[MLB-WIRELESS] Wireless Cluster ???

[dwayne]

oh look, a subject line designed to catch Dwayne's eye :)


> Matt Pearce wrote:
> 
> Hey all,
> 
> Would it be possible and feasible once we have a reasonable mesh of
> pc's running together to run a wireless clustering system using linux
> and "possibly" sell our combined proccessing power to research
> companies to pay for shared internet access for all ???.


Ahhhh, hello  :)

Yes indeedy, this is far and away the reason I got into wireless stuff
in the first place.


> Please dont dredge up all other old wars on this one.  All I am asking
> is would it be possible and feasible and what sorta performamce would
> the slow wireless cards make instead of hardwired LAN's ??  Obviously
> it would also depend on the performance of the server its running on
> as well but that an equation that is akin to how long is a peice of
> string.

This depends on a bunch of things, internode latency, memory latency,
processing power per node, the cluster architecture *what sort of
problems you plan on solving using this gear* etc. etc. etc.

We can do *stuff*, what sort of stuff and how valuable it is we will
probably have to find out :)
I'm pretty sure that pretty much every aspect of this is being covered
right now but i'm pretty sure that I'm ot aware of anyone doing wireless
cluster research.

Hmm. Google.

Awp:

Searched English pages for wireless cluster.    
Results 1 - 100 of about 120,000. Search took 1.81 seconds 

Anyway, yes, go for it, in the middle of distributed computing fiddlings
at home so it will be fun to try linking systems up over our network  :)

As far as using it to pay for things, excellent idea. Fiddly. Worth a
go. Count me in.



Dwayne

To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message