[MITgcm-devel] AMD/64bit/Portlant Group compiler/scali mpi

[chris hill]

Hi Martin,

 I'm guessing this is a futile question but I'll ask. Is there any
chance we could get a testing account on your cluster. It would be nice
to make sure things run OK on this system.

Thanks,

Chris
On Mon, 2004-12-20 at 11:06, Martin Losch wrote:
> Hey, that was a little quick (o:
> I wanted to give it a proper name first! For example, 
> linux_amd64_pgf77+mpi_ocl (ocl=name of the machine), because this opt 
> file will not work anywhere else, I am pretty sure, because FC and LINK 
> refer to some script which is specific to you maching. Oh well.
> 
> Martin
> 
> On Dec 20, 2004, at 4:53 PM, Martin Losch wrote:
> 
> > Hi,
> > just for the record, today we tried to compile and run the MITgcm on 
> > our new cluster with AMD 64-bit processors (opteron?). We got error 
> > messages like these:
> >> /opt/scali/bin/mpimon -stdin all  ../bin/mitgcmuv  --  ocl01 1 ocl02 
> >> 1 ocl03 1 o
> >> cl04 1 ocl05 1
> >> Dec 20 16:47:58: (mitgcmuv-0 at ocl01)(7023) Error: MPI_Type_hvector is 
> >> deprecated
> >> on architectures with 32-bit integer and 64-bit pointers; Use 
> >> MPI_Type_create_hv
> >> ector instead. MPI_Type_hvector can be enabled by setting 
> >> environmentvariable SC
> >> AMPI_ALLOW_DEPRECATED
> > [... for each processor ...]
> >> --- mpimon --- Aborting run after process-1 terminated abnormally 
> >> Childprocess 7
> >> 603 exited with exitcode 255 ---
> >>
> > seting this environment variable SCAMPI_ALLOW_DEPRECATED got rid off 
> > the error message. Does this tell you anything?
> >
> > Martin
> >
> > Compiler:
> >> pgf77 5.2-4
> >> Copyright 1989-2000, The Portland Group, Inc.  All Rights Reserved.
> >> Copyright 2000-2004, STMicroelectronics, Inc.  All Rights Reserved.
> > SCALI-MPI (whatever that means)
> >
> > The build options file is attached (I'll check that in shortly)
> >
> > <linux_amd64_pgf77+mpi>
> >
> > _______________________________________________
> > MITgcm-devel mailing list
> > MITgcm-devel at mitgcm.org
> > http://dev.mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-devel