CPU overheat crash during kernel update.

John Bowling <johnlb2002.DeleteThis@cox.net> writes:

>My personnal work horse computer (AMD dual core) crashed yesterday due to
>overtemp CPU.

It is possible, but from wht you say unlikely. The cpu temp should be
over 100C for it to crash.
Remember however that the "bios" reading is well after the crash ( ie a
few minutes, ) which is time for the cpu to cool down. (Note that a
computer I had began to act very flakey and crash, but the temp was
180C.)


>The temp reading under Kde 3.5 MyComputer was/is 40C, and has stayed there.
>Right after the crash the BIOS told me it was 53C.
>I slowed the processor down (it was a default speed), reinstalled to fix the
>disk problem. The disk did not crash - the system was in the middle of a
>kernel update when the CPU quit. And the system repair restored a kernel
>(the origional? 11.1 distro kernel) but grub couldn't find it. So I
>re-installed.

If you mean a kernel compile, then yes it does stress the cpu.


>After that the CPU temp, per the bios is still under 35C, yet MyComputer
>still says 40C.

>So the temperature readings given by MyComputer are not to be trusted.

>Also, my CPU speed was stated as either 1G or 2.6G prior. It is still 2.6G
>as max, even though I switched the bios to 200Mhz base from 220MHz and
>multiplier from Auto to 6x. That calculates out to 1.2G!) I don't have
>documentation on what x Auto really is (highest manual setting is 11x), but
>it seems slower.

>I'm still looking for any program to read CPU & MB temperatures (and provide
>warnings), and the only one I've found requires a full install of Gnome
>desktop. It might have prevented the problem and hours of down time.

gkrellm.
lmsensors

John Bowling wrote:

> My personnal work horse computer (AMD dual core) crashed yesterday due to
> overtemp CPU.
> The temp reading under Kde 3.5 MyComputer was/is 40C, and has stayed
> there. Right after the crash the BIOS told me it was 53C.
> I slowed the processor down (it was a default speed), reinstalled to fix
> the disk problem. The disk did not crash - the system was in the middle of
> a kernel update when the CPU quit. And the system repair restored a kernel
> (the origional? 11.1 distro kernel) but grub couldn't find it. So I
> re-installed.
>
> After that the CPU temp, per the bios is still under 35C, yet MyComputer
> still says 40C.
>
> So the temperature readings given by MyComputer are not to be trusted.
>
> Also, my CPU speed was stated as either 1G or 2.6G prior. It is still 2.6G
> as max, even though I switched the bios to 200Mhz base from 220MHz and
> multiplier from Auto to 6x. That calculates out to 1.2G!) I don't have
> documentation on what x Auto really is (highest manual setting is 11x),
> but it seems slower.
>
> I'm still looking for any program to read CPU & MB temperatures (and
> provide warnings), and the only one I've found requires a full install of
> Gnome desktop. It might have prevented the problem and hours of down time.

It would help is you said what motherboard and cpu type you were using.
from what you wrote something is not right in my eyes.

John Bowling wrote:
> My personnal work horse computer (AMD dual core) crashed yesterday due to
> overtemp CPU.
> The temp reading under Kde 3.5 MyComputer was/is 40C, and has stayed there.
> Right after the crash the BIOS told me it was 53C.
> I slowed the processor down (it was a default speed), reinstalled to fix the
> disk problem. The disk did not crash - the system was in the middle of a
> kernel update when the CPU quit. And the system repair restored a kernel
> (the origional? 11.1 distro kernel) but grub couldn't find it. So I
> re-installed.
>
> After that the CPU temp, per the bios is still under 35C, yet MyComputer
> still says 40C.
>
> So the temperature readings given by MyComputer are not to be trusted.
>
> Also, my CPU speed was stated as either 1G or 2.6G prior. It is still 2.6G
> as max, even though I switched the bios to 200Mhz base from 220MHz and
> multiplier from Auto to 6x. That calculates out to 1.2G!) I don't have
> documentation on what x Auto really is (highest manual setting is 11x), but
> it seems slower.
>
> I'm still looking for any program to read CPU & MB temperatures (and provide
> warnings), and the only one I've found requires a full install of Gnome
> desktop. It might have prevented the problem and hours of down time.

I have to clean the carpet dust, dead bugs and other muck out of the
power supply, cpu heat sink and fans every 3 month or so to prevent
overheating. Keep the fan's leading edges clean or loose significant
airflow.

Hans <no-spam.RemoveThis@no-spam.xx> writes:

>John Bowling wrote:
>> My personnal work horse computer (AMD dual core) crashed yesterday due to
>> overtemp CPU.
>> The temp reading under Kde 3.5 MyComputer was/is 40C, and has stayed there.
>> Right after the crash the BIOS told me it was 53C.

If your computer crashed with a cpu temp of 53 then your cpu is
defective. It should be able to survive up to 100C.
Get a new computer.

>> I slowed the processor down (it was a default speed), reinstalled to fix the
>> disk problem. The disk did not crash - the system was in the middle of a
>> kernel update when the CPU quit. And the system repair restored a kernel
>> (the origional? 11.1 distro kernel) but grub couldn't find it. So I
>> re-installed.
>>
>> After that the CPU temp, per the bios is still under 35C, yet MyComputer
>> still says 40C.

How did you determine the bios temp? shut down the computer, reboot and
read the bios? That is lots of time for the cpu to cool down.

>>
>> So the temperature readings given by MyComputer are not to be trusted.

What is MyComputer?

>>
>> Also, my CPU speed was stated as either 1G or 2.6G prior. It is still 2.6G
>> as max, even though I switched the bios to 200Mhz base from 220MHz and
>> multiplier from Auto to 6x. That calculates out to 1.2G!) I don't have
>> documentation on what x Auto really is (highest manual setting is 11x), but
>> it seems slower.
>>
>> I'm still looking for any program to read CPU & MB temperatures (and provide
>> warnings), and the only one I've found requires a full install of Gnome
>> desktop. It might have prevented the problem and hours of down time.

lmsensors and gkrellm
No you do not need the full gnome desktop for gkrellm


>I have to clean the carpet dust, dead bugs and other muck out of the
>power supply, cpu heat sink and fans every 3 month or so to prevent
>overheating. Keep the fan's leading edges clean or loose significant
>airflow.

On Thursday 20 Aug 2009 16:37, Unruh played with alphabet spaghetti and left
this on the plate:

> Hans <no-spam.DeleteThis@no-spam.xx> writes:
>
>>John Bowling wrote:

>>> After that the CPU temp, per the bios is still under 35C, yet MyComputer
>>> still says 40C.
>
> How did you determine the bios temp? shut down the computer, reboot and
> read the bios?

Takes about 15 seconds on my older laptop running 11.2m5. Enough time for
the processor to cool a little, but not enough for any significant cooling
to take place.

> That is lots of time for the cpu to cool down.

It's enough time for there to be a bit of a drop, but it's not going to drop
by more than a few degrees, and you can estimate the drop by watching how
much it drops while just watching it on the BIOS screen.

>>> So the temperature readings given by MyComputer are not to be trusted.
>
> What is MyComputer?

Under KDE, there's an icon that gives you an overview of the computer. This
includes using ACPI to read the temperature of the processor. For some
reason, no matter what the load, or what sensors says the processor
temperature is, it always seems to read 40C.

> lmsensors and gkrellm
> No you do not need the full gnome desktop for gkrellm
>
>
>>I have to clean the carpet dust, dead bugs and other muck out of the
>>power supply, cpu heat sink and fans every 3 month or so to prevent
>>overheating. Keep the fan's leading edges clean or loose significant
>>airflow.

Pet hairs, especially those from long-haired cats, are another killer of
PSUs and fans.

Regards,
David Bolt

--
Team Acorn: www.distributed.net OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s
openSUSE 10.3 32b | openSUSE 11.0 32b | |
openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b |
RISC OS 3.6 | RISC OS 3.11 | openSUSE 11.1 PPC | TOS 4.02

David Bolt wrote:

> Under KDE, there's an icon that gives you an overview of the computer.
> This includes using ACPI to read the temperature of the processor. For
> some reason, no matter what the load, or what sensors says the processor
> temperature is, it always seems to read 40C.

With KDE 3.5, I use the Ksensors front end to sensors. The initial settings
give obviously incorrect readings - the CPU is below room temp! I use an
IR thermometer to get the highest temp in the case and around the CPU then
apply a multiplier to get the displayed numbers to a more realistic value.
On my machine, the temps obviously track CPU and disk load so it's sort of
like my truck - when the needle rises above it's normal resting place, I
take note. If it shoots toward the high side in a hurry, I got problems!

>> lmsensors and gkrellm
>> No you do not need the full gnome desktop for gkrellm
>>
>>
>>>I have to clean the carpet dust, dead bugs and other muck out of the
>>>power supply, cpu heat sink and fans every 3 month or so to prevent
>>>overheating. Keep the fan's leading edges clean or loose significant
>>>airflow.
>
> Pet hairs, especially those from long-haired cats, are another killer of
> PSUs and fans.

Suggestion: clean/vacuum as much as you can get then reverse the machine and
blow back into the power supply through the fan. Vacuum the inside ps
vents thoroughly after that. Reason is that the ps fan pulls a pretty good
flow through it that piles up the junk between wires and parts which you
pretty much have to hit with a direct reverse flow to dislodge. I'm a
smoker with 3 dogs and a wood shop in the next room so this is a must do
for me about once a month.

--
Will Honea

Hans wrote:

> John Bowling wrote:
>> My personnal work horse computer (AMD dual core) crashed yesterday due to
>> overtemp CPU.
>> The temp reading under Kde 3.5 MyComputer was/is 40C, and has stayed
>> there. Right after the crash the BIOS told me it was 53C.
>> I slowed the processor down (it was a default speed), reinstalled to fix
>> the disk problem. The disk did not crash - the system was in the middle
>> of a kernel update when the CPU quit. And the system repair restored a
>> kernel (the origional? 11.1 distro kernel) but grub couldn't find it. So
>> I re-installed.
>>
>> After that the CPU temp, per the bios is still under 35C, yet MyComputer
>> still says 40C.
>>
>> So the temperature readings given by MyComputer are not to be trusted.
>>
>> Also, my CPU speed was stated as either 1G or 2.6G prior. It is still
>> 2.6G as max, even though I switched the bios to 200Mhz base from 220MHz
>> and multiplier from Auto to 6x. That calculates out to 1.2G!) I don't
>> have documentation on what x Auto really is (highest manual setting is
>> 11x), but it seems slower.
>>
>> I'm still looking for any program to read CPU & MB temperatures (and
>> provide warnings), and the only one I've found requires a full install of
>> Gnome desktop. It might have prevented the problem and hours of down
>> time.
>
> I have to clean the carpet dust, dead bugs and other muck out of the
> power supply, cpu heat sink and fans every 3 month or so to prevent
> overheating. Keep the fan's leading edges clean or loose significant
> airflow.

as root open a konsole and run this command "sensors-detect" then follow
what ever is says to do then install gkrellm this will give you a good temp
reading of the cpu and other stuff. then you will findout what your problem
is.

On Friday 21 Aug 2009 04:43, Will Honea played with alphabet spaghetti and
left this on the plate:

> David Bolt wrote:
>
>> Under KDE, there's an icon that gives you an overview of the computer.
>> This includes using ACPI to read the temperature of the processor. For
>> some reason, no matter what the load, or what sensors says the processor
>> temperature is, it always seems to read 40C.
>
> With KDE 3.5, I use the Ksensors front end to sensors. The initial
> settings
> give obviously incorrect readings - the CPU is below room temp!

I have one system like that. It's usually under virtually zero load, not
doing much but running a web server. At the moment it's also rebuilding a
mirror of the Factory repo. The load is still very low and it's supposedly
running just a little lower than the room present temperature (21C):

davjam@cobra-mk3:~> sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp:
+1°C
Core0 Temp:
+3°C
Core1 Temp:
+9°C
Core1 Temp:
+4°C

The one I'm using now displays a much more reasonable approximation:

davjam@moray:/local/music/music> sensors
acpitz-virtual-0
Adapter: Virtual device
temp1: +40.0°C (crit = +95.0°C)

k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp: +33.0°C
Core0 Temp: +38.0°C
Core1 Temp: +38.0°C
Core1 Temp: +48.0°C

> I use an
> IR thermometer to get the highest temp in the case and around the CPU then
> apply a multiplier to get the displayed numbers to a more realistic value.

Did you check to see if using a multiplier left it accurate at both the top
and bottom of the range? My guess is that it's actually tracking the
temperatures just fine, but it needs to have the offset tweaked to bring it
up to the correct values.

>> Pet hairs, especially those from long-haired cats, are another killer of
>> PSUs and fans.
>
> Suggestion: clean/vacuum as much as you can get then reverse the machine
> and
> blow back into the power supply through the fan. Vacuum the inside ps
> vents thoroughly after that.

I do that, roughly every 6 months or so, although I usually strip the covers
off the PSUs and vacuum the insides to get rid of all the dust and other
stuff. I also tend to strip the fans, clean off the stuff that's collected
on the fan blades, and re-lubricate the fan spindles.

> Reason is that the ps fan pulls a pretty good
> flow through it that piles up the junk between wires and parts which you
> pretty much have to hit with a direct reverse flow to dislodge.

A can of spray air comes in handy for that sort of issue, as do insulated
needle-nosed pliers.

> I'm a
> smoker with 3 dogs and a wood shop in the next room so this is a must do
> for me about once a month.

I just have one fairly short-haired dog and one long-haired cat. While the
dogs hairs are more noticeable, the cat's hairs are the ones that cause the
problems.


Regards,
David Bolt

--
Team Acorn: www.distributed.net OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s
openSUSE 10.3 32b | openSUSE 11.0 32b | |
openSUSE 10.3 64b | openSUSE 11.0 64b | openSUSE 11.1 64b | openSUSE 11.2m5
RISC OS 3.6 | RISC OS 3.11 | openSUSE 11.1 PPC | TOS 4.02