Issue building pio and processor outline question

[climatemodelling@gmx_com]

Dear Forum Members,

I am trying to build CESM using Intel compilers, openmpi and netcdf 3.6 (build with Intel compiler) on our 64bit linux cluster (nodes with 48 processors each).

However during the compiling process I run into issues with the pio module. I do not intend to use parallel input/output and hence opted for the netcdf 3 version.
Is it still necessary to build the pio module? If not, how do I switch this option off?

It seems the compiler gets the correct options:

Machine=x86_64
OS=Linux
using NETCDF_PATH from environment
configure: WARNING: PNETCDF_PATH not found in environment,                            defaulting to /usr/local/pnetcdf
configure: WARNING: pnetcdf.inc not found in PNETCDF_PATH/include                              disabling pnetcdf support 
configure: WARNING: libpnetcdf.a not found in PNETCDF_PATH/lib                              disabling pnetcdf support 
checking Fortran 90 support for sizeof()... yes

However, I still run into a lot of pointer issues with the pio module, e.g. a short example from the error messages:

pio_msg_callbacks.F90(13): error #6457: This derived type name has not been declared.   [FILE_DESC_T]
  type(file_desc_t), pointer :: file
-------^
pio_msg_callbacks.F90(22): error #6404: This name does not have a type, and must have an explicit type.   [FILE]
  file=> lookupfile(fh)
--^
pio_msg_callbacks.F90(22): error #6795: The target must be of the same type and kind type parameters as the pointer.   [FILE]
  file=> lookupfile(fh)
--^
pio_msg_callbacks.F90(28): error #6285: There is no matching specific subroutine for this generic subroutine call.   [PIO_CLOSEFILE]
     call pio_closefile(file)
----------^
pio_msg_callbacks.F90(33): error #6284: There is no matching specific function for this generic function reference.   [PIO_DEF_VAR]
     ierr = pio_def_var(file, name, type, dimids, vardesc)
------------^
pio_msg_callbacks.F90(39): error #6284: There is no matching specific function for this generic function reference.   [PIO_INQ_VARNDIMS]
     ierr = pio_inq_varndims(file, id, len)
------------^
pio_msg_callbacks.F90(41): error #6284: There is no matching specific function for this generic function reference.   [PIO_INQ_VARNATTS]
     ierr = pio_inq_varnatts(file, id, len)
------------^

Further, I wanted to ask experienced users which processor layouts are reasonable configurations for stand-alone CAM4 runs on either 2 degree or 1 degree horizontal resolution?
I am able to use a maximum of 48 processors (1 node) per run.

Thank you very much in advance!

 

[eaton]

Regarding compilation problems see these related posts:

http://forum.cgd.ucar.edu/showthrea...ilation-error&highlight=derived+type+declared

and

http://forum.cgd.ucar.edu/showthrea...-build-failed&highlight=derived+type+declared

Both 2 deg (1.9x2.5) and 1 deg (0.9x1.25) FV grids will run with 48 tasks. But the 2 deg grid will require using a 2D decomposition which is controlled by the namelist variable npr_yz (see the namelist doc). For example you could set npr_yz=24,2,2,24

 

[climatemodelling@gmx_com]

Thanks for the reply! I tried the suggested source modifications in the other thread as bug fixes, however I still get these errors in many more pio subroutines than the previous poster.

Are these problems fixed when using a newer Ifort version (13.0)?

Thanks!

 

[jedwards]

> Are these problems fixed when using a newer Ifort version (13.0)?

Yes, please try again with the newer intel compiler.