On 9/22/05, Martijn van Buul <martijnb@atlas.ipv6.stack.nl> wrote:
> It occurred to me that Andy Ruhl wrote in gmane.os.netbsd.general:
> > On 9/22/05, Martijn van Buul <martijnb@atlas.ipv6.stack.nl> wrote:
> >> It occurred to me that Andy Ruhl wrote in gmane.os.netbsd.general:
> >> > On 9/19/05, Andy Ruhl <acruhl@gmail.com> wrote:
> >> >> Is anyone using the atu driver with a USB 801.11b adapter? If so, I=
'd
> >> >> like to know what adapter it is.
> >> >
> >> > Didn't get a response, 1 more try and copying netbsd-help this time.
> >>
> >> Sorry, I must've missed it.
> >>
> >> I used to be using an USB atu device indeed - I was using an SMC2662 V=
.4.
> >> However, I wasn't very happy with the performance of it, so I switched=
 to
> >> using a Cardbus card using ral(4).
> >>
> >> However, I still have the SMC 11b nic, so if there's anything you want=
 me
> >> to do..
> >
> > Any idea if the performance problem was related to the adapter or the d=
river?
>
> I'm not sure; it's probably a combination of both:
>
> 1) Reception dropped off really fast. Using the same access point, the
>    atu device had mediocre reception unless I was in the very same room a=
s
>    the access point, compared to my current ral(4) device. To put it this=
 way:
>    unless I was sitting almost next to the access point, it would always =
be a
>    bit of a gamble whether I could get a usuable connection. The ral(4)
>    device just works (TM). I don't think this is a NetBSD specific issue,
>    as I had the same problems under Windows.

Ok, that sounds like a bad adapter then. Either it's a bad one or it's
low quality in general if that's the case.

> 2) Running NFS over wireless using the atu device was next to impossible,
>    even when using NFS-over-TCP. Reducing the blocksize helped, but only =
a
>    little bit. Same applies to samba; it just wasn't reliable, leading to
>    rather nasty stalls all of the time. Using the same samba mounts under
>    windows *did* work good, so I'm not sure the NetBSD drivers have nothi=
ng
>    to blame. For the record: NFS over my current ral device works like a
>    charm. I can still compile a kernel over NFS when I'm sitting in the
>    garden when using the ral(4) network card. Using the atu(4) device,
>    chances that trying to play an MP3 while in the same room as the acces=
s
>    point will likely result in a stale NFS mount.

Yeah but all bets are off if you have to sit next to your access point
for the thing to actually work. There's only so much the protocol can
do if the hardware doesn't cooperate.

> > I'm mulling over getting one of these so I have a fairly complete
> > "toolbox" of usb devices which work under NetBSD.
>
> Well, maybe I caught the "bad specimen", but I wouldn't advise them to
> anyone.
>
> If someone would like to pick up the challenge: I'll happily ship my
> USB atu card to whoever feels like (s)he could impove the performance of
> these critters.

Heh, not *that* card, thank you very much! I'll see if I can find
another one and I'll report my results if possible. If the card's
reception is that bad it's probably not even worth bothering with.

Thanks, this has been useful.

Andy