Martin wrote:

> Per Jessen wrote:
> [...]
>> I got hold of 25.6 like you suggested, and I'm running 4 threads of
>> the small FFT test now.  It drove the CPU temp right up to 60, then
>> carried on to 65-66C, which none of my other tests have done. It's
>> already been running longer than any other test with 4 instances.
> 
> Mmmm...
> 
> To have a +15 deg C temperature rise or more from off-load to on-load
> suggests that your heatsink is not adequate or that you have a poor
> thermal contact between the heatsink and the CPU...

OK, I don't know how long it took to rise from 50 to 65, but it wasn't
immediate, probably took 10-15seconds.  At first I heard the CPU fan
increase RPMs, then I took a look at the temp. 

For now, with what I consider a fully loaded CPU, it's holding at 67C,
which is fine.  AMD says max 70C.  I don't intend to run 4 copies of
mprime, the real workload will likely be very different, so even
investing in one of the many very fancy cooling devices (with heatpipes
and whathaveyou) is probably overkill. 

> How hot does the heatsink itself get? Does it feel like it is too hot
> to touch? (Carefully use your fingers!)

Here's a funny story: 

I've added a simple paper-funnel on top of the CPU fan.  The air inside
the cabinet gets quite hot from harddisks and the graphics card, but by
funnelling air from the outside of the cabinet, I can reduce CPU
temperature by 2-3 degrees. 
Now, to get to touch the heatsink I removed the funnel.  This almost
immediately made the fan spin up from 3300 to 4500 RPMs.  It also made
the temperature jump to 70C, which made me put the funnel back on right
away. 
The heatsink is not really hot - certainly less hot than both the
harddisks and the heatsink on the graphics card.



/Per Jessen, Zürich

_______________________________________________
Prime mailing list
[email protected]
http://hogranch.com/mailman/listinfo/prime

Reply via email to