Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site unisoft.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!godot!harvard!seismo!hao!hplabs!pesnta!amd!dual!unisoft!pc
From: pc@unisoft.UUCP (Paul Campbell)
Newsgroups: net.lang.pascal,net.lang.c
Subject: Re: Re: opt cmplrs vs. opt prgrmrs (really coding VMS in assembler)
Message-ID: <392@unisoft.UUCP>
Date: Thu, 17-Jan-85 11:11:40 EST
Article-I.D.: unisoft.392
Posted: Thu Jan 17 11:11:40 1985
Date-Received: Mon, 21-Jan-85 04:36:36 EST
References: <285@harvard.ARPA>, <4922@utzoo.UUCP>
Organization: UniSoft Corp., Berkeley
Lines: 14
Xref: watmath net.lang.pascal:205 net.lang.c:3984

(...)

	I heard a different story .... they coded the VMS kernel in assembler
because they didn't have a Vax .... (VMS was designed on a simulator running
on a pair of 11-70s) they built the hardware alongside the operating system
and tuned the instruction set as they wrote the operating system. The
context switch instructions were only decided on once the new what contexts
VMS would need to switch. Most of the compilers available for VMS on the day
the Vax was released were written in compatability mode and came straight from
RSX.

				Paul Campbell	ucbvax!unisoft!pc