CPU too hot on boot up

I have a quad core (9450) on a P5E3. PC Probe shows a big spike on
the CPU temp after booting up. It's drops pretty quickly though. Stll
it's booting up around 70 degrees centigrade and then drops down to
about 30 just idling. Not sure why it runs so hot.

wrote:
I have a quad core (9450) on a P5E3. PC Probe shows a big spike on
the CPU temp after booting up. It's drops pretty quickly though. Stll
it's booting up around 70 degrees centigrade and then drops down to
about 30 just idling. Not sure why it runs so hot.

The only conclusive condition, is when the processor is loaded.
For example, run Prime95 while in Windows. This will start a thread
per core and run the cores pretty close to 100% each.

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

Now, while Prime95 is running, check the temperature. Did the
temp shoot up ? Is it now over 70C ? If so, you need more cooling,
or you need to check that your Vcore is not set incorrectly,
or your core frequency is too high.

Power is proportional to F*C*V**2 where F is core frequency
(actually it is toggle rate but that is too hard to explain),
V squared is the Vcore value squared, and C is a constant of
proportionality. The equation is normally used to compute power
on a CMOS gate, but can be extended as a relationship, to describe
processors doing constant work. (i.e. As long as you are comparing
the processors doing the same amount of work, such as both cases
running Prime95, then the proportionality should apply.) Since
power is proportional to Vcore squared, the power is strongly
dependent on voltage used.

If you have decent cooling, the temp shouldn't go over 70C. At
that temperature, the processor might start to throttle. To
evaluate throttling, you can use a program like RMClock from
rightmark.

RMClock download page - yes, there is a free version.
http://cpu.rightmark.org/download.shtml

Demo of RMClock on a processor which is throttling because it is too hot.
The downward spikes are the loss in performance, due to clock throttling.

http://www.xbitlabs.com/articles/cpu...c-guide_7.html

It could be something as simple, as one of your push-pins on the
heatsink is not holding properly, and the heatsink isn't sitting
flat on the CPU.

Paul

Paul, thanks for your suggestions - Prime ran less then a minute
before the temp was 92 and climbing.
The hsf is seated fine , it is the stock cooler, but still while I
expect not so great results with it, at stock right now for boot into
OS it's hitting 70. Something is odd.

Vcore is 1.12 - 1.2
This is the default bios 4.0.2 for my motherboard. So still digging.

Stuart

wrote:
Paul, thanks for your suggestions - Prime ran less then a minute
before the temp was 92 and climbing.
The hsf is seated fine , it is the stock cooler, but still while I
expect not so great results with it, at stock right now for boot into
OS it's hitting 70. Something is odd.

Vcore is 1.12 - 1.2
This is the default bios 4.0.2 for my motherboard. So still digging.

Stuart


Has the heatsink/fan been removed and reinstalled a number of
times ? Any chance the thermal paste or pad is messed up ?

If the processor was at 92C, then if the heatsink was making
good contact, the fins of the heatsink should be going up
in temperature as well. You'd also expect the exhaust air
from the CPU fan to be a little on the warm side. If you
use your finger as a sensor, do the observable conditions
seem consistent with the temperature readout software ?
Or is the heatsink cold to the touch ?

Paul

On Jul 26, 5:26*pm, Paul wrote:
wrote:
Paul, thanks for your suggestions - *Prime ran less then a minute
before the temp was 92 and climbing.
The hsf is seated fine , it is the stock cooler, but still while I
expect not so great results with it, at stock right now for boot into
OS it's hitting 70. *Something is odd.

Vcore is 1.12 - 1.2
This is the default bios 4.0.2 for my motherboard. * So still digging..

Stuart

Has the heatsink/fan been removed and reinstalled a number of
times ? Any chance the thermal paste or pad is messed up ?

If the processor was at 92C, then if the heatsink was making
good contact, the fins of the heatsink should be going up
in temperature as well. You'd also expect the exhaust air
from the CPU fan to be a little on the warm side. If you
use your finger as a sensor, do the observable conditions
seem consistent with the temperature readout software ?
Or is the heatsink cold to the touch ?

* * Paul

Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch.
So, no the conditions are not consistent with the readout.

Also the HSF has not been removed. I put down a nice coat of arctic
before laying the heatsink down.

Stuart

p.s. Aside from that this system is driving me nuts . The combo of
this board and the ATI 4870 is pretty unstable at this point. That's
a whole other issue.

wrote:
Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch.
So, no the conditions are not consistent with the readout.

Also the HSF has not been removed. I put down a nice coat of arctic
before laying the heatsink down.

By "a nice coat" how thick do you mean. It is recommended, IIRC, that
the coat should be 0.003" thick! Too thick a coat can cause
overheating, too!

On Jul 26, 7:17*pm, tobe wrote:
wrote:
Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch.
So, no the conditions are not consistent with the readout.

Also the HSF has not been removed. *I put down a nice coat of arctic
before laying the heatsink down.

By "a nice coat" how thick do you mean. *It is recommended, IIRC, that
the coat should be 0.003" thick! *Too thick a coat can cause
overheating, too!

Well I didn't have my micrometer but I believe it was about close
enough. Not thick.
I am going to reset it but I think replacing the HSF wouldn't be a bad
idea either. I

wrote:
On Jul 26, 7:17 pm, tobe wrote:
wrote:
Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch.
So, no the conditions are not consistent with the readout.
Also the HSF has not been removed. I put down a nice coat of arctic
before laying the heatsink down.
By "a nice coat" how thick do you mean. It is recommended, IIRC, that
the coat should be 0.003" thick! Too thick a coat can cause
overheating, too!

Well I didn't have my micrometer but I believe it was about close
enough. Not thick.
I am going to reset it but I think replacing the HSF wouldn't be a bad
idea either. I

I think when you remove it, you're going to see a pattern in the AS
that tells you it wasn't making good contact. If you clean off the
AS, put a single rice grain of AS in the center of the CPU, install
the heatsink, remove it and examine the pattern, the AS probably won't
have spread over the entire surface. And that implies a problem with
the fit.

Paul

On Jul 27, 3:20*am, Paul wrote:
wrote:
On Jul 26, 7:17 pm, tobe wrote:
wrote:
Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch.
So, no the conditions are not consistent with the readout.
Also the HSF has not been removed. *I put down a nice coat of arctic
before laying the heatsink down.
By "a nice coat" how thick do you mean. *It is recommended, IIRC, that
the coat should be 0.003" thick! *Too thick a coat can cause
overheating, too!

Well I didn't have my micrometer but I believe it was about close
enough. *Not thick.
I am going to reset it but I think replacing the HSF wouldn't be a bad
idea either. *I

I think when you remove it, you're going to see a pattern in the AS
that tells you it wasn't making good contact. If you clean off the
AS, put a single rice grain of AS in the center of the CPU, install
the heatsink, remove it and examine the pattern, the AS probably won't
have spread over the entire surface. And that implies a problem with
the fit.

* * Paul

I readl through an AS pdf online. It shows a line in the center,
longer then a rice grain but short of the height of the chip. The
heat spreader is apparently wider then the actual cpu so it need not
spread corner to corner. Unless this document is incorrect I'll try
that.

Stuart

On Jul 27, 5:13*am, " wrote:
On Jul 27, 3:20*am, Paul wrote:



wrote:
On Jul 26, 7:17 pm, tobe wrote:
wrote:
Now , that's interesting but let me understand - why if at 92c you say
heatsink is making good contact ?
However that's an excellent question because it's cool to the touch..
So, no the conditions are not consistent with the readout.
Also the HSF has not been removed. *I put down a nice coat of arctic
before laying the heatsink down.
By "a nice coat" how thick do you mean. *It is recommended, IIRC, that
the coat should be 0.003" thick! *Too thick a coat can cause
overheating, too!

Well I didn't have my micrometer but I believe it was about close
enough. *Not thick.
I am going to reset it but I think replacing the HSF wouldn't be a bad
idea either. *I

I think when you remove it, you're going to see a pattern in the AS
that tells you it wasn't making good contact. If you clean off the
AS, put a single rice grain of AS in the center of the CPU, install
the heatsink, remove it and examine the pattern, the AS probably won't
have spread over the entire surface. And that implies a problem with
the fit.

* * Paul

I readl through an AS pdf online. *It shows a line in the center,
longer then a rice grain but short of the height of the chip. *The
heat spreader is apparently wider then the actual cpu so it need not
spread corner to corner. *Unless this document is incorrect I'll try
that.

Stuart

Postedit - removed HSF, cleaned and prepped with AS kit both,
reseated, on load completion in Windows BEFO 67C CURRENT 47.
About to try Prime