[Beowulf] Exascale by the end of the year?

[Christopher Samuel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 06/03/14 03:07, Joe Landman wrote:

> I've not done much with MPI in a few years, have they extended it
> beyond MPI_Init yet?  Can MPI procs just join a "borgified"
> collective, preserve state so restarts/moves/reschedules of ranks
> are cheap?  If not, what is the replacement for MPI that will do
> this?

At SC13 there was a talk by Laxmikant Kale from UIUC about CHARM++
(used by NAMD for instance) at the "Practical Fault Tolerance on
Today's HPC Systems" tutorial which sounded like it had the abilities
to cope with faults as it is more a computational workload
distribution system (going on what was said about it).

Now you can build CHARM++ to use MPI as its underlying mechanisms but
I don't know how well it copes with failures in that mode.

cheers,
Chris
- -- 
 Christopher Samuel        Senior Systems Administrator
 VLSCI - Victorian Life Sciences Computation Initiative
 Email: samuel at unimelb.edu.au Phone: +61 (0)3 903 55545
 http://www.vlsci.org.au/      http://twitter.com/vlsci

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iEYEARECAAYFAlMX4NMACgkQO2KABBYQAh8uyACgk2t3COaBalPC/Lt5TT+/nrxW
X08An3ZPJXa7FuaXsWWua3h6sw+S9r3A
=KKKM
-----END PGP SIGNATURE-----