*BSD News Article 29975


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!yeshua.marcam.com!MathWorks.Com!news.duke.edu!news-feed-1.peachnet.edu!emory!swrinde!gatech!news.byu.edu!cwis.isu.edu!u.cc.utah.edu!cs.weber.edu!terry
From: terry@cs.weber.edu (Terry Lambert)
Newsgroups: comp.os.386bsd.misc
Subject: Re: FreeBSD system
Date: 27 Apr 1994 20:33:27 GMT
Organization: Weber State University, Ogden, UT
Lines: 52
Message-ID: <2pmi6n$44k@u.cc.utah.edu>
References: <1994Apr20.155710.2992@palantir.p.tvt.se> <5O+tb8n.dysonj@delphi.com> <CotA5p.30u@pegasus.com>
NNTP-Posting-Host: cs.weber.edu

In article <CotA5p.30u@pegasus.com> richard@pegasus.com (Richard Foulk) writes:
]>>: > The BT545S ISA SCSI controller is good too, but EISA is THE way to go!!!!
]>>: 
]>>: EISA is dead, PCI is *the* way to go ?!
]> 
]>Well, I am going to suggest the more conservative way to go.  PCI is still
]>fairly new and I believe that if someone is asking, then they are not
]>necessarily prepared to "experiment".  I agree PCI is in our future, but
]>EISA is solid.
]
]There is also VESA local bus, which is less expensive and much more
]widely available that either of the others.

And vastly less reliable than either of the others.

o	Do you have multiple VESA slots?  Which one is a master slot?
	IS there a master slot on your machine?

o	Without a master slot, you are not guaranteed cache writeback
	and/or flush on VESA bus-mastering DMA.  Almost all modern disk
	controllers are capable of bus mastering DMA and almost all
	modern OS's use it.  Your machine will crash if it gets the
	wrong data because bus mastering DMA is done wrong.

o	Many VESA cards hold the bus for a relatively long time during
	transfers.  With all VESA slots populated, are you sure that
	the agregate bus-on time won't screw up you DRAM refresh like
	it does for almost everyone else with populated slots?  You
	could reduce the bus on time for your disk controller (presumably
	living in a master slot), but then where is your VESA purchased
	performance?

Seriously, the previous poster was not blowing smoke when they said that
EISA is more solid.  I expect EISA to have 3-6 months of continued
dominance, and that in 6-9 months it will die out in favor of PCI.  The
main problem with PCI is that people like Gateway have rev'ed to the new
Saturn chipset, but still have not updated all their motherboards (the
486 motherboards have been rev'ed, but not their P5-60), so they still
can't do cache-writeback correctly (the fix for the old Saturn involved
hacking the hardware itself).  And you've got to be careful to avoid buying
old stock of now-working motherboards (ie: old chipset/workaround) or pay
a hefty (~50%) performance penalty.

So for right now, EISA is the most consistently reliable high performance
bus to get.


					Terry Lambert
					terry@cs.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.