Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!cbosgd!ihnp4!qantel!intelca!oliveb!glacier!decwrl!pyramid!amiga!skipper!cbmvax!daveb
From: daveb@cbmvax.cbm.UUCP (Dave Berezowski)
Newsgroups: net.micro.amiga
Subject: Re: Picking a terminal emulator
Message-ID: <637@cbmvax.cbmvax.cbm.UUCP>
Date: Sat, 16-Aug-86 10:07:15 EDT
Article-I.D.: cbmvax.637
Posted: Sat Aug 16 10:07:15 1986
Date-Received: Tue, 19-Aug-86 04:27:39 EDT
References: <232@neoucom.UUCP> <148600124@uiucuxc>
Organization: Commodore Technology, West Chester, PA
Lines: 62

> >>AmigaTerm (our terminal program, soon to be shipped with the modem
> >>for the Amiga, and maybe seperately) has a slightly higher throughput
> >>than a DEC VT101 (but slightly lower than a CIT-101).
> >
> >I'm not sure this is all that impressive (sorry, Andy).  A VT100 has only
> >a 64 character buffer, which fills up pretty quickly at 9600 baud.  The
> >real question is, what is the maximum baud rate for which AmigaTerm doesn't
> >need flow control?
> 
>     that could depend... unlike the vt100, the amiga has many kilobytes
> available for buffering.  use enough, and even long bursts of data won't
> overrun before statistics average the rate down.  just set the io_RBufLen
> field while you're setting baud rate, etc, in a SDCMD_SETPARAMS.  default
> is 512, so there's plenty of room for improvement there.
> 

	Under v1.1, AmigaTerm had a measured throughput of approx
9100-9200 baud. I don't know what it will be under 1.2 but since 1.2
has faster text output it should be at least this fast. What Andy meant
by the VT100 - CIT-101e comment was that if you put a VT100, AmigaTerm,
and a CIT-101 beside each other and say 'cat' some text to it; the CIT-101
will finish first, then AmigaTerm, then the VT100.

	You comment regarding using a super large buffer bears some discussion.
What happens when you xoff the host? Does the terminal program xoff the host and
then a)empty its buffer or b)halt output to the screen immediately. If you chose
'a' and you have a very large buffer (say >= 1K) then a delayed effect is produced
which makes the user think that the xoff didn't work. (You know, the user hits
xoff, but 1K of data gets printed after that.)  Method b solves the above problem
but in a way gets you out of sync with the host. If a user xoff's the host and
then decides to abort the output altogether (lets say with a ctrl-c to the host)
the user will still get the 1K of data that's in the local buffer.

	VT100's and CIT-101's get around this problem by using method 'a' and
a relatively small buffer (as you  mentioned, only 64 bytes). Since the smallest
buffer possible is 512 bytes (when using the serial device) AmigaTerm (under 1.1)
uses method 'b' as beta-testers found method 'a' un-acceptable.

	I don't know what method the final released version of AmigaTerm uses
as 1.2 enhancements/bug fixes were made after I left CBM. I should also mention
that beta-testers and myself found little difference (if any) using a CIT-101
or AmigaTerm at 9600 baud (ie. 400 out of 9600 baud [a 4% difference] is not
easy for a human to notice under most conditions).

	Besides throughput speed, the other features that may make AmigaTerm
desirable are:

	a) full support of ALL VT100 character sets
	   (US, UK, graphics, special graphics, extended gfx, vt52, vt52 gfx).'
	b) a 128 column (with the above true as well) which under most
	   conditions operates as well as 132 columns.
	c) 24(non-interlace) or 49(interlace) lines.
	d) full VT100 emulation (with the exception of blinking LED support,
	  (double height/width characters, etc).
	  (ie. it was tested with emacs, vi, edt, etc).

	Hope this helps.

	Regards, David Berezowski

Disclaimer: I no longer work for CBM and my comments are products of
	sparatic movements by my fingers.