My temp for CPU is 44 for the northbridge - is this normal?

Ok, so it sounds like I don't have anything to worry about. Now, if
it does over-heat - what is the worse thing that will happen? Will it
fry my HD's? The motherboard? The processor? Does it have auto-shut
down's if it gets too hot?

>
> Your temperatures and variations you are seeing are 100% normal and are
> actually on the lower operating range for a 3.0Ghz P4D setup. Just to give
> you a comparison from a newer setup, my E8400 running at stock voltage but
> at 3.6Ghz runs at 47C idle and at 57 to59C under 100% load on both cores
> with Prime95 in an Antec 182 case with added cooling. (Room temp:22C)
> Running at 100% stock (3.0Ghz ) with a decent Coolermaster hs/f it would
> idle at 37C and go to 45-47 full load.

<daviddschool@gmail.com> wrote in message
news:e215137a-b4ca-4787-94ef-eec708bae7bc@d45g2000hsc.googlegroups.com...
> Ok, so it sounds like I don't have anything to worry about. Now, if
> it does over-heat - what is the worse thing that will happen? Will it
> fry my HD's? The motherboard? The processor? Does it have auto-shut
> down's if it gets too hot?
>
>>
>> Your temperatures and variations you are seeing are 100% normal and are
>> actually on the lower operating range for a 3.0Ghz P4D setup. Just to
>> give
>> you a comparison from a newer setup, my E8400 running at stock voltage
>> but
>> at 3.6Ghz runs at 47C idle and at 57 to59C under 100% load on both cores
>> with Prime95 in an Antec 182 case with added cooling. (Room temp:22C)
>> Running at 100% stock (3.0Ghz ) with a decent Coolermaster hs/f it would
>> idle at 37C and go to 45-47 full load.
>
Well, like Ed or Phil said, the CPU will simply lock up without causing any
physical damage or issues. But believe me, you are nowhere near this level.
My brother had (bought new at the time) a Dell XPS box with a P4D Extreme
Edition 3.8Ghz unit in it and the temps off that thing were extreme to say
the least...almost double yours...yet it was 100% stable in gaming and
everything else.

Augustus wrote:
> <daviddschool@gmail.com> wrote in message
> news:e215137a-b4ca-4787-94ef-eec708bae7bc@d45g2000hsc.googlegroups.com...
>> Ok, so it sounds like I don't have anything to worry about. Now, if
>> it does over-heat - what is the worse thing that will happen? Will it
>> fry my HD's? The motherboard? The processor? Does it have auto-shut
>> down's if it gets too hot?
>>
>>> Your temperatures and variations you are seeing are 100% normal and are
>>> actually on the lower operating range for a 3.0Ghz P4D setup. Just to
>>> give
>>> you a comparison from a newer setup, my E8400 running at stock voltage
>>> but
>>> at 3.6Ghz runs at 47C idle and at 57 to59C under 100% load on both cores
>>> with Prime95 in an Antec 182 case with added cooling. (Room temp:22C)
>>> Running at 100% stock (3.0Ghz ) with a decent Coolermaster hs/f it would
>>> idle at 37C and go to 45-47 full load.
> Well, like Ed or Phil said, the CPU will simply lock up without causing any
> physical damage or issues. But believe me, you are nowhere near this level.
> My brother had (bought new at the time) a Dell XPS box with a P4D Extreme
> Edition 3.8Ghz unit in it and the temps off that thing were extreme to say
> the least...almost double yours...yet it was 100% stable in gaming and
> everything else.
>

Both the Athlon64 or later, and contemporary LGA775 socketed processors from Intel,
have THERMTRIP. Both companies make processors which have the ability to
detect overheat. THERMTRIP is then wired into the logic that controls the
ATX power supply. So if the heatsink clip snaps, and the heatsink falls off, in a
second or two, the computer will shut off.

The second protection mechanism is throttling. It might start to be applied
at a point about 20C below THERMTRIP. Throttling is a function applied to the
effective processor clock - the processor is put to sleep for short intervals,
as a thermal control measure. What this does, is rob the user of the performance
they paid for. What you see when this is working, is that the processor seems
to have a "ceiling" for temperature, and won't rise above a certain point.
But your performance might not be as good as normal, if you reach that point.

This is why people go to the trouble of monitoring CPU temperature, in order
to avoid throttling.

An example of a program that loads the processor nicely, is Prime95. This version
runs in Windows, and starts a thread of execution per core. There are other downloads
for Linux, and I've also run multiple copies of an older version in a Linux
environment.

http://www.mersenne.org/gimps/p95v255a.zip

Speedfan, a monitoring program from almico.com , will monitor both motherboard
related temperatures, and will also display hard drive temperatures as provided
by the SMART standard for hard drives. You can use Speedfan, and watch your
temperatures, while Prime95 is running. You don't have to use the fan control
functions, if you are just interested in monitoring.

To make up some numbers, current processors might be somewhere around 70C
for throttle, and 90C for THERMTRIP. You might want to make the CPU cooling
good enough, so that when an application like Prime95 or Orthos is running,
the CPU temperature stays below 65C.

There is a program called RMClock, and it has a graphic display of some
system properties. One of the properties, is monitoring the throttle
bit in a hardware register. Using a program like that, it is actually
possible to see exactly when the processor starts to throttle.

The last processor with thermal issues was the Athlon (socket 462 or socket A).
With the early motherboards for those, if the CPU heatsink fell off, the
processor could be cooked alive in the matter of a few seconds. At
the point that I bought a socket 462 board, when AthlonXP processors
were being used, my motherboard had a small eight pin chip that did
nothing but watch the CPU internal die temperature. The chip had
a threshold, and worked in the same fashion as THERMTRIP. Since it
was a chip, and did not rely on BIOS or operating system code,
it could autonomously shut off the computer. So even for
computers of that vintage, it was possible to get protected
ones. It was some of the early Socket 462 motherboards and processors,
that might have been vulnerable to burnup if the heatsink fell off.

The key message in the above description, is to do the testing with
a 100% CPU load. Then you know how the system works under worst case
conditions. (The only thing you're not simulating, is an elevated
room temperature, but you can add an offset to the results to
compensate for that. If, for example, room temp is 25C today,
and you expect to run the computer while room temp is 35C, then
you know your temps will shift by about 10C.)

My CPU peaks at around 44C right now, with a room temp of 25C.
So I'm quite far from any temperature which would cause an issue.
(Old P4 processor running slightly above 3GHz, Northwood type.)

With regard to sensors and labeling of displays, sometimes the
CPU temperature and the Northbridge or motherboard sensors get
swapped. Watching the dynamics of each temperature display
(which temperature rises suddenly, when you start running
Prime95), will help tell a person which sensor is which.

Another minor issue, is motherboard heatpipe coolers, and
computer cases like some of the ones from Lian-Li. At least
some of the earlier motherboards, that use heatpipes to join
the heat from Northbridge, Southbridge, and MOSFET coolers,
don't work as well, if the motherboard is rotated 180 degrees.
Some Lian-Li cases mount the motherboard upside-down, and
that has been known to cause abnormally high Northbridge
temperature readings. The solution is to try to find a
regular computer case, that mounts the motherboard the way
the motherboard manufacturer expected it would be used.

HTH,
Paul

'Paul' wrote, in part:
> Both the Athlon64 or later, and contemporary LGA775 socketed processors
> from Intel,
> have THERMTRIP. Both companies make processors which have the ability to
> detect overheat. THERMTRIP is then wired into the logic that controls the
> ATX power supply. So if the heatsink clip snaps, and the heatsink falls
> off, in a
> second or two, the computer will shut off.
..
..
_____

Intel x86 CPUs have had an on-CPU-die thermal diode at least since the
Pentium 75 MHz (the Pentium 60 MHz and 66 MHz likely did, but I don't have
any documentation on the first two Pentium CPUs).

The real problem for the AMD CPUs (before AMD got religion and added an
on-CPU-die thermal diode) when a heatsink failed was the IDLE power
dissipation, not the high stress power dissipation. Because, just like the
Intel CPUs, when the CPU is actually doing a lot of work and overheats the
CPU locks up and cools off to whatever the idle temperature would be. Idle
power dissipation as a percentage of total power dissipation is greatly
reduced as the CPU core voltage is reduced.

It is easy to understand the spectacular burn-ups of AMD processors when you
consider the power dissipation per unit area of an x86 CPU is close to that
per unit area of a burner on an electric stovetop eye on high, glowing at
red heat. For example:

8 inch diameter eye dissipating 1250 watts, spiral element surface area
approximately equal to the area of the 8 inch disk (the top surface plus the
bottom surface) 4 inches X 4 inches X 3.14 = ~ 50 square inches, 1250
watts / 50 sq. in. = 25 watts per sq. in. : Pentium 4 Prescott 2.4 GHz,
die size about 10 mm X 11 mm, assume the idle power dissipation to be 5
watts, 10 mm X 11 mm = 110 sq. mm, 110 sq. mm / (25.4 X 25.4) = ~ 0.17 sq.
in., 5 watts / 0.17 sq. in = 30 watts per sq. in.

Many temperature monitoring programs are available, but none are nearly as
universal as the old standby MotherBoard Monitor once was. Unfortunately
MotherBoard Monitor hasn't been maintained for the last four or so years, so
it is useless for almost all motherboards designed since then. I haven't
been able to find any temperature monitoring applet that works very well
with my year old EVGA 680i motherboard, even those applets that correctly
identify the Winbond system monitoring chips used on the motherboard.

Orientation of the motherboard is only a factor when system case ventilation
is inadequate. I know, I know, heat rises, but the pressure differential
over 5 to 20 inches is very low, so passive cooling is negligible compared
to even the feeble 0.05 psi air flow the CPU and system fans generate.

Phil Weldon


"Paul" <nospam@needed.com> wrote in message
news:g0lva8$h8j$1@registered.motzarella.org...
> Augustus wrote:
>> <daviddschool@gmail.com> wrote in message
>> news:e215137a-b4ca-4787-94ef-eec708bae7bc@d45g2000hsc.googlegroups.com...
>>> Ok, so it sounds like I don't have anything to worry about. Now, if
>>> it does over-heat - what is the worse thing that will happen? Will it
>>> fry my HD's? The motherboard? The processor? Does it have auto-shut
>>> down's if it gets too hot?
>>>
>>>> Your temperatures and variations you are seeing are 100% normal and are
>>>> actually on the lower operating range for a 3.0Ghz P4D setup. Just to
>>>> give
>>>> you a comparison from a newer setup, my E8400 running at stock voltage
>>>> but
>>>> at 3.6Ghz runs at 47C idle and at 57 to59C under 100% load on both
>>>> cores
>>>> with Prime95 in an Antec 182 case with added cooling. (Room temp:22C)
>>>> Running at 100% stock (3.0Ghz ) with a decent Coolermaster hs/f it
>>>> would
>>>> idle at 37C and go to 45-47 full load.
>> Well, like Ed or Phil said, the CPU will simply lock up without causing
>> any physical damage or issues. But believe me, you are nowhere near this
>> level. My brother had (bought new at the time) a Dell XPS box with a P4D
>> Extreme Edition 3.8Ghz unit in it and the temps off that thing were
>> extreme to say the least...almost double yours...yet it was 100% stable
>> in gaming and everything else.
>
> Both the Athlon64 or later, and contemporary LGA775 socketed processors
> from Intel,
> have THERMTRIP. Both companies make processors which have the ability to
> detect overheat. THERMTRIP is then wired into the logic that controls the
> ATX power supply. So if the heatsink clip snaps, and the heatsink falls
> off, in a
> second or two, the computer will shut off.
>
> The second protection mechanism is throttling. It might start to be
> applied
> at a point about 20C below THERMTRIP. Throttling is a function applied to
> the
> effective processor clock - the processor is put to sleep for short
> intervals,
> as a thermal control measure. What this does, is rob the user of the
> performance
> they paid for. What you see when this is working, is that the processor
> seems
> to have a "ceiling" for temperature, and won't rise above a certain point.
> But your performance might not be as good as normal, if you reach that
> point.
>
> This is why people go to the trouble of monitoring CPU temperature, in
> order
> to avoid throttling.
>
> An example of a program that loads the processor nicely, is Prime95. This
> version
> runs in Windows, and starts a thread of execution per core. There are
> other downloads
> for Linux, and I've also run multiple copies of an older version in a
> Linux
> environment.
>
> http://www.mersenne.org/gimps/p95v255a.zip
>
> Speedfan, a monitoring program from almico.com , will monitor both
> motherboard
> related temperatures, and will also display hard drive temperatures as
> provided
> by the SMART standard for hard drives. You can use Speedfan, and watch
> your
> temperatures, while Prime95 is running. You don't have to use the fan
> control
> functions, if you are just interested in monitoring.
>
> To make up some numbers, current processors might be somewhere around 70C
> for throttle, and 90C for THERMTRIP. You might want to make the CPU
> cooling
> good enough, so that when an application like Prime95 or Orthos is
> running,
> the CPU temperature stays below 65C.
>
> There is a program called RMClock, and it has a graphic display of some
> system properties. One of the properties, is monitoring the throttle
> bit in a hardware register. Using a program like that, it is actually
> possible to see exactly when the processor starts to throttle.
>
> The last processor with thermal issues was the Athlon (socket 462 or
> socket A).
> With the early motherboards for those, if the CPU heatsink fell off, the
> processor could be cooked alive in the matter of a few seconds. At
> the point that I bought a socket 462 board, when AthlonXP processors
> were being used, my motherboard had a small eight pin chip that did
> nothing but watch the CPU internal die temperature. The chip had
> a threshold, and worked in the same fashion as THERMTRIP. Since it
> was a chip, and did not rely on BIOS or operating system code,
> it could autonomously shut off the computer. So even for
> computers of that vintage, it was possible to get protected
> ones. It was some of the early Socket 462 motherboards and processors,
> that might have been vulnerable to burnup if the heatsink fell off.
>
> The key message in the above description, is to do the testing with
> a 100% CPU load. Then you know how the system works under worst case
> conditions. (The only thing you're not simulating, is an elevated
> room temperature, but you can add an offset to the results to
> compensate for that. If, for example, room temp is 25C today,
> and you expect to run the computer while room temp is 35C, then
> you know your temps will shift by about 10C.)
>
> My CPU peaks at around 44C right now, with a room temp of 25C.
> So I'm quite far from any temperature which would cause an issue.
> (Old P4 processor running slightly above 3GHz, Northwood type.)
>
> With regard to sensors and labeling of displays, sometimes the
> CPU temperature and the Northbridge or motherboard sensors get
> swapped. Watching the dynamics of each temperature display
> (which temperature rises suddenly, when you start running
> Prime95), will help tell a person which sensor is which.
>
> Another minor issue, is motherboard heatpipe coolers, and
> computer cases like some of the ones from Lian-Li. At least
> some of the earlier motherboards, that use heatpipes to join
> the heat from Northbridge, Southbridge, and MOSFET coolers,
> don't work as well, if the motherboard is rotated 180 degrees.
> Some Lian-Li cases mount the motherboard upside-down, and
> that has been known to cause abnormally high Northbridge
> temperature readings. The solution is to try to find a
> regular computer case, that mounts the motherboard the way
> the motherboard manufacturer expected it would be used.
>
> HTH,
> Paul

Phil Weldon wrote:

>
> Orientation of the motherboard is only a factor when system case
> ventilation is inadequate. I know, I know, heat rises, but the pressure
> differential over 5 to 20 inches is very low, so passive cooling is
> negligible compared to even the feeble 0.05 psi air flow the CPU and
> system fans generate.
>
> Phil Weldon
>

The issue with the Lian-Li case, was noted with an Asus A8N-SLI family
board. It seems some heatpipe coolers are more sensitive to orientation
than others. The internal finish of the heatpipe has an effect on
how well it works when the orientation changes from the expected one.
Changing the computer case, to a non-Lian-Li was only one possible
solution. This should not be interpreted as a convection cooling
issue, because the temperature rise was more pronounced than that.

When the internal operation of a heatpipe is thwarted, whether by
"thermal swamping" (pushing too much power for the heatpipe design),
a lack of operating fluid (a leak), or a change in orientation,
all of those can have a severe impact on performance. When the
liquid-vapor phase cooling cycle is working, that is 100 to 500 times
more effective, than the conductivity of the copper tubing itself.
So if the internal operation is degraded, it can lead to a significant
change in performance.

Paul