[comp.unix.sysv386] Multiport I/O cards: intelligent or dumb?

toad@cellar.UUCP (Tony Shepps) (04/11/91)

We're running a BBS under SCO Unix.  Right now, we have two lines, but in the 
near future we'd like to expand to three or four.

The tough question is whether to go with the cheap non-intelligent multi-port 
card or the more expensive intelligent version.

We are running HST dual-standard modems exclusively, so we expect to always 
communicate with the modems at 38,400 baud.  Beyond the BBS users and news 
processing, there is little activity on the system.  Of course, we can 
visualize a situation where 4 people are all downloading at 14,400 baud; but 
would there be a noticeable difference with the intelligent I/O?

Any hints would be *greatly* appreciated.  Thanks in advance!

----------------------------------------------------------------------------
-       Tony Shepps     toad@cellar.UUCP     (...uunet!cellar!toad)        -
- The Cellar BBS; since late 1990.  Reliable hardware, responsible sysops, -
- and line noise is neve~ a prob&~Jq@#}.}a\[>~~~    NO CARRIER

grant@bluemoon.uucp (Grant DeLorean) (04/13/91)

toad@cellar.UUCP (Tony Shepps) writes:

>The tough question is whether to go with the cheap non-intelligent multi-port 
>card or the more expensive intelligent version.

 IMHO, you will be extremely pleased if you get an intelligent board. Just
with the CPU not having to service every character coming in makes it
worth buying the intelligent board.

 Make sure you put DigiBoard at the top of your list to check out. They
have superb tech support if you need it (free, too), their boards work
extremely well, they always have the current drivers available for download
if you want to update the driver (another free service), they have drivers
for darn near everything on an ISA, EISA or MCA type machine, the boards
and cables are very well made, they support hardware as well as software
handshaking, up to 56kbps on each port sustained...   I switched to them
from another intelligent board and am now completely sold on them.
-- 
\  Grant DeLorean  (grant@bluemoon)    {n8emr|nstar}!bluemoon!grant  /
"You need only reflect that one of the best ways to get yourself a reputation 
as a dangerous citizen these days is to go about repeating the very phrases
which our founding fathers used in the struggle for independence."-C.A. Beard

gemini@geminix.in-berlin.de (Uwe Doering) (04/13/91)

toad@cellar.UUCP (Tony Shepps) writes:

>We're running a BBS under SCO Unix.  Right now, we have two lines, but in the 
>near future we'd like to expand to three or four.
>
>The tough question is whether to go with the cheap non-intelligent multi-port 
>card or the more expensive intelligent version.
>
>We are running HST dual-standard modems exclusively, so we expect to always 
>communicate with the modems at 38,400 baud.  Beyond the BBS users and news 
>processing, there is little activity on the system.  Of course, we can 
>visualize a situation where 4 people are all downloading at 14,400 baud; but 
>would there be a noticeable difference with the intelligent I/O?
>
>Any hints would be *greatly* appreciated.  Thanks in advance!

You could install the FAS 2.08 serial driver for dumb ports (posted to the
net some months ago, it's freeware). If you use it together with NS16550A
UART chips, even four full speed downloads at the same time are no
problem with that combination. And, most important, it _won't_ bring your
machine down to it's knees during the downloads. FAS 2.08 has full-duplex
RTS/CTS hardware flow control and other nice features not found in
conventional dumb port drivers. It runs unter SCO UNIX and most other
286/386 UNIX or XENIX flavors.

     Uwe
-- 
Uwe Doering  |  INET : gemini@geminix.in-berlin.de
Berlin       |----------------------------------------------------------------
Germany      |  UUCP : ...!unido!fub!geminix.in-berlin.de!gemini

davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (04/14/91)

In article <9Xw8Z2w164w@cellar.UUCP> toad@cellar.UUCP (Tony Shepps) writes:
| We're running a BBS under SCO Unix.  Right now, we have two lines, but in the 
| near future we'd like to expand to three or four.
| 
| The tough question is whether to go with the cheap non-intelligent multi-port 
| card or the more expensive intelligent version.

  I would get an AST 4 port (someone was closing them out for $99) and
try it. If it drops chars or eats too much CPU install FAS. If it still
has problems convert the USARTs to 16550s so FAS can use the fifo.

  Minimal price, minimal non-SCO software. Stop enhancing when you get
the performance you want.
-- 
bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen)
    sysop *IX BBS and Public Access UNIX
    moderator of comp.binaries.ibm.pc and 80386 mailing list
"Stupidity, like virtue, is its own reward" -me

larry@nstar.rn.com (Larry Snyder) (04/16/91)

toad@cellar.UUCP (Tony Shepps) writes:

>processing, there is little activity on the system.  Of course, we can 
>visualize a situation where 4 people are all downloading at 14,400 baud; but 
>would there be a noticeable difference with the intelligent I/O?

yes - 4 high speed lines all locked at 38400 on a
dumb multiport board will load the system -

we at one time used all dumb serial ports - but about a year ago
we installed intelligent boards - along with more high speed modems -

and will all the modems going wide open (with zmodem transfers at high speeds)
the system is so fast it can't be noticed!

-- 
   Larry Snyder, NSTAR Public Access Unix 219-289-0287 (HST/PEP/V.32/v.42bis)
                        regional UUCP mapping coordinator 
               {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}

root@equinox.UUCP (Super user) (04/18/91)

In article <9Xw8Z2w164w@cellar.UUCP> toad@cellar.UUCP (Tony Shepps) writes:
>We're running a BBS under SCO Unix.  Right now, we have two lines, but in the 
>near future we'd like to expand to three or four.
>
>The tough question is whether to go with the cheap non-intelligent multi-port 
>card or the more expensive intelligent version.
>
>We are running HST dual-standard modems exclusively, so we expect to always 
>communicate with the modems at 38,400 baud.  Beyond the BBS users and news 
>processing, there is little activity on the system.  Of course, we can 
>visualize a situation where 4 people are all downloading at 14,400 baud; but 
>would there be a noticeable difference with the intelligent I/O?
>
IMHO, with the number of lines you have (two with plans to expand to four),
a dumb serial card should fit your needs. Of course, if money isn't an
object, get a smart card (suchas the...oh skip it...).

In November 1989 a friend of mine, Terry Neu, benchmarked several
intelligent cards. He also included a dumb 8-port card. What was kinda
surprising at the time was that the dumb card did better than about
half of the "intelligent" cards all the way up to 19.2K baud. Benchmark
tests don't favor dumb cards, they output continously. For an dumb card
to outperform an intelligent card at 19.2K baud over 8 ports is a
surprise. In the real world, where users wouldn't press the card as
heavy as the benchmark, the dumb board should even work better. The
smart card will make a difference in buffering (giving users 'snappy'
screen refreshes) under the light load. Now, the serial drivers for 
the intelligent cards have probably improved their efficiency since
then too. Still, it's something to think about....


_________________________________________________________________
Arf, Yipe, Grrrrrrrr          Equinox Systems (305) 255-3500 x289
Death to the Shah!            14260 SW 119th Ave, Miami FL 33186
              WolF Kozel : Hobbies...Unix,kali,chess & beer

larry@nstar.rn.com (Larry Snyder) (05/12/91)

root@equinox.UUCP (Super user) writes:

>IMHO, with the number of lines you have (two with plans to expand to four),
>a dumb serial card should fit your needs. Of course, if money isn't an
>object, get a smart card (suchas the...oh skip it...).

not if the ports are all locked at 38400 - plus - the overhead
associated with those speeds will slow the system down..

We did run multiple dumb ports - and throughput was reasonable -
but the system seemed sluggush - and throughput for console IO
tasks under X were quite slow.  After adding a smartboard, 
the throughput on the serial lines increased around 15% but the
console IO was real fast.  The sluggush response problems also
ceased.

-- 
      Larry Snyder, NSTAR Public Access Unix 219-289-0287/317-251-7391
                         HST/PEP/V.32/v.32bis/v.42bis 
                        regional UUCP mapping coordinator 
               {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}