[Beowulf] portable clusters
James.P.Lux at jpl.nasa.gov
Wed Nov 30 07:28:04 PST 2005
At 10:36 PM 11/29/2005, DGS wrote:
>On Tue, Nov 29, 2005 at 09:43:19PM -0800, Jim Lux wrote:
> > At 09:26 PM 11/29/2005, Mark Hahn wrote:
> > >> The goal here is to have a credible concept to improve the "system"
> > >> performance by adding on a computatational element to an existing
> > >portable
> > >> system that is Windows based without requiring any hardware changes to
> > >the
> > >> windows system, or without requiring significant software mods to the
> > >> windows system (i.e. running some new program is ok, running a windows
> > >> emulator is not).
> > >
> > >this seems fairly quixotic to me, since you also seem to be planning on
> > >some sort of small cluster of token (via, etc) processors. it's rather
> > >easy to imagine cases where a decent ruggedized 4x2 linux box would easily
> > >outperform a 16x via cluster.
> > Yep.. but you're still faced with the network boot issue, cluster or
> > no. Actually, it appears that the real challenge isn't the "cluster" per
> > se, but the whole "diskless linux boot from Windows laptop" process.
>How about a custom live CD with Warewulf? Boot your laptop from the CD,
>boot up your nodes, re-boot into Windows, and off you go. The nodes need
>DHCP, PXE, and "warewulfd" (modified HTTP server) at boot-up, but can
>stand-alone after that. You'd probably have to make a new CD if you
>want to modify the cluster. After you've done that once, it shouldn't be
>too hard to do again.
Gotta be able to run the Windows based external hardware AND feed the data
off to the cluster for processing at once. In theory, one might be able to
run it in something like Virtual PC, but that's been problematic in the past.
James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
More information about the Beowulf