[App_rpt-users] Jim and Ramesh (chan_voter and RPi)

[Shane Morris]

Hi all,

I was thinking, even though the RaspberryPi can't handle the audio
stream from a URI due to the limitations of the USB bus, what is the
ethernet performance like? I imagine it wouldn't be terribly good, due
to the fact the ethernet swings off the USB, but hear me out...

chan_voter is a module that communicates via ethernet, am I correct?
It uses an RTCM connected over an ethernet bus to communicate the
voice data around, and I can imagine, at a low codec rate like GSM or
even Speex, throughput wouldn't be as excessive as raw sound data
coming from a fob.

So my proposal is, before Jim and Ramesh throw out the baby with the
bathwater, and consign the RPi effort to the dust, why not take the
work, which I believe involves getting Zaptel/ DAHDI to work with the
RPi, and apply it to a purely chan_voter system? Yes, I know the
chan_voter is dependant upon app_rpt, and in the future, someone,
somewhere, is going to try to hook up a sound fob, and miserably
complain it doesn't work, but we get the same questions about Asterisk
1.8 over and over too, right?

The advantages here are manyfold - again, low power consumption comes
into the game, everyone knows I personally need as low a power
consumption as I can get. The fact the RTCM data (ie, VoIP over
ethernet) is much more suited to the infrastructure on the RPi is a
good thing here too. The fact that normal Asterisk just runs, and runs
well, on RPi is another motivating factor - my friend Andrew is
running his house PABX on an RPi running Asterisk 1.8 I believe, there
is an image for this. And yes, simply the cool factor - who else can
say they run their voting repeater system off something the size of a
credit card, with all the proprietary, messy, rack mounted stuff out
there?

If Jim and Ramesh are unwilling, or unable, to help, due to the fact
that the RPi is a bit of a letdown (I heard about the USB performance
issues - ouch!), I would like to take their work, and apply it to a
platform I think may have a bit better USB performance - its based on
a Cortex-A8 (ARMv7) - the Cubieboard. For all you I/O junkies out
there, what can YOU do with 96 I/Os? For $49, gets you something good,
look it up, www.cubieboard.org - I got one for Christmas, and I
thought it would be good for this. Its a Chinese Allwinnner A10
device.

Since I have gone beyond just simple URI systems since Jim introduced
me to the RTCM, I would like to get this to work. I recognised the
RTCM did something I was talking to Andrew about on the way home from
Expedition in October of this year - "Wouldn't it be nice, Andrew, if
we could have a voting system that allows us the flexibility to run an
Expeditions communications in such a way that make it simply easy for
us, and easy for the users?" Before, we've had users change channels
in different places, this inevitably leads to confusion. With a
chan_voter system, we can run the same UHF CB channel across the
arena, and access it anywhere.

Andrew had said to me "Shane, do you know how much such a system would
cost...?" Yes, Andrew, I had the Simoco, and I bought the RTCM for
$250 second hand ^.^ Sorry for being a smart-alec, I know you meant
how much would a PROPRIETARY system cost us?

Anyway, I hope I have Jim and Rameshes blessing on this one - they did
some great work, and I know it must've been a letdown to find out the
hardware simply wasn't up to the task. Believe me, this isn't the
first story I've heard about the lacklustre I/O performance of the
RPi, and it won't be the last. But by using their work on a much
stronger (I hope!) platform that it still ARM Linux based, I hope to
circumvent at least one problem with the RPi. And its only $11 more
before shipping!

Looks like I'm going to be reading the Allwinnner A10 device spec
sheet now, huh? =)

Thanks for putting up with my ramble everyone.

Shane.