Newbie: 3 lots of CPU temp in SpeedFan? 2 of them are overheating ...

[Dundonald]

Hello,

I have a problem with my new Mesh computer where it keeps switching
itself off. I bought this PC. Last time I had a new PC I built my own
back in 1998 so things have moved on since then with CPUs.

I haven't called Mesh support because to be honest I'll probably be
quicker diagnosing this myself, either that or I'll definately learn a
bit before I do call Mesh.

Anyway to my questions -

I'm thinking the problem of the computer switching off is due to an
over heating problem somewhere, so I've been doing some googling and I
found out about and installed speedfan. Very good software. It has
identified 3 lots of temperatures other than the Hard Drive, Temp1,
Temp2, Temp3. My question is, is the CPU temp split in to three in
speedfan?

The second question is related to a test I ran. I have some video
creating software that is obviously CPU/memory intensive. When I use
this software it very often shuts down shortly after starting a video
creation. So I ran a test but this time monitored the temperatures in
SpeedFan. Sure enough the temperatures shot through the roof, but
strangely, Temp1 temperature remained roughly the same (37 degrees
increasing to about 40 degrees), but Temp2 and Temp3 went from 37
degrees each straight up to 50 degrees and still increasing so I
stopped the video creation software. As soon as I stopped the video
creation software the temperatures (Temp2 and Temp3) went straight back
down to about 37 degrees.

Can anyone hopefully help me understand what's going on here?

Thanks.

 

[kony]

Hello,

I have a problem with my new Mesh computer where it keeps switching
itself off. I bought this PC. Last time I had a new PC I built my own
back in 1998 so things have moved on since then with CPUs.

How long have you had it? Did it seem to do this from the
very start (of use at higher load)?

I haven't called Mesh support because to be honest I'll probably be
quicker diagnosing this myself, either that or I'll definately learn a
bit before I do call Mesh.

If it's brand new still, I wouldn't hesitate to call them
any longer, if nothing else this is a documentation of the
problem on a new system should it come to the point of
having it replaced it is better that this was documented
right away, that it shipped with a *defect* per se rather
than the typical OEM presumption that it must've been a
customer/user change causing a problem (which we can't rule
out as of yet but you make no mention of anything you
changed that might cause this).

Anyway to my questions -

I'm thinking the problem of the computer switching off is due to an
over heating problem somewhere, so I've been doing some googling and I
found out about and installed speedfan. Very good software. It has
identified 3 lots of temperatures other than the Hard Drive, Temp1,
Temp2, Temp3. My question is, is the CPU temp split in to three in
speedfan?

First, you can't just install Speedfan and assume correctly
reported temps, some boards use offsets or the sensor to
part correlation is mixed up. Double-check and compare with
what your bios reports in it's hardware/health monitor page
(presuming there is one).

The second question is related to a test I ran. I have some video
creating software that is obviously CPU/memory intensive. When I use
this software it very often shuts down shortly after starting a video
creation.


So I ran a test but this time monitored the temperatures in
SpeedFan. Sure enough the temperatures shot through the roof, but
strangely, Temp1 temperature remained roughly the same (37 degrees
increasing to about 40 degrees), but Temp2 and Temp3 went from 37
degrees each straight up to 50 degrees and still increasing so I
stopped the video creation software.

I wouldn't call that shot through the roof, 13C change is
fairly minor, and typical... what you just described is
pretty common with any system and not a cause for concern in
and of itself... providing the temp report is accurate at
50C.

As soon as I stopped the video
creation software the temperatures (Temp2 and Temp3) went straight back
down to about 37 degrees.

Can anyone hopefully help me understand what's going on here?

Check your bios for a CPU shutdown temp setting- it might be
set too low.

Open the case and point a desk fan at it, too see if this
reduces temps and if so, if the system still crashes. We
can't determine yet if it is temp related as 50C is nowhere
near hot enough to be a problem. Any system should still be
stable up around 65C, sometimes (often) even higher unless
it's overclocked.

Check that all fans are running and heatsink looks to be
installed correctly - unless there's some issue of voiding
warranty if you open it, I'm unfamiliar with their policies.

 

[Dundonald]

How long have you had it? Did it seem to do this from the
very start (of use at higher load)?

Only a couple months and aye it's been like this from the start.

If it's brand new still, I wouldn't hesitate to call them
any longer, if nothing else this is a documentation of the
problem on a new system should it come to the point of
having it replaced it is better that this was documented
right away, that it shipped with a *defect* per se rather
than the typical OEM presumption that it must've been a
customer/user change causing a problem (which we can't rule
out as of yet but you make no mention of anything you
changed that might cause this).

Point taken I was planning to and will call them (I'll do so tomorrow).
I just wanted to try and figure things out for myself also.

First, you can't just install Speedfan and assume correctly
reported temps, some boards use offsets or the sensor to
part correlation is mixed up. Double-check and compare with
what your bios reports in it's hardware/health monitor page
(presuming there is one).

I ran a test again with the video software but this time I didn't stop
the video processing instead I just let it run whilst monitoring the
temperatures in SpeedFan. I knew and expected the computer to crash
when the temperature got too high. What I wanted to check was at what
temperature it would crash. The temperatures reported in SpeedFan rose
to 60 degrees celcius then the computer crashed. When trying to
restart the machine I checked the bios and the temperature in there was
reported at 50 degrees celcius. So either the CPU cooled by 10 degrees
whilst the machine was off for a minute (and I was trying to restart
it) or SpeedFan and the BIOS are a little bit out of sync when
reporting the temperature. Incidentally the BIOS reported the mother
board temp at around 37 degrees C.

I wouldn't call that shot through the roof, 13C change is
fairly minor, and typical... what you just described is
pretty common with any system and not a cause for concern in
and of itself... providing the temp report is accurate at
50C.

When I ran a second test the computer rebooted at about 60 degrees
celcius.

Check your bios for a CPU shutdown temp setting- it might be
set too low.

Good point, but unfortunately I went through every option in my BIOS
but I couldn't find anything that would signify a shutdown temperature.
I did see a target temperature for CPU (set to 50) in the BIOS and I
think this is used to control fan speeds. I did see something call AMD
cool n'quiet but not sure what that is, but even the option for that
was either disable or enable. Is there definately such an option? If
so I'll have to keep looking.

Open the case and point a desk fan at it, too see if this
reduces temps and if so, if the system still crashes. We
can't determine yet if it is temp related as 50C is nowhere
near hot enough to be a problem. Any system should still be
stable up around 65C, sometimes (often) even higher unless
it's overclocked.

Check that all fans are running and heatsink looks to be
installed correctly - unless there's some issue of voiding
warranty if you open it, I'm unfamiliar with their policies.

I'll call the Mesh support before I open the case just in case of
warranty, I wouldn't want to break the warranty.

Thanks for your help.

 

[kony]

I ran a test again with the video software but this time I didn't stop
the video processing instead I just let it run whilst monitoring the
temperatures in SpeedFan. I knew and expected the computer to crash
when the temperature got too high. What I wanted to check was at what
temperature it would crash. The temperatures reported in SpeedFan rose
to 60 degrees celcius then the computer crashed. When trying to
restart the machine I checked the bios and the temperature in there was
reported at 50 degrees celcius. So either the CPU cooled by 10 degrees
whilst the machine was off for a minute (and I was trying to restart
it) or SpeedFan and the BIOS are a little bit out of sync when
reporting the temperature. Incidentally the BIOS reported the mother
board temp at around 37 degrees C.

In the minute between power off and the restart, the temp
could easily drop by 10C or more. I suggest instead that
you power on the system and leave it sitting on the bios
health monitor page, letting the temp normalize up near it's
peak. See if that results in the restart, or abandon the
test and turn system off it climbs beyond 70C.

When it looks like the bios has reached a peak temp
(shouldn't take but a few minutes), then reset and boot
windows. At that point windows' ACPI power management will
begin idling the CPU to a lower temp, so the sooner you
check that temp in speedfan after the boot has finished, the
closer it's expected to be to the bios temp... maybe barely
higher from some processing of files during OS load but only
a little, closer to the bios temp then than if you had left
the system sitting idle at windows desktop for a few
minutes.

When I ran a second test the computer rebooted at about 60 degrees
celcius.

I'd consider 60C borderline, about the point at which some
system change is called for to reduce temps, especially if
the 60C wasn't at a full load like running Prime95's Torture
Test, Large In-Place FFTs setting.

60C is not a critical temp threshold though, many systems
run fine at 60C. I would suspect you might have a bios
setting (or windows software?) set to shut it down at 60C,
which just happened to be the lowest value possible to be
set thus it was the default, and should be changed up to
70C.

However, if the above described Prime95 test results in
temps closer to 70C, IMO you should insist that the OEM fix
the problem (if they will, this kind of issue is about a
quality design / parts spec problem but they have a little
grey area to argue over it, IF the system hadn't been
turning off which is actually in your favor when it comes to
claiming they have an obligation to fix it.

I'd consider it fixed if it never goes above 60C at full
load, but this is with a room ambient temp at 80C or below,
if you happened to have it sitting next to a heater/vent,
that'll have the potential to make the CPU temp rise too.

Good point, but unfortunately I went through every option in my BIOS
but I couldn't find anything that would signify a shutdown temperature.
I did see a target temperature for CPU (set to 50) in the BIOS and I
think this is used to control fan speeds.

Maybe, but I seldom see specific temps used to set fan
speeds, normally it is for the emergency shutdown. Usually
fan speed is set more like a percentage value.

I did see something call AMD
cool n'quiet but not sure what that is, but even the option for that
was either disable or enable. Is there definately such an option? If
so I'll have to keep looking.

Cool 'n Quiet shouldn't be shutting your system off like
this, but maybe the OEM hid the bios setting for emergency
shutdown? You need to get in touch with them and mention
these details about the shutdown and the 60C temp.

Could even be that the bios is reading CPU temp wrong and if
it read it right it wouldn't be shutting down, in which case
a bios update would be the solution... but I wouldn't go
flashing the bios till Mesh had advised you to, unless you
again want to risk the warranty.

 

[Dundonald]

kony wrote:

<snip>

I just installed the most recent version of speedfan as I noticed there
was a note on its release history explaining that AMD K8 chips are now
supported with the thermal diode. Since installing there is now an
additional temperature in speedfan named "core". I'm guessing this is
the thermal diode temp. Interestingly the settled temp of this is
between 55 and 60 degrees whereas the other two CPU temps are as
mentioned above usually around 38 - 40 degrees ish when settled. I''m
guessing then that the BIOS temperature reading is using the thermal
diode?

Should it be that high?

Is the only answer to replace the heat sink and fan? Or am I avoiding
the root cause of the problem with this? I've taken the existing heat
sinc and fan off and there is grey solution on top of the CPU (I found
out about checking this after googling). Any other ideas?

 

[kony]

kony wrote:

<snip>

I just installed the most recent version of speedfan as I noticed there
was a note on its release history explaining that AMD K8 chips are now
supported with the thermal diode. Since installing there is now an
additional temperature in speedfan named "core". I'm guessing this is
the thermal diode temp. Interestingly the settled temp of this is
between 55 and 60 degrees whereas the other two CPU temps are as
mentioned above usually around 38 - 40 degrees ish when settled. I''m
guessing then that the BIOS temperature reading is using the thermal
diode?

We can't know because you didn't tell us how any of these
temps compare to the reading from my prior suggestion, to
leave system sitting at the bios health monitor page and
compare that peak temp to the temp reported immediately
after system had finished booting windows.

Ideally yes, the bios will use the CPU thermal diode but
whether it did is an unknown to us, you have to determine
this yourself unless you can find confirmation from a fellow
owner of that board.

Should it be that high?

There's no particular setting it "should" be, it merely has
to be cool enough at full load to stay stable. "Usually"
that's around 70C or less, though if the temp exceeds 60C at
any time it's worth determining why it's so high, whether it
merely resulted from a mediocre heatsink or is a sign of a
chassis airflow problem that could be overheating other
parts besides the CPU. Remember that all parts are supposed
to have cooling accomdations relative to a certain amount of
airflow and if CPU heats up while wearing a decent enough
'sink, so would the other parts.

Is the only answer to replace the heat sink and fan?

We don't know, only you can see the system or know what the
'sink is like, if it's mounted good, etc. You might pull
the side panel off the case and point a fan at it to
determine if more chassis airflow would help, and if it does
then weigh the opportunities for doing so.

Or am I avoiding
the root cause of the problem with this? I've taken the existing heat
sinc and fan off and there is grey solution on top of the CPU (I found
out about checking this after googling). Any other ideas?

Once you take the 'sink off, you need to clean off the
original thermal interface material and use a fresh coat of
heatsink grease, preferribly something synthetic (rather
than silicone oil) based like Arctic Silver, Alumina or
Ceramique (or whatever the brand, that it be synthetic oil
based as this formulation performs better in long term use
at higher heat levels).