Debugging bad memory problems

[Michael Powell]

jd1008 wrote:
[snip]
> One slight, and perhaps remote, possibility is that memory
> is a hair slower than what the memory controller expects,
> especially, as Valerie mentioned, under heavy memory load.
> On systems where the cpu clocking is unlocked, one might
> be able to slow down the cpu clock just slightly to see if the
> problem is mitigated.

Another consideration is the power supply. An aged power supply can be right 
on the raggedy edge ripple wise such that it seems OK when not under load. 
Place a load on it and the ripple shoots up. Easy to see with an 
oscilloscope. Not to mention the machine's behavior gets extremely erratic. 

If it works fine on a known good battery under battery only (pure DC), but 
flakes when plugged up to the AC mains...

I only bring this up because I have chased my tail looking for memory 
problems when it turned out to be a power supply about to go south. I doubt 
this is his situation though. He's most likely right that it is memory. I 
just include this to get away from 'tunnel' vision.

-Mike