Software modems

[Brian Kelsay]

I don't think that's how the manufacturers justify removing the chips from
the modem and moving it to the faster PCI bus.  In fact, the software modems
I have used have been considerably slower than a hardware modem on ISA bus.
Granted the CPU I had was probably just the minimum I could use but still
the speed difference was ridiculous.  Also the system crashed about five
times the day I finally yanked the card out.  Maybe someday Lucent or
somebody will have good software to improve a PCI soft modem the way it
should be done, but by then it will be irrelevant.

Brian

-----Original Message-----
From: A Duston [mailto:hald at sound.net]
Sent: Thursday, August 17, 2000 7:29 PM
To: KCLUG
Subject: kclug - Software modems

Here is an article I found that might give a valid reason 
for doing the modem in software.  
"Bandwidth and Latency: It's the Latency, Stupid" by Stuart Cheshire
Part 1 http://www.tidbits.com/tb-issues/TidBITS-367.html#lnk4
Part 2 http://www.tidbits.com/tb-issues/TidBITS-368.html#lnk4

I'm not Ed, so I will summarize.  Basically what it says is that 
since hardware modems are doing compression, they wait for 50 
milliseconds of no data before flushing the buffer on down the 
wire.  As most modems are used for ppp connections today, the PC 
sends a packet, and then the modem waits 50 ms before it sends 
it down the wire.  If the modem is in the software, the packet can 
be sent immediately since the computer is in control of buffering. 

Hal Duston
hald at sound.net
If Al Gore invented the internet, why is it named after George W. Bush?

P.S. Jim, thanks for the comment at the meeting about my modem, 
I checked and an interrupt conflict was the problem.