C for a particular Athlon.)

So if the tool you're using, is pretending to measure a CPU socket
temperature, then 65C might be a reasonable upper limit (for stability
as much as anything else).

If you had a more modern board, say one with the W83L785TS-S on it, then
it may be enforcing a silicon die limit of 85C or 90C. (For different
generations of AMD processor, you may find the datasheet lists one
of those values.)

if the above seems a bit imprecise, well, it is imprecise. AMD (CPU manuf.)
does have a stated limit, as the 24309 document shows. But what your
measurement circuit is doing, is a completely separate issue. Maybe a
65C reading there, is equal to the 90C limit of the processor. Who knows...


For Northbridge or Southbridge chips, similar issues may exist. On older
motherboards, a thermistor may be used to measure the temperature of
one of them. On newer designs, a silicon diode inside the chip, is used.
They can give different readings (as one is a socket/case reading, the
other a silicon die reading).

Some examples of limits:

875P Tcase-max is  99C degrees. (PDF page 17). This is a Northbridge

ICH5 Tcase-max is 115C degrees. (PDf page 7). This is a Southbridge

The human finger, can press on a 65C surface for about two
seconds, before deciding it is "too hot". And unless your chipset
is sandwiched inside a laptop, it really shouldn't be forced to
run that hot.

The thing is, you need adequate cooling for your hard drives, and by the
time there is sufficient air moving through the computer case to cool
the drives, chances are the (properly attached) chipset heatsinks are
cool enough. Setting a limit of 65C there, should help keep the memory
controller on the Northbridge stable and error free. You might
expect to see a low value in practice, like 43C or so. And in the
case of the ICH5 (which didn't have a heatsink on it), there was
barely any heat detectable when you touched it. So no reading there
is really necessary.

You can try looking up your chipset on the AMD site, like look for
the AMD761 and see if it gives thermal info. The VIA website is a
waste of time, and they don't typically give out useful info
to end users (only under NDA). Your Southbridge is made by VIA,
and the Northbridge by AMD (same as the brand of processor).

AMD761 (Tcase-max is 85C, page 58)


You can check your hard drive temperature with Speedfan from
or with HDTune from ( )
Usually, mine are in the 35C range, and may get up to 45C or so if the A.C. is
off. (Hard drive temperature, if supported, is provided via the S.M.A.R.T
interface.) Hard drives are sensitive to both humidity and temperature, and the
closer you are to "living in a rain forest", then the closer to 35C you
should maintain the drive. You can run them hotter, if the air is
extremely dry. Some hard drive manufacturers provide a graph, if you
can find the appropriate document (that is where I discovered
they were also interested in the humidity).


Things that help your CPU stay cool:

1) Adequate case intake vents, to ensure good case cooling, so the case fan
    doesn't have to work too hard. Some computer cases, have too small an
    intake area.

2) Clean the heatsink fins (you've done that).

3) Don't bother cleaning the fan blades - I ruined a fan bearing by
    applying a little too much pressure to the fan blades. Be careful if you
    do that kind of cleaning.

4) If you know the temps shot up too high, chances are your thermal
    paste or thermal interface material, needs to be changed. Special
    care must be used, when working with bare-die processors such as
    yours. It helps if a "shim" or "rubber bumps" are present on the
    top of the processor, so the heatsink cannot be rocked from side
    to side, when putting the heatsink back. If you rock the heatsink,
    it chips the edges of the silicon die. If a crack on the edge
    of the chip, propagates deep enough, it will kill the CPU.

    Apply thermal paste sparingly, because if it oozes out, it will get
    on the resistors on the top of the processor. To do a "test pressing",
    put a half-grain of paste on the CPU die, and gently put the heatsink
    in place. You should get a round circle of paste squashed onto the
    die. The size of the circle, tells you how much additional material
    you'd need, to get complete coverage. Too much paste, is just as bad
    as too little. The AMD CPU has one of the smallest contact surfaces
    of any processor, so doing this right is critical to getting a good
    result. Processors with thermal spreaders on top, would make this
    much easier.

After re-doing the paste, you can run a benchmark like Prime95, and record
the peak CPU temp (after the paste has bedded in). Record the computer
case air temperature, and the room temperature. In later years,
you can go back to those measurements, and compare them. If the CPU
runs 10C hotter than it used to, for the same degree of case or room
temp, then it may be time to re-do the paste again.

The original thermal interface material, may be a phase change type.
It can last a long time, but may need to be replaced if you remove
the heatsink a lot. Paste, on the other hand, is not a permanent
solution, and may need to be re-done in a few years. I'm still using
the original tube of paste I bought, so a tube lasts a long time. I
prefer paste, because I know, sooner or later, I'll be disassembling
it anyway.


Re: How hot is too hot?

Quoted text here. Click to load it

My laptop shuts down abruptly at ~85 degrees Celsius for the CPU. I cleaned
it a while ago with a can of compressed air and only recently has it begun
to reach that temperature again, but only if I accidentally block the cpu
vent which is not so cleverly located on the underside.

Today is a cool day and I am only using the browser and the email client ..
the CPU is sitting at ~49 to ~54 degrees Celsius .. the cpu fan is not
'blowing' at this temperature.

If I watch a YouTube video .. The cpu temperature jumps to ~59 degrees
Celsius (today the room temperature is about 18 degrees I estimate) and the
cpu fan kicks in.
The hard drive reports ~43 degrees Celsius.

These measurements are reported by 'Speedfan 4.34'
My laptop is HP Pavilion ze2000 running XP .. on Intel Pentium M 1.60mhz
with 2GB RAM

Re: How hot is too hot?

On Fri, 30 Jul 2010 15:47:14 +0100, "123Jim"

Quoted text here. Click to load it

Wow, that's much hotter than I've been since I had the temp program
installed.  I have it set to complain at 71^C, and I've let it get one
degree over that, maybe 2 at most for very short times. Of course,
nothing has abruptly shut down, or even gone haywire, and like I say,
when it did to that, I had no way to know what the temp was.

Quoted text here. Click to load it

I vacuumed mine, with a small shop vac, and the running CPU tenp on
non-hot days dropped by 10 to 20 degrees.  The room could be much
hotter without causing problesms and indded sometimes it was the mobo
temp that went above what I had set it at 41 or 42C, something like
99F.   I still plan to use a bigger shop vac and a lampshade
attachement, the round brush with the hole in the middle. Actually any
brush that would get into the heat sink prongs would be a big help.
Myabe next time the fan is off I should just look for a brush that
will do that.

Quoted text here. Click to load it

This confused me.  My CPU is never below 60 I think, and I think the
fan blows all the time no matter what, but it's hotter than yours, and
this is an old mobo for a desktop, so it may not know how to turn off
the fan, and there is no need to save battery life anyhow.

Quoted text here. Click to load it

I couldn't watch a video when it was hot here.   It was an average of
7 degrees hotter every day in July than the previous year, which
admittedly was cooler than usual, but this June and July were much
hotter than usual.

Quoted text here. Click to load it
Thanks for your detailed answers 123.  At first they were so much
hotter than mine, I thought I was misundertanding them, and that's one
reason I didn't reply for a long time (sorry) but otoh, if I added 10
to 20 degrees to my temps (the same amount that vacuuming it made my
temp go down) it might have been your temps when it went haywire.

And thanks Paul.

Site Timeline