Article View: comp.terminals
Article #33Re: Network based character terminals
From: Chip Colby
Date: Tue, 03 Nov 1992 00:28
Date: Tue, 03 Nov 1992 00:28
70 lines
3818 bytes
3818 bytes
Adam Thompson (thompsn@ccu.umanitoba.ca) wrote: ----- MUCH DELETED ---- : : Perhaps you aren't aware ... most ethernet connections displaying ASCII on : a screen don't get throughput much better than 9600. For example, I have an : X terminal on my desk, with an "ASCII-terminal" mode, that is basically an : "ANSI" terminal with a telnet connection. It gets throughput of about : 10000 bps, on average. On some of our 486sx PCs, we have tested NCSA telnet : and MS-Kermit v3.11[3.12] under FTP software's tcp/ip drivers, and Novell's : LAN Workplace for DOS[Windows]' tcp/ip drivers. We have acheived ascii : dump rates of about 20000bps, _MAX_. There was also a test configuration I : saw once, which was: : 486DX-66Mhz/8Mb RAM : 16-bit WD ethernet card : FTP Software's TCP/IP drivers (commercial) : FTP Software's Telnet/FTP package (commercial) : : Results were, reportedly, ascii dump rates of about 45kbps. Now, this : should give you some idea of what you're up against. I find these remarks rather curious. Here is what I have observed over months of testing: (figures converted, then rounded, from cps to bps match with your numbers) 1. A Dumb terminal connected through a terminal server performs at "bursts" of up to about 30,000 bps, depending on terminal server and network load. 2. A PC performing VT100 emulation over a network connection can perform at a peak rate of near 50,000 bps, this time depending entirely on network load. 3. X terminals are inhernetly slow, so this is one reason why I don't want to use one. 10,000 bps is probably low, with 20,000 bps being typical. Don't confuse X terminal performance with raw ASCII performance as there is no comparison. Your display manager must interpret the data and, display it in an open window. Either your PC emulation package is incredably slow (a condition not unheard-of), or I am missing something. To belabor the point, a 286 at 16 Mhz, running FTP software rloginvt performs at a sustained rate of about 45,000 bps. In other words, it screams. Of course, that's all it's doing. I just don't want to license this software with every terminal we need to attach to our product. Ok, so I'm cheap. : : I _CAN_ suggest another solution: use an intelligent mux/demux setup, : which can interpret (for example) VT100 emulation. This can remove some : of the redundancy in the data stream. Be prepared to search long and hard, : however -- and to pay through the nose. I've only heard of these, not : actually seen them or detail on them. I suggest you contact DEC,Tektronix, : Cisco,Western Digital... perhaps also Wyse, Wang... This is an interesting idea, one I had not considered. I shall investigate. Of course, since I'm cheap... ------------------------------------------------------------------------------- | Chip Colby - Software Engineer | Chip Colby - Watchmaker | | Digital Systems International, Inc. | Vintage Timepieces | | "Voicelink (R) Intelligent Dialing Systems" | Member NAWCC and AWI | | 6464 185th Ave. NE | 20006 NE 168th PL | | Redmond, WA 98052 | Woodinville, WA 98072 | | (206) 869-4551 | (206) 788-4900 | -------------------------------------------------------------------------------
Technical Details
Message-ID:
<1756@dsinet>Path:
arc1-ca.pugleaf.net!sparky!uunet!nwnexus!ole!rwing!dsinet!jaws!chipcReferences:
<1992Oct31.030551.23693@ccu.umanitoba.ca>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!nwnexus!ole!rwing!dsinet!jaws!chipc From: chipc@jaws (Chip Colby) Newsgroups: comp.terminals Subject: Re: Network based character terminals Message-ID: <1756@dsinet> Date: 3 Nov 92 00:28:26 GMT References: <1992Oct31.030551.23693@ccu.umanitoba.ca> Sender: root@dsinet Lines: 70 X-Newsreader: Tin 1.1 PL5