Trevor Boicey <tboicey@xxxxxxx> wrote: > Michael Thomson wrote: ... > > Memtest reported zero errors over 10 passes through the standard test suite. > > So it looks like the memory may be OK after all. ... > At the time, I didn't actually use memtest to diagnose it, I just > swapped memory around and found that crashing just followed that memory > stick. I'd be curious if that stick would actually fail memtest. It might not (so Michael, beware!). I've had DIMMs that would give rise to spurious data corruption (ugh) and crashes - but memtest86 would find no errors. Other DIMMs worked fine, though, so there was no doubt the DIMMs were at fault. Just to prove a point, I wrote my own little memory tester - one which should find any possible flaw given enough time :-). The trick is writing random data to random addresses (and it should read back in random order too, but there was no need to go that far). The good DIMMs tested okay; the faulty ones gave errors _within_seconds_. So much for the usefulness of regular patterns for memory testing :-). Michael, you may want to try another memory stick anyway... Good luck! - Reinoud