Anvil Serial Boards -- cheap trash or what?

Karl Denninger karl at ddsw1.MCS.COM
Mon Dec 26 15:43:31 AEST 1988


Anyone out there have an idea on what's going on with this setup?

We have just acquired an Anvil Designs 8-port I/O board.  The firmware
onboard is 1.0, diskette software 2.4.1 (so it says on boot).

The problem we're seeing is that one (count it, one!) port receiving at 19200 
is taking up to 90% of the board's CPU capacity!  Two 19,200 connections and
you start losing things.  This is a per-direction total (which makes things
even worse; a 19,200 SLIP connection would completely trash communications!)  
With a Telebit modem connected, and a 19200 terminal line, I can cause an 
incoming newsfeed on the modem to timeout and disconnect by doing a few ps's!

One good thing about this board is that it came with a nice diagnostic that
allows me to place the blame where it belongs.  It gives you a board-CPU
usage percentage on a real-time basis (updated once per second or so).....
No differences have been noticed when the monitor program was or was not
running (which indicates that it's contribution to load is inconsequential).

This board is rated for 8 ports @ 9600 baud in throughput.  That means I 
should be able to run 4 19200 lines -- at 19200 -- and not lose anything!  
What's going on here?  

Oh yeah -- for comparison SCO's stock "dumb AT" drivers do better than
this; we've been running a dumb Digiboard Com/8 at 38,400 for several months
now and had NO problems!  I expect more from a nearly-$1000 product when our
last purchase of an 8-port (dumb) card for $300, and $50 worth of 16550
chips worked better than this board does!

Ideas and suggestions (such as V2.4.1 drivers bite, get x.x.x) welcome!  Get
the ideas in quick if you have any; I just might drop-kick this thing through
the wall before too long ;-)

---
Karl Denninger (karl at ddsw1.MCS.COM, ddsw1!karl)
Data: [+1 312 566-8912], Voice: [+1 312 566-8910]
Macro Computer Solutions, Inc.    	"Quality solutions at a fair price"



More information about the Comp.unix.xenix mailing list