Mailinglist Archive: opensuse (3666 mails)

< Previous Next >
Re: [SLE] memtest86 over 4 hours
  • From: Michael James <Michael.James@xxxxxxxx>
  • Date: Wed, 23 Mar 2005 09:59:06 +1100
  • Message-id: <200503230959.07126.Michael.James@xxxxxxxx>
On Wed, 23 Mar 2005 01:17 am, Fergus Wilde wrote:
> On Tuesday 22 March 2005 13:07, steve wrote:
> > How long should an athlon 1800 take to complete the memory
> > test? Can I assume from this that my memory is bad?
>
> Did it show any errors? If not, then no,
> it doesn't mean anything bad that it ran for hours.
> Usually if there are egregious problems with the memory
> they
> will show themselves relatively quickly,
> but memtest writes a whole bunch of patterns
> to the memory to test it.
> The whole thing could go on exhaustively for ages.
> But I must say if I had an error free 4 hour run
> I would give up
> and assume the memory is OK
> unless you have unexplained crashes
> or other reasons to believe it's bad.

Exactly.
If you just want to see if a second hand stick of ram is any good,
a run of a few hours should settle it.
No errors listed below the bar = good.

_IF_ you are having crashes, memtest is a good first test.
If a run of a few days doesn't produce anything,
then it's probably safe to rule out memory errors
and look elsewhere for the problem. (Don't forget the power supply)

I once had a memory error so subtle
I could run memtest all weekend and only get 1 glitch.
It was giving me a couple of crashes a week though
and each crash left files that hung any process that read them.
You can imagine how unusable my desktop got
once there were a few of these pitfalls around.
Replacement RAM and fsck.reiserfs --fix-fixable cured it.

FWIW,
michaelj

--
Michael James michael.james@xxxxxxxx
System Administrator voice: 02 6246 5040
CSIRO Bioinformatics Facility fax: 02 6246 5166

< Previous Next >
Follow Ups
References