Problems with intel fortran compiler version 8

From: BB (mail_at_mail.com)
Date: 06/28/04 

Date: Mon, 28 Jun 2004 16:08:45 GMT

I have found two problems with the NASA overflow code using Intel Fortran
8 for linux. I'm using the latest version (just updated today) and have
posted my problems to premier.intel.com. Item number 1 concerns reading
a namelist file.

Here is the test program :

      PROGRAM FPREAD
C
C All procs do some initialization, global master reads
C MPIINP namelist input.
C
C Called by: OVERRU.
C
C Calls: none.
C
      IMPLICIT NONE

      CHARACTER*150 OFILENAME
      INTEGER*4 ILEN
      INTEGER*4 JMETHOD, KMETHOD, LMETHOD, JCHUNK
      INTEGER*4 KCHUNK, LCHUNK
      INTEGER*4 FINEPART(5,10)
      INTEGER*4 SHARZONE(5,10)

C
C
      NAMELIST/MPIINP/ JMETHOD,KMETHOD,LMETHOD,
     & JCHUNK ,KCHUNK ,LCHUNK,
     & FINEPART,SHARZONE
C
      INTEGER N,I

      OFILENAME = 'over.1.14procs.inp'
      ILEN = INDEX(OFILENAME,' ')-1
      OPEN(5,FILE=OFILENAME(1:ILEN),STATUS='OLD')

C
C Defaults for method and chunk size.
C
      JMETHOD = 4
      KMETHOD = 4
      LMETHOD = 4
      JCHUNK = 10
      KCHUNK = 10
      LCHUNK = 10
C
C
      READ(*,MPIINP,END=120,ERR=110)
C
C Write a summary of the 'MPI' input parameters.
C
      WRITE(*,1) JMETHOD,KMETHOD,LMETHOD,JCHUNK,KCHUNK,LCHUNK
    1 FORMAT(/' MPI PARAMETERS ($MPIINP)'/
     & 4X,'J DIRECTION SOLVER (JMETHOD) = ',I5/
     & 4X,'K DIRECTION SOLVER (KMETHOD) = ',I5/
     & 4X,'L DIRECTION SOLVER (LMETHOD) = ',I5/
     & 4X,'J DIRECTION CHUNK SIZE (JCHUNK ) = ',I5/
     & 4X,'K DIRECTION CHUNK SIZE (KCHUNK ) = ',I5/
     & 4X,'L DIRECTION CHUNK SIZE (LCHUNK ) = ',I5)
      WRITE(6,*) FINEPART
      WRITE(6,*) SHARZONE
      GOTO 130
C
 110 CONTINUE
      WRITE(*,"(' ** ERROR ** ERROR READING NAMELIST $MPIINP')")
      GOTO 130
 120 WRITE(*,"(' ** ERROR ** EOF $MPIINP')")
 130 CONTINUE
      STOP
      END

and here is my namelist input :

 $MPIINP
     FINEPART= 2, 3, 1, 1,
                5, 3, 1, 1,
                7, 1, 2, 1,
                4, 2, 1, 1,
                3, 2, 1, 1,
     SHARZONE= 1, 1,
                2, 6,
  $END
                
If I change the * (preconnected unit) to a 5 it works. If I run it the
way it is, it hangs. It works correctly under Intel 7 and Portland
group.

I'm also getting a segfault using an automatic array. I'm running on a
workstation with 2GB memory, so it shouldn't be a memory size. Linux
version is Redhat 7.3. This program also works under Intel 7 and
Portland Group.

Here is the auto program :

      Program auto
      n = 5000
      m = 500
      print *,n*m
      call sub1(n,m)
      stop
      end
      Subroutine sub1(n,m)
      Dimension a(n,m)
      call sub2(a,n,m)
      return
      end
      Subroutine sub2(a,n,m)
      dimension a(n,m)
      do i=1,n
        do j=1,m
           a(i,j) = 0.0
        enddo
      enddo
      return
      end

Maybe I just don't understand how automatic arrays work.

If anyone can shed any information on the problems I'm having, I would
appreciate it.

BB
The Boeing Company

Relevant Pages

  • Re: [SLE] Gnome disappointment
    ... the problem is that Intel can't make AMD ... and SPARC was going to blow away the mainframe. ... NT/W2K have replaced the Sun kit, perhaps the IBM HMC based on Linux may ...
    (SuSE)
  • Re: CRN is Outta Their Tree--DARL MCBRIDE?
    ... >Linux box, but I have done a moderate amount of reading) nor do I ... Apple's version of Unix. ... SCO has been about the most backward compatible company to ... that virtually all of the iNTEL based *n*x systems disappeared. ...
    (comp.unix.sco.misc)
  • Re: New system reqd, AMD or Intel ?
    ... on the INTEL you could afford an AMD 64 bit system which with the 64bit ... A faster fsb does not always mean a faster system nor does a faster chip ... ram to give you the dual channel memory (not much of a boost but is ... noticeable in linux which utilises the memory better than other O/S`s ...
    (alt.os.linux.suse)
  • Re: Getting people to say nice things about Microsoft (at least Intel is trying)
    ... then 'apt install synaptic' and that worked - poked around there for awhile ... Freespire will default to using traditional Linux user accounts and running ... these attempts to 'dumb down' KDE. ... as Intel's new 45 nanometer transistors, ...
    (Fedora)
  • Re: [SLE] And another 10.1 showstopper
    ... Had I been using Linux properly at the time I ... That also means they want the hardware to ... Intel who forces many to sell only their hardware ... -- but even it is missing lots of support from any of its peripherals. ...
    (SuSE)