From owner-freebsd-hackers  Mon Nov  6 10:07:51 1995
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id KAA05283
          for hackers-outgoing; Mon, 6 Nov 1995 10:07:51 -0800
Received: from blob.best.net (blob.best.net [204.156.128.88])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id KAA05260
          for <freebsd-hackers@freebsd.org>; Mon, 6 Nov 1995 10:07:28 -0800
Received: from geli.clusternet (rcarter.vip.best.com [204.156.137.2]) by blob.best.net (8.6.12/8.6.5) with ESMTP id KAA20354; Mon, 6 Nov 1995 10:05:57 -0800
Received: from localhost (localhost [127.0.0.1]) by geli.clusternet (8.6.12/8.6.9) with SMTP id KAA24338; Mon, 6 Nov 1995 10:02:53 -0800
Message-Id: <199511061802.KAA24338@geli.clusternet>
X-Authentication-Warning: geli.clusternet: Host localhost didn't use HELO protocol
X-Mailer: exmh version 1.6.4 10/10/95
To: Mark Diekhans <markd@grizzly.com>
cc: bde@zeta.org.au, freebsd-hackers@freebsd.org, jkh@time.cdrom.com,
        julian@ref.tfs.com
Subject: Re: NPX still broken in 2.1.0-951104-SNAP... 
In-reply-to: Your message of "Mon, 06 Nov 1995 08:09:20 PST."
             <199511061609.IAA08229@Grizzly.COM> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Mon, 06 Nov 1995 10:02:53 -0800
From: "Russell L. Carter" <rcarter@geli.com>
Sender: owner-hackers@freebsd.org
Precedence: bulk

This is an interesting discussion of an area with an immense amount
of work behind it.  More portable numerical software is relying on
specific numerical tests that have well defined (by the IEEE standard)
results for given inputs, but are ruined by the various unimplemented
pieces such as the given examples show.

David Hough was until recently probably one the most energetic people
in this area.  He produced a test suite that compared a system's actual
response to the "correct" response in many different ways.  I'm going
to try dig this up.

It's easy to get religious about the area, but it is hard figure out
the best tradeoff between implementation and performance.  I'd prefer
if the IEEE behaviour could match the behaviour of Sun, SGI, IBM, and
DEC, in that order.

Russell
rcarter@geli.com

Russell