Xref: utzoo comp.sys.mac.comm:536 comp.sys.mac.hardware:4257
Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uwm.edu!zaphod.mps.ohio-state.edu!usc!apple!portal!cup.portal.com!ts
From: ts@cup.portal.com (Tim W Smith)
Newsgroups: comp.sys.mac.comm,comp.sys.mac.hardware
Subject: Re: nubus ethernet card for macII
Message-ID: <31507@cup.portal.com>
Date: 7 Jul 90 08:18:02 GMT
References: <1187@s5.Morgan.COM>
Organization: The Portal System (TM)
Lines: 39

< Dove FastNet III is like $395 most places, but MacUser says it's slow.

Has anyone been able to duplicate the MacUser tests?  For example, one
of the tests was supposed to be raw throughput, which they say they
tested with some expensive network analyzer.

I tried to measure raw throughput of my FastNet III card.  I don't have
any network analyzer, though.  Here is what I did:

	1. Modify a Dove FastNet SCSI box so that it is running an
	Ethernet echo program.  This program simply takes incomming
	packets and blasts them back out.

	2. Write a program for my Mac II that opens the .ENET driver
	and blasts out packets for a fixed amount of time.

	3. The program from step two counts how many echo packets
	come back in that fixed amount of time.

Using this program, I got reasonable performance from FastNet III.
Using packets of 1500 bytes, I got a lot higher throughput than MacUser
claimed for the fastest card.  If I can find a way to duplicate the
MacUser results, I can find out why FastNet III was slower and then
fix it (I wrote the drivers that Dove is using).

Now the FastNet III card should theoretically be slower than some
other NuBus cards, because it has 16 bit memory mapped into a 32
bit address space (i.e., addresses 0,1,2,3,4,5,... on the card
appear in the Mac address space as S00000,S00001,S00004,S00005,
S00008,S00009,..., where S is the slot number), which means that
getting data on and off of the card is a little slower than it
would be for a full 32 bit card, but this is not a big enough
difference to cause the MacUser results.

I seem to remember some MacWorld benchmarks that also had FastNet III
slower than some other cards, but not by a lot.  In fact, by about the
amount one would expect based on the 16 bitness of the card.

						Tim Smith