Return to BSD News archive
Newsgroups: comp.os.386bsd.bugs
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!bunyip.cc.uq.oz.au!munnari.oz.au!spool.mu.edu!sol.ctr.columbia.edu!howland.reston.ans.net!vixen.cso.uiuc.edu!newsfeed.ksu.ksu.edu!moe.ksu.ksu.edu!hobbes.physics.uiowa.edu!ljg
From: ljg@space.physics.uiowa.edu (Larry Granroth)
Subject: Re: 3com509 driver for FreeBSD??
Message-ID: <Ct58Dz.6CF@space.physics.uiowa.edu>
Sender: news@space.physics.uiowa.edu
Nntp-Posting-Host: ljgpc.physics.uiowa.edu
Organization: The University of Iowa, Department of Physics and Astronomy
X-Newsreader: TIN [version 1.2 PL2]
References: <19940626.122208.045870.NETNEWS@VM.BIU.AC.IL> <JKH.94Jun29065700@nx.ilo.dec.com> <2urbmb$gv@ici-paris.ensta.fr> <CsutED.I89@space.physics.uiowa.edu> <2vvot7$mhv@orion.cc.andrews.edu>
Date: Mon, 18 Jul 1994 15:58:46 GMT
Lines: 30
Andrew Gillham (gillham@andrews.edu) wrote:
: In article <CsutED.I89@space.physics.uiowa.edu> ljg@space.physics.uiowa.edu (Larry Granroth) writes:
: >
: >I'm using the ep drivers with a 3c509 in a FreeBSD-1.1.5.1 system
: >and notice similar behavior writing to a SCSI disk on an Ultrastore 14F
: >(I think) controller. I notice some weird options on the DOS setup disk
: >for modem baud rates and client/server . . .
: >Does anyone know what these settings really mean? Is the 3c509 programmed
: >to lock out interrupts for variable amounts of time or somesuch?
: >(Maybe the SCSI controller has similar configurations to be tweaked?)
: Well... I don't *really* know what they mean.. but if you do a file
: save in 3c5x9cfg to create a 'config file', then edit the config file
: it has an option that says something about minimum interrupt release
: time, or something like that. (ok, so I can't remember... ) Anyway
: I would think that picking server and a high baud rate would 'optimize'
: the settings for allowing other interrupts/devices access to the cpu/bus
: whatever...
: If you select client/no modem and save it, then server/38400 baud and
: compare the files you'll get an idea what I'm babbling about.
: It's late.. I'll give up on this for now. :-)
I finally got around to trying this, but the results were the opposite
of what I thought you were suggesting. For the {server, 38400} setting
the ftp transfer rate for a large binary file to disk was only 43 kB/s and
over 900 kB/s to /dev/null. For the {os/2 client, 1200} setting the
rates were a little more even at around 200 kB/s and 700 kB/s.
I'd be happier if I could get the disk transfer up by another 100 kB/s.
larry-granroth@uiowa.edu