> I believe V.34 is full duplex, but the sum of the bandwidth in both directions
> is 28.8k. Ie, if you you're punding it in both directions at once, you
> should see about 14.4k thruput each way. A 56k DDS has 56k in each direction.
Lets look at v.32bis which I know well. This is considered to be 14400 bps
and I know from experience transferring precompressed files (.ZIP) that
you can generally get 1650 chars per second (Z-modem) through that kind of
link. Since 14400 divded by 8 bits per char is 1800 and there must be some
overhead to handle the v.42bis protocol this seems to show that a v.32bis
modem really can send 14400 bps in one direction at a time. In the BBS
world there are alternate bi-directional protocols like Hydra that will
send in both directions simultaneously and I have tested them and found
that I still get at least 1500 cps or better on transfers. This indicates
that v.32bis is capable of 14400 bps in both directions simultaneously.
The documentation for Hydra and other bi-directional protocols points out
that all modems 2400 bps and greater are fully bi-directional except
those using USR's HST protocol. Since this info predates v.34 it may not
apply there but I have never read anywhere that v.34 achieved it's high
speeds by stealing bandwidth from the backchannel so I don't think this
to be true. I believe that v.34 and v.34+ are fully bi-directional at
28,800 bps for both channels.
Since the discussion started with async/sync DSU's does anyone remember
some timing tests reported on an ISP list sometime within the past 6
months were they found that the async/sync conversion dropped the
effective bandwidth to little more that the async alone?
Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-542-4130
http://www.memra.com E-mail: michael@memra.com