[MITgcm-support] linux_ia32_ifort
Martin Losch
mlosch at awi-bremerhaven.de
Tue Sep 28 02:05:20 EDT 2004
Dimitris,
I use linux_ia32_ifort as it is on a redhat 9 something system. After
having a look at the man-pages I think that
-DWORDLENGTH=1
is equivalent to the combination of
-DWORDLENGTH=4
and
-assume byterecl
because according to man-pages:
> · -assume byterecl
>
> Specifies (for unformatted data files) that the units
> for the
> OPEN statement RECL specifier (record length)
> value are in
> bytes, not longwords (four-byte units). For formatted
> files,
> the RECL unit is always in bytes. The default is
> -assume
> nobyterecl.
BTW, I think the -D_BYTESWAPIO could be replaced by a compiler option
-convert big_endian
I haven't tried using this with the MITgcm yet, but with other
applications. I don't know if that's any faster than the -D_BYTESWAPIO.
Martin
On Sep 27, 2004, at 8:26 PM, Dimitris Menemenlis wrote:
> Ed, the following changes appear to be needed in linux_ia32_ifort
> for ifort 8.0. Anyone else using ifort 8.0 succesfully with original
> linux_ia32_ifort ?
>
> D.
>
> #!/bin/bash
> #
> # Ed Hill
> #
> # tested on faulks, 20030818
> #
> # Build options for the intel 8.0 fortran compiler
> # the compiler is now called ifort (not ifc) and it appears to
> # need -DWORDLENGTH=1 and not -DWORDLENGTH=4
> # (which is what ifc 6.0) used to have.
>
>
> CPP='/lib/cpp -traditional -P'
>
> # Note that the -mp switch is for ieee "maintain precision" and is
> # roughly equivalent to -ieee
> if test "x$IEEE" = x ; then
> FFLAGS='-132 -r8 -i4 -w95 -W0 -WB -assume byterecl'
> else
> FFLAGS='-mp -132 -r8 -i4 -w95 -W0 -WB -assume byterecl'
> fi
>
> FOPTIM='-O3 -align'
> #P3 FOPTIM = ( $FOPTIM '-tpp6 -xWKM' )
> #P4 FOPTIM = ( $FOPTIM '-tpp7 -xWKM' )
>
> FC=ifort
> DEFINES='-D_BYTESWAPIO -DWORDLENGTH=4'
>
>
> --
> Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
> Jet Propulsion Lab, California Institute of Technology
> MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
> tel: 818-354-1656; fax: 818-393-6720
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
More information about the MITgcm-support
mailing list