Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site arizona.UUCP
Path: utzoo!watmath!clyde!burl!mgnetp!ihnp4!arizona!whm
From: whm@arizona.UUCP (Bill Mitchell)
Newsgroups: net.unix-wizards
Subject: Gould UNIX Systems
Message-ID: <15264@arizona.UUCP>
Date: Fri, 7-Sep-84 03:43:47 EDT
Article-I.D.: arizona.15264
Posted: Fri Sep  7 03:43:47 1984
Date-Received: Wed, 12-Sep-84 02:05:54 EDT
Organization: Dept of CS, U of Arizona, Tucson
Lines: 43

Several weeks ago I posted an article asking for comments about what
people thought of Gould UNIX Systems.  I received many notes asking
for copies of what I found out, but I received no notes that offered
information about the Gould systems.

I've had some opportunities to use some Gould systems during the last
few weeks and I've found them to be very satisfactory for the most part.
I have noted a few problems with them however:
	
	Their current release is 4.1c.  They're talking about having
	 4.2 by January.

	They have some general problems with debuggers.  Apparently,
	 breakpoints can only be set on word boundaries.  This seems to
	 have the effect that breakpoints are sometimes missed.  Also,
	 if one tries to set an adb breakpoint at x, it will actually
	 be set at x+4.  I found this to be very irritating.  In addition,
	 dbx has loads of problems.  I was using it on a very small
	 program and I blew it up a number of times without even trying.

	They currently don't have Lisp or Pascal.  However, a third
	 party has alledgedly ported Franz Lisp.  I think they will
	 also be offering LNS Pascal.

	On a 32/67, which they claim to be 50% faster than a 780, I've
	 run a number of benchmarks (no floating point though) that
	 have come out with performance about equal to that of a 780.
	 I'm going to be getting some time on a 97 very soon and they
	 claim that it's 4-5x a 780, but I'm predicting that I'll see
	 about 2.5-3x a 780.  (On the other hand, I'll be very surprised
	 if the 790 doesn't have exactly the performance that DEC is
	 promising.)

I suppose that if I was making the decision tonight as to whether to go
with a 790 or a big Gould system (the 97), I'd go with the 790 because
while Gould does present attractive performance advantages and attractive
price advantages (perhaps made marginal by some whopping DEC discounts)
I think that the software hassles (and accompanying lack of them on a 790)
would cancel out Gould's strong points.

					Bill Mitchell
					whm.arizona@csnet-relay
					{noao,mcnc,utah-cs}!arizona!whm