🚀 go-pugleaf

Here in comp.os.os2.setup, David Moorhouse
<davidm{spamfence}@ariel.ucs.unimelb.edu.au> spake unto us, saying:

>Stan Koper wrote:
>> 
>> Right now I'm getting downloads of about 3250 cps (is it
>> cps or bps?), not 56k for sure, but 25% faster than my
>> old 28.8 card.
>> 
>It's cps - Characters Per Second.  A 56K modem running at
>its theoretical maximum of 56x1024 bps - Bits Per Second -
>should be able to deliver 56x1024/10 = 5734 cps, assuming
>1 start bit, 8 bit data, 1 stop bit, and no compression.

Keep in mind, however, that most error-correction protocols (at least
MNP4 and LAP-M, the two common ones in our own asynch modem context)
actually reframe the data, essentially removing the start and stop bits
which otherwise would accompany each data byte and replacing them with
a smallish header every hundred or so bytes (I admit I don't remember
precise numbers and it varies a little with the specific implementation,
but that's ballpark IIRC).

In a nutshell, a typical connection using EC will approach 8 bits/data
byte, and not 10 bits/data byte.

>Bps is not the same as 'baud', which strictly is transitions
>per second.  In the old days, a 300baud modem was 300bps,
>since each transition was 1 bit.  Nowadays, the modems
>encode 4 bits or more for each transition.

You bet.  I've lost track since the release of Rockwell's V.FastClass,
but I remember that even the old 14.4kbps HST protocol was 6 bits/baud.

-- 
-Rich Steiner  >>>--->  rsteiner@skypoint.com  >>>---> Bloomington, MN
         OS/2 Warp 4 + Linux + Executor = PC Hobbyist Heaven!
          CIRCUMVENT:  The result of a botched circumcision!