Article View: mod.computers.masscomp
Article #33bug in FP on 5600/5700
From: Stan Barber, Moderator
Date: Tue, 05 Aug 1986 23:08
Date: Tue, 05 Aug 1986 23:08
29 lines
1368 bytes
1368 bytes
If you use exponentials in your FORTRAN code, you should avoid
linking in floating point hardware support. The functions will return
garbage if the exponet is any larger than 2.
Here is a test program you can verify with:
program powtest
integer exponet
real base
real result
base = 2.0
do 10 exponet=1,10
result = base**exponet
print *,base,'**',exponet,'=',result
10 continue
end
This problem does not occur when using the pow function in c.
Masscomp has been notified.
Stan
--
Stan uucp:{shell,rice,drillsys}!soma!sob Opinions expressed here
Olan domain:sob@rice.edu or sob@soma.bcm.tmc.edu are ONLY mine &
Barber CIS:71565,623 BBS:(713)660-9252 noone else's.
Technical Details
Message-ID:
<1036@soma.UUCP>Path:
arc1-ca.pugleaf.net!utzoo!decvax!bellcore!petrus!sabre!zeta!epsilon!gamma!ulysses!cbosgd!soma!masscompShow Full Headers
Complete Article Headers
Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: arc1-ca.pugleaf.net!utzoo!decvax!bellcore!petrus!sabre!zeta!epsilon!gamma!ulysses!cbosgd!soma!masscomp From: masscomp@soma.UUCP (Stan Barber, Moderator) Newsgroups: mod.computers.masscomp Subject: bug in FP on 5600/5700 Message-ID: <1036@soma.UUCP> Date: Tue, 5-Aug-86 19:08:51 EDT Article-I.D.: soma.1036 Posted: Tue Aug 5 19:08:51 1986 Date-Received: Tue, 5-Aug-86 23:42:58 EDT Reply-To: sob@soma.UUCP (Stan Barber) Organization: Neurology, Baylor College of Medicine, Houston, Tx Lines: 29 Keywords: FORTRAN 68881 lightening Approved: masscomp@soma.uucp