On Fri, Aug 29, 2008 at 01:38:14PM -0700, Jeff Johnson wrote:

Your other options are a benchtop DIMM tester (very expensive, several grand) or using a hardware only PCI test card from Ultra-X.

The problem I've found, and why of the numerous memory problems, only
one was caught by memtest is that the problem often isn't the memory
device in isolation.  This is especially true in light of the question
in the subject which involves running a memory part out of spec.

I do feel that memtest is a very useful tool.  I had a machine that
was corrupting files written to disk and occasionally crashing.
memtest allowed me to quickly identify a stuck bit in a memory module
and replace that memory.

But, that's the only time I've ever had it find anything.  My other
problems have always been interaction problems.

Also, with the Linux memory management system, a kernel compile will
pretty fully exercise at least a couple of GB of memory.  Agreed, not
all problems will be caught or noticeable, but since gcc builds a lot
of data structures with pointers, bad memory tends to cause segfaults
which show up fairly readily.

David


--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to