Xref: utzoo comp.sys.att:3473 unix-pc.general:795
Path: utzoo!utgpu!water!watmath!clyde!bellcore!faline!thumper!ulysses!andante!princeton!udel!rochester!bbn!uwmcsd1!ig!agate!pasteur!ames!ncar!noao!arizona!naucse!sbw
From: sbw@naucse.UUCP
Newsgroups: comp.sys.att,unix-pc.general
Subject: Re: 68020/68881 for Unix-PC
Message-ID: <730@naucse.UUCP>
Date: 9 Jun 88 14:06:37 GMT
References: <9592@g.ms.uky.edu>
Followup-To: comp.sys.att
Organization: Northern Arizona University, Flagstaff, AZ
Lines: 32

From article <9592@g.ms.uky.edu>, by david@ms.uky.edu (David Herron -- One of the vertebrae):
> I'd asked just that question a couple of months ago and didn't
> receive much in the way of response.

Ditto, though a number of people commented that there was difference
in the way the '010 and '020 expect the system stack to appear.
> 
> One thing that'll be different is the way you do floating point.
> Apparently on 68010's you have to do some fairly complicated things
> to access the '881 and usually that stuff is in the run-time

I think that starting with 3.5 there are environment variables to
get the C compiler to put out '881 instructions instead of calls
to floating point simulation software.  However, I also think
that there are some files (startup? libraries?) that turn out
to be missing if one tries to pretend there is an '881.

> It would be loverly to have an 020 in there tho'.

I would love that, as well.

By the way, I was (long ago and far away) once able to actually
get a price for part number 105160253 ("Math. Acc. Unit MC 68881")
that's listed in my "AT&T UNIX PC Service and Parts Ordering
Information" - $2,520.00 list.  Thanks, but no thanks.  (I also
believe that they decided not to market it - hmmm,  wonder why
it didn't sell?  After all, doesn't everyone want to spend $2500
for faster floating point?)

-- 
	Steve Wampler
	{....!arizona!naucse!sbw}