[CUWiN] good news everyone

[David Young]

On Sun, Aug 06, 2006 at 11:56:00AM +0200, Aaron Kaplan wrote:
> > None of this is meant to diminish from your point, though, that having
> > two antennas on virtually every WiFi device presents opportunities!
> > 
> Yes, the mail by jim thompson seems like quite an opportunity.
> Atheros home brewn-beamforming hm... nice 
> 
> anyway, point is: we guys in berlin, vienna and Cuwin should work more
> closely together. There is lots of know how and we tend to reinvent the
> wheel far to often. For example in VIE we are planning to create our own
> linux FW for www.osbridge.com (5gxi model) . We have nice experiences
> with it (1.7mBytes ~ netto transfer rate) and we want to have olsr on
> it. This setup is very nice pointo to point link with little
> interference in other parts of the mesh (5ghz -> many channels). Maybe
> at CUWin you would be interested to see how it performs with HSLS

Which MIPS CPU and bootloader is on the 5gxi?  Atheros system-on-a-chip?
I wish it had more ROM and RAM, and a different antenna.

> (BTW: what cpu power do you need for HSLS?)

In practice, hslsd uses too much CPU because we run it with so many
debugging messages turned on.  It is possible for the daemon to get into
a "livelock" condition, where it is too busy printing debug messages to
actually get work done.

In theory, hslsd will use less CPU than any other linkstate algorithm
running on a network of similar (large) size, both because of the
hazy-sightedness (fewer LSUs to process), and because my implementation
is a bit "lazy" about running shortest paths first.

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933