Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wotan!moxie!lobster!ralph!mondy!mdm
From: mdm@mondy.UUCP (Michael D. Mondy)
Newsgroups: comp.unix.sysv386
Subject: Re: Summary: 4-port cards
Message-ID: <313@mondy.UUCP>
Date: 8 May 91 23:50:32 GMT
References: <FW0P84N@geminix.in-berlin.de> <2PBZ13w163w@aegis.or.jp> <8PBQUOT@geminix.in-berlin.de>
Organization: Private System -- Lafayette, LA
Lines: 37

davidg%aegis.or.jp@kyoto-u.ac.jp (Dave McLane) writes:
> [ ... ]
>The reason that I ask I am running FAS 2.0.8 with both NS16550AFN 
>chips and the UARTs that are integrated with other functions on my 
>Dell 333D; FAS accepts both and works with both ... until you see 
>how much CPU time is being used. The significant reduction in time
>would seem to indicate FAS is using the FIFO buffers in the
>NS16550AFNs but not in the others....
> [ ... ]

In article <8PBQUOT@geminix.in-berlin.de> gemini@geminix.in-berlin.de (Uwe Doering) writes:
> [ ... ]
>If you feel that the UARTs on your main board consume too much CPU
>time, then they are no NS16550A chips. At least I don't know of
>any NS16550A chips that are integrated with other functions, so
>I think that there are rather NS16450 or 8250 chips on the main
>board. [ ... ]

   My Dell 325 has a VL16C452-QC chip in it.  This is a chip made by
VLSI (what a name) and is functionally equivalent to two 16450s and
and parallel port chip.  I believe that it may be replaced with
a VL16C552-QC to get the equivalent of 16550s with fifos but I have not
tried this.  Before deciding to go with an intelligent serial port I found
out that this chip may be purchased from Arrow at +1 800 932 7769 for $59.03.

   Does anybody know how VLSI's chips compare to the other manufacturer's
chips?  Just checked a databook - national makes a 15552.

   Has anyone at Dell tried using the fifo version of the chip on the 
motherboard?  Is Dell still using the non-fifo chips in their machines?
Even the 486 monsters intended to be used as unix workstations?!?
For a just few bucks more, a very expensive machine could be transformed from
something with ridiculous serial i/o performance to a box with reasonable
serial i/o performance.  [ And if my experiences are representative, I'd bet
that a ton of money might be saved in customer support. ]

Michael Mondy