Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!zds-ux!gerry
From: gerry@zds-ux.UUCP (Gerry Gleason)
Newsgroups: comp.arch
Subject: Re: Fiber channels (was Re: The Killer Micro From Hell)
Message-ID: <107@zds-ux.UUCP>
Date: 16 Jan 90 18:33:08 GMT
References: <34030@mips.mips.COM> <4322@nttmhs.ntt.JP> <39807@ames.arc.nasa.gov> <3101@umn-d-ub.D.UMN.EDU> <40043@ames.arc.nasa.gov> <47800@sgi.sgi.com> <94@zds-ux.UUCP> <28758@amdcad.AMD.COM> <EMV.90Jan12140628@urania.math.lsa.umich.edu> <28795@amdcad.AMD.COM>
Reply-To: gerry@zds-ux.UUCP (Gerry Gleason)
Organization: Zenith Data Systems
Lines: 22

In article <28795@amdcad.AMD.COM> ching@pepsi.AMD.COM (Mike Ching) writes:
>-In article <28758@amdcad.AMD.COM> ching@snap.amd.com (Mike Ching) writes:
>-   At AMD we have toyed with the idea of building a SCSI cable using our TAXI
>-   transceivers (32-140Mbit serial data rate) and fiber.

<I was a bit careless. The $75 is component cost. Retail would be 2-3x that?
<You could send data over 2 kilometers using fiber. Using copper would make
<the component cost $40 and the distance 30-100 meters depending on the
<quality of the coax. Anyway we decided that the low demand for differential
<SCSI meant it wouldn't have wide appeal either. (BTW, we weren't trying
<to build these. Just considering it for an application note for the chips.
<Our business is ICs.)

Let's be clear about this.  I assume you are talking about an assembly
with SCSI connectors on both ends, and modules that convert to and from
the fiber.  In this case wouldn't the costs be much more equivalent since
the transceivers and conversion electronics more or less replace the
SCSI interface circuitry?  Of course SCSI drives are available from many
people now, and to do what I'm suggesting means you'd have to almost make
your own drive controllers, but why worry about such details ;-).

Gerry Gleason