Re: utility for testing ram?

Rogier Wolff (R.E.Wolff@BitWizard.nl)
Wed, 30 Sep 1998 12:29:28 +0200 (MEST)


Mike A. Harris wrote:
> On Mon, 28 Sep 1998, Brandon S. Allbery KF8NH wrote:
>
> >The best thing I've ever found to test memory to destruction is Linux, with
> >OS/2 a close second. :-} Boot Linux and do a full kernel compile, if your
> >memory is at all flaky the compile *will* die with a signal.
>
> That is not necessarily true. It *MAY* die. Flaky memory,
> depending on the trouble, may or may not crash the computer. A
> lot of memory problems are intermittent. Also, even if a
> particular bit was stuck on, or stuck off, it may not affect
> something for a while, depending on what data or code actually
> hits that memory page.
>
> Use a memory checker program that does pattern testing. This
> tests memory very well for problems. There are such utils out
> there for Linux, and someone is likely to point them out in this
> thread.

The problem is that modern computers don't suffer from the problems
that the memory-tester-experts have come up with.

Some decades ago, they thought that memory when broken would be stuck-at
one or stuck-at zero. So the memory test

\updown W0 \updown R0 \updown W1 \updown R1

was designed.
(i.e. doesn't matter which way you do it, but write all zeroes. Then
in any order you like, check for all zeroes etc etc.)

As time went on, more possible failure modes were found and
correspondingly more complex memory tests were designed.
For example

\updown W0 \up R0W1 \down R1W0 \up R0

is not much more expensive, but catches quite a lot more errors. The
above memory test does 5 references per memory location. The most
elaborate memory tests have about 20 references per memory location.

However running those on a system that causes gcc to sig-11 rarely
shows a problem.

Roger.

-- 
| Most people would die sooner than think....  |    R.E.Wolff@BitWizard.nl 
| in fact, most do.  -- Bertrand Russsell      |     phone: +31-15-2137555 
We write Linux device drivers for any device you may have! fax: ..-2138217

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/