[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
| Title: | Digital Fortran |
| Notice: | Read notes 1.* for important information |
| Moderator: | QUARK::LIONEL |
|
| Created: | Thu Jun 01 1995 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1333 |
| Total number of notes: | 6734 |
1222.0. "-transform_loops bug from Internet" by GEMGRP::GROVE () Mon Mar 17 1997 08:25
Steve Hobbs passed along a FORTRAN (GEM) bug from the Internet.
I compiled it with GEM's current BL33 sources, and it works OK
at both -O4 and -O5. The FORTRAN team should check that it is
fixed in V4.1 or some other released version.
Thanks
/Rich
From: STEVEN::hobbs "Steven Hobbs" 14-MAR-1997 18:58
To: craig@DEC:.steven, pieper@DEC:.steven, noyce@DEC:.steven, liu@DEC:.steven
CC: grove@DEC:.steven
Subj: bug in Digital Fortran 77 V4.0 for UNIX Alpha Systems (-transform_loops) - comp.unix.osf.osf1 #20805
In article <[email protected]>, Stefan Rombouts <[email protected]> writes:
There is a bug in the Digital Fortran 77 compiler V4.0 for UNIX Alpha
Systems: the following program does goes wrong if it is compiled with
the optimalization option -transform_loops (and thus also with -O5):
PROGRAM fortranbug
C bug in the fortran compiler
C compilation with -O5 gives a different result than -O4
REAL X(8,8)
T1=9.
X(6,1)=0.0
T2=1.
T3=2.
DO K=2,5
T4=T2+T3
T5=T2*T3
T6=SQRT(T2**2+T1)
X(1,K)=T1
X(2,K)=T2
X(3,K)=T3
T7=T2*SQRT(T3**2+T1)
X(4,K)=T7
X(5,K)=T3
X(6,K)=X(6,k-1)+T7
T2=T2+1.
T3=T3+1.
ENDDO
WRITE(6,*) 'result:',X(6,5)
DO I=2,5
WRITE(6,'(5F12.6)') (X(J,I),J=1,5)
ENDDO
END
/o\ |+\ Stefan Rombouts
\_ | / Institute for nuclear sciences Tel. +32 (0)9/264 65 41
\ |\ RUG - Universiteit Gent Fax. +32 (0)9/264 66 99
\_/ | \ Proeftuinstraat 86 ; B-9000 Gent ; Belgium
gemeun.zko.dec.com> pf77 -O4 bug.f -o bugo4
gemeun.zko.dec.com> pf77 -O5 bug.f -o bugo5
gemeun.zko.dec.com> bugo4
result: 50.41464
9.000000 1.000000 2.000000 3.605551 2.000000
9.000000 2.000000 3.000000 8.485281 3.000000
9.000000 3.000000 4.000000 15.000000 4.000000
9.000000 4.000000 5.000000 23.323807 5.000000
gemeun.zko.dec.com> bugo5
result: 50.41464
9.000000 1.000000 2.000000 3.605551 2.000000
9.000000 2.000000 3.000000 8.485281 3.000000
9.000000 3.000000 4.000000 15.000000 4.000000
9.000000 4.000000 5.000000 23.323807 5.000000
| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1222.1 | Looks fixed | TLE::EKLUND | Always smiling on the inside! | Mon Mar 17 1997 09:20 | 12 |
| While DFA400 gives different results, DFA401 and DFA410
both give the answers you did. Seems to have been OK for a
while.
Cheers!
Dave Eklund
PS You may continue to discount the value of random postings
on the internet...
Dave E
|
| 1222.2 | | QUARK::LIONEL | Free advice is worth every cent | Mon Mar 17 1997 09:42 | 3 |
| I replied to the newsgroup posting.
Steve
|