Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!highspl!fithp!mhw
From: mhw@fithp (Marc Weinstein)
Newsgroups: comp.sys.3b1
Subject: Re: v42bis 2400baud at 960cps throughput!!!
Message-ID: <1991May18.180655.2944@fithp>
Date: 18 May 91 18:06:55 GMT
References: <159@beyonet.UUCP>
Organization: Weinstein Consulting
Lines: 47

From article <159@beyonet.UUCP>, by beyo@beyonet.UUCP (Steve Urich):
>
> 	My friend and I uucp with our 3b1's. He uses the 3.5 and I use
> 	3.51m. He has a multitech modem with v42bis fixed link rate at
> 	9600. I have a T2500 at 19200 :-). He discovered how to switch
> 	from UUCP G protocol to E protocol. Boy what a difference in 
> 	thoughput!  
> 
> 	His throughput jumped from an approx avg. 200-226 to 317-960cps.
> 	My throughput jumped also the same.
> 	Not bad with a connect rate of only 2400 baud.

Jeez - that's a hell of a compresion ratio.  You sure you're seeing
this kind of compression?

We've seen UUCP (the xferstats file) report very high throughput, but
it's typically off due to short duration granularity problems.  Have you
demonstrated this kind of throughput sending, say, files of at least 50KB
in size?

>       The compression
> 	was really working. Even with compressed files it really works
> 	better.

E allows you to turn off the checksum ACKS in UUCP, which improves
throughput.  At 2400 baud, we were seeing about a 10% improvement in
speed using e vs g.  However, e assumes an error free connection, so
you should be using MNP or V.42 during the connection.  Also, using MNP4
for files already compressed using the UNIX compress utility is better
than MNP5, because the MNP5 compression algorithm is less efficient
than the UNIX version.  And, since MNP provides a pseudo-synchronous
connection, you eliminate the start/stop bits between modems and that
gives you another 20%.


> 	To enter the change to the protocol you must add ',eg' to the
> 	end of the Devices prefix: 
> 
> 	ACUM24,eg     tty000 

I've never seen an entry with both protocols specified here.  I'm not sure
what that means - sending vs receiving??  We just would use 'ACUM24,e'.

-- 
Marc Weinstein
{simon,royko,tellab5}!linac!fithp!mhw		Elmhurst, IL
-or- {internet host}!linac.fnal.gov!fithp!mhw