[Gllug] Underclocking an Athlon

[Nick Hill]

I run an Athlon 800. I have slowed the CPU fan down to reduce noise and wear on the bearings 4700 to 3500rpm. 

The CPU temperature is normally 7 centigrade higher than the overall temperature in the machine case. This raises to 15 centigrade when the processor is busy (eg compiling). I leave the side of the case off, so the system temp is normally ~30c.

In general, Athlons dissipate less heat, for a given performance, than a P4. 

It is very unlikely for a heatsink to fall off a CPU as in tom's hardware test. I don't think the test proves anything other than it is really important to make sure the heatsink is sitting properly on the CPU with a clean surface on both the CPU and heatsink, with high performance heat transfer paste joining the CPU and heatsink.

There are so many factors affecting system reliability other than the processor. All these can give the impression of processor unreliability:

1) Memory reliability
2) Quality of motherboard layout
3) System thermal management
4) Quality of motherboard chipset
5) Power supply quality
6) Capabilities of motherboard-mounted switch mode voltage regulator. This supplies the high-current low-voltage requirements of the core. (how stable is the output? Can it cope with max current for long periods? Can the capacitors be relied upon long term? Does it cope well with fluctuations in current requirement by CPU?)

Also remember, Hard drive reliability (far more important than CPU reliability!) is heavily dependant on airflow over the drive and drive PCB. I know of many hard drives which have failed due to apparent overheating of the spindle motor drive chip. Spindle motor bearings also suffer badly under the effects of temperature. (I have fixed many HDDs! it is 30:70 head crash:PCB failure)

Put a thermometer in your machine case. If (temp_case - temp_room)>20 centigrade, this is bad. Also, be careful where you run ribbon cables. These can cause a spaghetti in the case, slowing/ preventing airflow. Be particularly careful not to cover motherboard chips with ribbon cables.

Regards

Nick.


On Mon, 1 Oct 2001 12:49:33 +0100
John Edwards <john_ed at cornerstonelinux.co.uk> wrote:

> On Mon, Oct 01, 2001 at 11:42:25AM +0100, Jackson, Harry wrote:
> > Hi all
> > 
> > For all you overclockers out there this may be of some use.
> > 
> > Download GIMPS and run the test if it fails reduce the clocking until it
> > does not. I done this on Saturday and reduced my 1GHz Athlon From 1.1GHz TO
> > 1050MHz before I could get it to work properly. I know clocking it to 1.1GHz
> > was a bit daft but I had to see how far I could take it.
> > 
> > Regards;
> > Harry Jackson.
> 
> As the big Athlons have a habit of being as hot as a burning coals 
> when running in a small machine, is there any way to UNDERclock them 
> (eg running a 1.4GHz at 1.2GHz) ?
> 
> I would rather have a 10% slower machine than a dead machine.
> 
> -- 
> #------------------------------------------------------------------------#
> |    John Edwards    Email: John.Edwards at uk.com                          |
> |                                                                        |
> |  "Many solutions were suggested for this problem, but most of these    |
> |   were largely concerned with the movements of small green pieces of   |
> |   paper, which is odd because on the whole it wasn't the small green   |
> |   pieces of paper that were unhappy." - Douglas Adam on unhappiness    |
> #------------------------------------------------------------------------#
> 
> -- 
> Gllug mailing list  -  Gllug at linux.co.uk
> http://list.ftech.net/mailman/listinfo/gllug

-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug