Path: utzoo!utgpu!watserv1!watmath!att!dptg!ulysses!andante!princeton!njin!rutgers!cs.utexas.edu!uunet!mcsun!hp4nl!philapd!ssp2!pb
From: pb@idca.tds.PHILIPS.nl (Peter Brouwer)
Newsgroups: comp.realtime
Subject: Re: 386 C compilers??
Message-ID: <623@ssp2.idca.tds.philips.nl>
Date: 16 Jul 90 08:51:33 GMT
Reply-To: pb@idca.tds.philips.nl (Peter Brouwer)
Organization: Philips Information Systems, Apeldoorn, The Netherlands
Lines: 23
Summary:
Expires:
References:<29@dvpdev.UUCP> <455@newave.UUCP>
Sender:
Followup-To:
Organisation: Philips Information Systems, Apeldoorn, The Netherlands
Disclamer: This is mine opinion alone.
Keywords:


 In article <455@newave.UUCP> john@newave.mn.org (John A. Weeks III) writes:
>In article <29@dvpdev.UUCP> ken@dvpdev.UUCP (Ken Greenebaum) writes:
>>We are interested in the 386 PC-Unix, or MSDOS platforms, and are evaluating
>>these products:
>>    Metaware Hi-C 386 C compiler
>
>Generates nice fast code, but don't ever consider using it for
>floating point math.  The floating point code it generates is
>inaccurate and has no provisions for error detection.

I have not seen inaccurate results but have experienced that it only works 
correct in combination with a coprocessor or 3.0 floating point emulation 
library. It will give not correct results with a 3.2 floating point emulation
library.
I have had contact with metaware about this. They state that both the microsoft
and AT&T 3.2 version are not correct.

-- 
Peter Brouwer,                # Philips Information Systems,
NET  : pb@idca.tds.philips.nl # Department SSP-P9000 Building V2,
UUCP : ....!mcsun!philapd!pb  # P.O.Box 245, 7300AE Apeldoorn, The Netherlands.
PHONE:ext [+31] [0]55 432523, #