[MITgcm-support] topog.bin read error in exp2

Jeff Haferman jeff at haferman.com
Fri Jul 24 15:29:18 EDT 2009


I got it to work with the serial version, mpi version is still bombing
out, I'll set the compiler to mpif90 and see if that helps...


Zhi-Ping Mei wrote:
> 
> HI,
> 
> I encountered similar problem when I compiled with pgi f77. When I specified
> pgif90 as fortran compiler, compilation was fine.
> 
> Zhi-Ping
> 
> 
> 
> --- On Thu, 7/23/09, Jeff Haferman <jeff.haferman at gmail.com> wrote:
> 
> From: Jeff Haferman <jeff.haferman at gmail.com>
> Subject: [MITgcm-support] topog.bin read error in exp2
> To: mitgcm-support at mitgcm.org
> Received: Thursday, July 23, 2009, 7:04 AM
> 
> Hi -
> I'm trying to get the verification/exp2 example running. I've built and 
> run this successfully on a number of platforms over the last few years, so 
> I'm not quite sure what the problem is, but on my current machine (running 
> Centos 5.2 with Linux kernel 2.6.18-92 and Intel Xeons, i.e. an x86-64), I 
> am building fine with both PGI and Sun Studio compilers. But upon execution
> I'm getting:
>
> PGFIO-F-253/unformatted read/unit=9/attempt to read non-existent record (x
> direct access).
> File name  topog.bin unformatted, direct access record
> 
> In source file mdsio_read_field.f, at line number 2028
>
> This is with the PGI compiler; I get a similar error with the Sun Studio 
> compile, i.e. the code bombs out when it tries to read the topog.bin file.
>  I know this is a file filled with 64-bit binary values, and there are endian
>  considerations to be made.
>
> 
> My build options are pretty simple, the important stuff is
> DEFINES= '-DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4'
>
>
>
> 
>  I'm wondering about -byteswapio and -r8 options... these have always worked
>  fine on other Linux x86-64 machines. I've copied the topog.bin file over
>  from a machine where exp2 is working, and it is identical to the topog.bin
>  on the machine where exp2 is not working.
>
>  Any ideas?
>




More information about the MITgcm-support mailing list