Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: net.unix-wizards
Subject: Re: Looking for VAX microcoding tools
Message-ID: <3359@utzoo.UUCP>
Date: Tue, 22-Nov-83 20:36:00 EST
Article-I.D.: utzoo.3359
Posted: Tue Nov 22 20:36:00 1983
Date-Received: Tue, 22-Nov-83 20:36:00 EST
References: <6268@unc.UUCP>
Organization: U of Toronto Zoology
Lines: 15

I'm not sure about the 750 and below, but I saw a study several years
ago which implied that microprogramming the 780 was a dead loss.  Seems
the user-microcode feature on the 780 is basically a marketing ploy, not
a useful piece of hardware.  The dispatching overhead for user-defined
instructions (i.e., the time needed for the 780 to realize that it's
got a user-microcode instruction and call your microcode) is so large
that your user-defined instruction better be doing something really heavy
or else it will be a net speed *loss*.

I have also heard that it is nearly impossible to get documentation from
DEC describing how 780 microcode works, even if you buy a 780 with the
writeable-microcode box!
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry