Scyld Nodes Freezing w/ SMP (fwd)

Alvin Oga alvin at Mail.Linux-Consulting.com
Fri Nov 14 16:00:46 EST 2003


hi ya

On Fri, 14 Nov 2003, Lombard, David N wrote:

> From: Alvin Oga
> [deletia]
> > 
> > if the latest kernel has no effect, than there's some other
> > serious hw problems ... timing issues ??
> > 	- make sure the kernel is compiled for athlon and not p4
> > 	and smp enabled
> > 
> > 	- memory clock speeds, marginal memeory sticks
> > 	( get rid of generic no-name-brand memory sticks
> > 		- swap memory sticks and see if the problem
> > 		follow the memory ( keep good track of it
> > 		so you can easily identify it if all the memory
> > 		was thrown on the floor all at the same time
> > 
> Instead of guessing, try memtest86 at
> http://www.memtest86.com/memtest86-3.0.tar.gz

i have yet to see memtest find a failure thats real
or pass the memory that works 
in a given system ... where you know the system crashes

and yet moving the mem stick to another system give you
identical failures or passes the "application running tests"

other memory testors
	http://www.Linux-1U.net/Diags/#Mem
	http://www.Linux-1U.net/Memory/#Test

> The easiest way to use it is download the package, copy the pre-built
> binary to a floppy, and boot the node with the floppy.  It will run
> until you stop it.
> 
> This should point to a memory problem if it exists.  Make sure you read
> the docs, as there are some Athlon-specific comments IIRC.

have fun
alvin

_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf



More information about the Beowulf mailing list