Xref: utzoo comp.sys.laptops:2657 comp.sys.ibm.pc.misc:9100
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!uunet!stanford.edu!agate!e260-1e.berkeley.edu!c60b-1eq
From: c60b-1eq@e260-1e.berkeley.edu (Noam Mendelson)
Newsgroups: comp.sys.laptops,comp.sys.ibm.pc.misc
Subject: Re: Comm programs, Crosstalk Mk IV
Message-ID: <1991Apr27.053539.10761@agate.berkeley.edu>
Date: 27 Apr 91 05:35:39 GMT
References: <1991Apr21.071509.13167@news.iastate.edu> <1991Apr21.085326.28167@agate.berkeley.edu> <1991Apr27.031643.23049@wlbr.imsd.contel.com>
Sender: root@agate.berkeley.edu (Charlie Root)
Organization: University of California, Berkeley
Lines: 28

In article <1991Apr27.031643.23049@wlbr.imsd.contel.com> mcc@WLV.IMSD.CONTEL.COM (Merton Campbell Crockett) writes:
>				In file transfers through a 2400 baud
>modem link using 512 byte packets and 3 windows, an effective baud rate of
>2510 to 2560 is typical transfer rate for text files.  The effective baud
>rate decreases as you decrease the packet size; however, I haven't noticed
>any improvement in the effective baud rate as you increase the packet size
>from 512 bytes.

The 2510-2560 bps figure for text data is due to Kermit's RLE compression,
_not_ due to windowing.  It's fairly difficult to get around the 240 cps
maximum for 2400 bps transfers.  Some protocols like SUPERK establish a
synchronous connection with the other end and eliminate the start/stop
bit bottleneck (MNP-3 and above do this too, but at a hardware level).
This allows for a theoretical 300 cps transfer rate at 2400 bps.  However,
this method is very error-prone, and SUPERK is the only file transfer
protocol that I know of which utilizes it.
If you want to get unbiased results, and you can't turn off Kermit's
compression, try running some text data through crypt(1) and then sending.
Also, if you have clean lines, I would recommend that you set the packet
size as high as possible, especially at higher speeds (to minimize the delays
caused by ACKs).  At 2400 bps, 512 byte packets are OK; at 9600, I would
definitely suggest 1K or greater.

-- 
+==========================================================================+
| Noam Mendelson   ..!ucbvax!web!c60b-1eq       | "I haven't lost my mind, |
| c60b-1eq@web.Berkeley.EDU                     |  it's backed up on tape  |
| University of California at Berkeley          |  somewhere."             |