[Click] Click CPU load

[Torquato Bertani]

Yes, it's a click performance problem. I think that without polling
drivers click is to heavy for a low resource hw. You have to optimize
everything to reach high performances.

I don't know how meraky guys can reach a 802.11g-like performances
with a 180MHz mips...Does anyone know?

Bye
Thor

On 11/10/06, Roberto Riggio <roberto.riggio at create-net.org> wrote:
> On Fri, 2006-11-10 at 12:11 -0500, Bob Keyes wrote:
> > I would be interested to see how you've done this. I believe that roofnet
> > requires support for pseudoIBSS mode, and that using adhoc may cause
> > problems with BSS splitting.
>
> I just put the interface in adhoc mode and I've changed the init script
> removing all the elements that deal with the monitor mode
> (athdecap, ...).
>
> Of course the BSSID partitioning remain a problem, but for some quick
> tests it is ok. I've managed to get the nodes ping each other and reach
> the Internet thought a gateway.
>
> Anyway as far as I've understood from a discussion with Torquato the
> performance are so bad because there is no support for polling in both
> the madwifi drivers and the broadcom drivers.
>
> Bye
>
>
> _______________________________________________
> click mailing list
> click at amsterdam.lcs.mit.edu
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>