[NTLUG:Discuss] PC Hardware Issue

[Robert Pearson]

On 12/21/06, steve <sjbaker1 at airmail.net> wrote:
> Dennis Myhand wrote:
> > Lenrek Xunil wrote:
> >> I built my own PC.  It's got an Abit motherboard, AMD Athlon 64 CPU,
> >> 1.5gigs of ram (two cards), dual graphics card,
>
> Dual graphics cards?  Aha!
>
> Are they nVidia cards using 'SLI' mode?
>
> If so, I would say for 100% sure that one of your two graphics cards
> has crapped out - possibly from the heat - but it could just be failing.
>
> If you have two graphics cards using nVidia's "SLI" then the two
> cards share the rendering.  If one of them fails, the result is indeed
> a horribly "jaggy" display.
>
> So - I recommend you remove one of the two cards and see if the system
> stabilises.  If it doesn't then try the other graphics card by itself.
>
> Then you know that either:
>
> 1) The card you removed is bad...or...
> 2) The system is overheating when there are two cards - but not when
>     there is only one.  This would not be surprising because those
>     high end graphics cards chuck out a lot of heat.
> 3) Your power supply isn't up to driving such a high end configuration.
>     Modern graphics cards eat a LOT of power (that's why they generate
>     so much heat).
>
> So if this fixes things - try replacing the 'working' graphics card
> with the one you pulled out.
>
> If the system is still stable - then there is nothing wrong with either
> graphics card but your cooling or your power supply is not adequate for
> both of them running at the same time.
>
> If they system does not stabilize - then you've found the bad graphics
> card.
>
> Even if your graphics cards are not in SLI mode - the odds are good that
> you are overheating or running out of power.
>
> To find out if you're overheating, get a really good household 'box
> fan', take the lid off the computer and blast air into it as hard as
> you can.  If it still fails - it's not overheating!
>

Great troubleshooting procedure.

With regard to heat problems.
Does monitoring, or taking the exhaust air temperature help determine problems?
Any ideas out there about the "normal" temperature across a box?
Measuring airflow CFM (cubic feet per minute) is a little tough but
that would be  good to know for designing a solution. Brute force is
probably cheaper. But isn't too much air flow worse than not enough?

In my box the SATA drive got so hot you couldn't touch it. The guy at
the computer store said his did the same and recommended either one of
those "cool" enclosures or the "attach to the SATA drive" dual fan
module. I just parked the SATA drive and used an IDE (PATA) I had. I
did determine the fans were adequate but the airflow pattern missed
the SATA drive. Bad case design.
I'm thinking eSATA for the future. Get the heat sources out of the case.