[net.arch] Slander on OC Async Buses

pauls@hammer.UUCP (Paul Sweazey) (06/12/86)

> Of course, with DEC emphasizing Unibus disks (e.g., RA81s) plugged into
> HSC50s and HSC70s, perhaps the I/O bandwidth issue is moot...I can't
> believe that the transfer rates over the Unibus is any faster than
> any of the more modern microprocessor busses.  I mean, last time I looked
> it was an open-collector, asynchronous bus (it did, after all, predate
> tri-state logic!)

I wouldn't dare to defend Unibus, but open-collector logic and
asynchronous protocols are not inherently slower than synchronous
buses or tri-state logic.  In fact, by a significant margin, the
two fastest industry standard 32-bit backplane buses are the
Futurebus and the Fastbus.  The keys to their speed are the use
of properly terminated transmission lines and the elimination of
clock skew.  Either of these buses can transfer data at rates in
excess of 100 Megabytes per second.
-- 

Paul Sweazey			{decvax,ucbvax,...}!tektronix!tekecs!pauls UUCP
Tektronix, Inc.			tekecs!pauls@tektronix		          CSNET
P.O. Box 1000, M/S 61-215	tekecs!pauls.tektronix@Udel-Relay       ARPANET
Wilsonville, OR 97070		Work: 503-685-2563	     Home: 503-692-5016