[Beowulf] ECC exerciser/exorciser?

Robert G. Brown rgb at phy.duke.edu
Mon Jan 26 10:26:00 PST 2009


On Mon, 26 Jan 2009, Mark Hahn wrote:

> - do you have or know of a good exerciser for testing ECC's?  yes, I know 
> about memtest86, but I'm more curious about a load that could be run under
> linux.  my thinking is that ecc's are triggered by bad reads, so something
> which allocates all memory and then continually reads it would be best.

I've never tried "allocating all memory", but you can definitely do
things close to what you want to do with benchmaster:

http://www.phy.duke.edu/~rgb/General/benchmaster.php

It's pretty simple source (and open) so perhaps you can hack it if
necessary to get it to do what you want.  Although if you run it for a
really really big memory allocation maybe it will "just work" as an
exerciser even where it is pretty tedious waiting for an actual
benchmarking run to finish.

It has a random option that lets you really whack the memory subsystem,
BTW, avoiding streaming access altogether and defeating the cache
scheme to the extent possible given the size of cache and the size of
the vector being tested.

Let me know if it works for you (if you try to use it).  Oh, and there
is also lmbench -- that should similarly be usable as a memory exerciser
with minimal tweaking.  I don't remember if Carl even implemented a
random access algorithm, though.

    rgb

>
> - how about layout of memory -> dimms?  take a single page, for example:
> I presume that the first cacheline (16B) will be "striped" across both 
> channels of one bank (for instance, the first dimm-pair.)  is it normal
> for the 17th byte to begin on the next dimm-pair (csrow)?  dmidecode seems to 
> indicate that 8 1GB dimms are mapped to contiguous addresses
> (which would imply no channel interleaving, which is wrong...)
>
> - does "numactl --hardware" work correctly for you?  I see something like:
> available: 2 nodes (0-1)
> node 0 size: 5375 MB
> node 0 free: 3550 MB
> node 1 size: 4095 MB
> node 1 free: 3874 MB
>
> 9470 MB total, which, on a machine with only 8x 1GB dimms is unexpected...
>
> thanks for any comments,
> mark hahn
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit 
> http://www.beowulf.org/mailman/listinfo/beowulf
>

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu





More information about the Beowulf mailing list