Path: utzoo!attcan!uunet!lll-winken!ncis.llnl.gov!helios.ee.lbl.gov!pasteur!ucbvax!decwrl!sgi!vjs@rhyolite.SGI.COM
From: vjs@rhyolite.SGI.COM (Vernon Schryver)
Newsgroups: comp.arch
Subject: Re: History Question: Origin of fork()
Summary: nostalgic meanderings
Message-ID: <25259@sgi.SGI.COM>
Date: 20 Jan 89 20:16:52 GMT
References: <43676@linus.UUCP> <5608@phoenix.Princeton.EDU> <23246@beta.lanl.gov>
Sender: daemon@sgi.SGI.COM
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 63

In article <23246@beta.lanl.gov>, hwe@beta.lanl.gov (Skip Egdorf) writes:
> In article <5608@phoenix.Princeton.EDU>, haahr@phoenix.Princeton.EDU (Paul Gluckauf Haahr) writes:
> > In article <43676@linus.UUCP> dee@linus.UUCP (David E. Emery) writes:
> > > Who "invented" fork?  What system was the first to use fork?
> > > (where fork did the Unix thing, i.e.  created a complete copy of the calling
> > > process as a new O.S. process)
> > 
> > in the unix retrospective from the 1978 bstj, dennis ritchie attributed
> > the idea to genie, the operating system for the sds-9 (from sigma data
> > systems, later xerox data systems)
>...
> Ref (2) is
> 2. L. P. Deutsch and B. W. Lampson, "SDS 930 Time-sharing System Preliminary
>    Reference Manual," Doc. 30.10.10, Project GENIE, Univ. Cal. at Berkeley
>    (April 1965)


Scientific Data Systems, SDS, built a nice line of "scientific"
machines (think of Nova's but the previous generation; if you don't
recall the Nova, then think of PC's used in labs to do data analysis
and instrument control).

I understand that people at Berkeley (the names above were among the
powerful) took an SDS-930 (24-bit words, 16-bit address, 1.75 usec
cycle time, LDA was 2 cycles, STA 3 cycles, cpu was 3 bits wide, all
silicon, mostly 4"x5" (?) boards with a couple of gates or maybe a flip
flop except the CTE (serial comm.) which was racks of surface mount
stuff--4 cards/channel) and added paging (without extending the cycle),
user/system mode, and the return-to-user-trap to make the 940.

There was serious talk of a follow-on from SDS, the 9000 series, but
instead, the Sigma series came out, to the disappoint of some.  Then
Xexox bought SDS, turned it into XDS, and ran it into the ground--or so
it seemed from a 1000 miles away.  The 940 was <<The>> timesharing
machine for a long time--all 45 (?) that were built were cherished.
They figured heavily in the histories of Tymshare and Comshare, as well
as a few odd places, including the famous San Francisco coop.  I've met
people who ran chunks of the 940 system under some kind of PDP-10 (?)
emulator.  There was the 1-board implementation in Hawaii, and the
magnificent dream of the BCC-5000 (or was it 500?).  "My" 940 was
scrapped only a few years ago by the Dept of Comm.

I don't think the 930 could run any flavor of TSS--no paging, no user mode.
The 910 and 920 were smaller and slower.  I think only 1-bit cpu's.

There was also a tape file system...

(This is certainly partly wrong, since it is based on my failing memory.)

Can anyone who thinks that fame from a job done well is lasting list
all of the Famous Names from project GENIE?  In their day, they were as
influential among people who might have read this newsgroup as any of
those who have worked on 4.xBSD.

My trivia questions:  was the 940 syspop "sti" a predecessor of the 4.x BSD 
	ioctl TIOSTI as fork(2) was?  Or just a case of parallel evolution?

    Was Deutsch's QED an ancestor of ex?


Vernon Schryver
Silicon Graphics
vjs@sgi.com