Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!decwrl!labrea!rutgers!columbia!cunixc!cck
From: cck@cunixc.columbia.edu (Charlie C. Kim)
Newsgroups: comp.protocols.appletalk
Subject: Re: KFPS Upgrade kit
Message-ID: <237@cunixc.columbia.edu>
Date: Tue, 10-Nov-87 23:09:35 EST
Article-I.D.: cunixc.237
Posted: Tue Nov 10 23:09:35 1987
Date-Received: Fri, 13-Nov-87 20:50:12 EST
References: <871110154145.1097@bl-rtsg>
Reply-To: cck@cunixc.columbia.edu (Charlie C. Kim)
Organization: Columbia University
Lines: 34


First, one thing... Let's not confuse people by calling any other
version except Bill Croft's version of software for the Kinetics box
KIP (revisions 10/86, 2/87, 9/87).  Previous versions of the Stanford
code can probably be best referred to as the Seagate or UDP code...
Kinetics offers their own DDP gateway or their combined DDP/UDP
gateway software.  In addition, I believe they have or will have
EtherTalk combatible software (e.g.  EtherTalk on EtherNet instead of
lap (sorta - closer to ddp) on Ethernet).

The upgrade, as I understand, basically consists of a new ROM and some
ancillary documentation (which I'm not even sure relates especially to
the new ROMs).  From what I've seen the new ROM (I think the version
they are updating are 3.0) has the following major points:
	o uses fewer buffers internally (at least 3.5 or so less)
	o fixes the old reset bug - you can now issue reset and
	  restart from Prompt (I thought Prompt II was supposed 
	  to work around this probem but it doesn't see to have)
 These ROMs seem very solid and robust on both the Fastpath I's &
II's.  I believe they have other code and bug fixes added, but I could
only guess at it.

I'm not sure what you mean by a single ethernet buffer - all versions
of the software (both Kinetics and Stanford) have been running for
quite a while with multiple buffers in both receive and transmit.  I
agree input from enet is too slow..  I just wish the hardware/software
was fast enough to keep up with 8-16 back-back packets from even a
VAX8700.  Hopefully, this can be fixed in software by speeding up the
ethernet service routines...

Charlie C. Kim
User Services
Columbia University