[Gllug] Body lotion

[Nix]

On 21 Jul 2010, Richard Jones uttered the following:

> On Wed, Jul 21, 2010 at 12:50:11AM +0100, Nix wrote:
>> <85C? How much < is the question, given that 85C is terribly close
>> to the emergency thermal cutout temperature of many CPUs...
>
> This box has serious thermal issues, particularly now the weather is
> warm.  It was shutting down (hard) under any sort of load.  I only
> just started to look into why exactly this is happening and it seems
> to be because the thermal paste was either badly applied originally or
> has just dried out.

It disturbs me more than slightly to learn that the stuff *can* dry out.
How long does it take? Do I need to worry about losing my machines to
a hardware failure I can't fix in five years or something?

(OK, OK, I can take them down the road and get new paste slathered on.
But if that shop wasn't there I'd be buggered.)

> Without the body lotion / paste, the CPUs were running at 90C
> normally, up to 101C (the automatic shut off point) once there was any
> significant load.

Holy searing ow batman.

(For comparison, my Nehalems here run at about 47C under no load, rising
to 60C under load, but without the temperature-controlled fans would
doubtless be much hotter: I don't even know how *many* fans are in one
of them, a sodding noisy one I'd never heard before near the RAID
controller kicked in two weeks ago for half an hour.

The Geode LX in the Soekris Net5501 upstairs, under 10% steady load from
ekeyd, is ticking along at 58C; but of course it has no fan of any kind
and relies purely on passive convection from the outside of its metal
case for cooling. The external temperature, according to the ekeyd, is
somewhere under 33C, but the ekeyd has an ARM in it and when powered on
gets notably warm, so probably it's quite a bit under. The nearby METAR
station says 15C, but that's outside...

... hey, I pay attention to temperatures in weather like this.)

> With the body lotion *and* an extra case fan that I've moved from its
> normal location so it's blowing over both CPUs, things are much
> better:
>
> temp1:       +59.0
> temp2:       +69.0

That's more like it. CPU 2 is still a bit hot for my liking.

> (The case fan is blowing more over CPU0 ("temp1") which is why that
> one is a bit cooler).

It sounds like your real problem was lack of airflow, then...
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug