Re: Numerical Computations With GCC

Article View: gnu.gcc.help

Article #52036

#52036

From: Mark Kent
Date: Tue, 23 Feb 1993 18:57

16 lines
1135 bytes

(I tried mailing to smith@minerva.harvard.edu, but minerva
 has the typical mail problem:

<<< 553 minerva I refuse to talk to myself
)

On a Sun when you get out of the range of normalized numbers
(i.e.,  smaller than min_normal() or bigger than max_normal())
the fpu *really* slows down.

Could it be that with slight diffs between gcc and sun cc that you
spend more time in the sub-normal range?  (since it converges I assume
you are not past max_normal()).

-mark

Technical Details

Message-ID: <1993Feb23.185717.21227@leland.Stanford.EDU>

Path: arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!nntp.Stanford.EDU!patience.Stanford.EDU!kent

References: <SMITH.93Feb21175110@minerva.harvard.edu>

Show Full Headers

Complete Article Headers

X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 11318c,497432ef72672d8
X-Google-Attributes: gid11318c,public
X-Google-ArrivalTime: 1993-02-23 10:57:17 PST
Newsgroups: gnu.gcc.help
Path: arc1-ca.pugleaf.net!sparky!uunet!stanford.edu!nntp.Stanford.EDU!patience.Stanford.EDU!kent
From: kent@patience.Stanford.EDU (Mark Kent)
Subject: Re: Numerical Computations With GCC
Message-ID: <1993Feb23.185717.21227@leland.Stanford.EDU>
Sender: news@leland.Stanford.EDU (Mr News)
Organization: Scientific Computing/Computational Mathematics (SCCM), Stanford U.
References: <SMITH.93Feb21175110@minerva.harvard.edu>
Date: Tue, 23 Feb 93 18:57:17 GMT
Lines: 17

Reply to Parent Message