Subject: ccm3.6.6 on DEC Alpha/Digital UNIX
From: Chien Wang (wangc@MIT.EDU)
Date: Fri Jan 14 2000 - 13:06:59 MST
Hi everyone,
I'm a new member of this community I guess. I've just downloaded the
ccm3.6.6 along with data files and ported it to my DEC Alpha ev56 station
(Digital UNIX 4.0D, Digital FORTRAN 5.2). The process seems quite smooth
for the non SPMD setting. I simply changed a few lines in the build script
and Makefile as well as a few source files and primarily follow the SGI
setup in most places:
1) use f90 instead of f77 (malloc dosn't work well for f77);
2) add -cpp to flags of compiler and linker and change OS identifier to
ifeq ($(UNAMES), OSF1) ;
3) delete duplicate declaration of "integer info" in lsm/restwart.F and
delete
"external srfsice
external srfcoe" in som/somoce.F
Oh, I installed the gmake and I think doing a "unlimit" (or whatever can
increase the stack size and datasize) is necessary for running most models
(otherwise getmem might give runtime problem). after all these the model
runs and produces results though I haven't gotten time to check them
carefully. I have two questions and would very much appreciate any advice
from you:
1. When using SPMD mode with Digital MPI, define REALTYPE as
MPI_DOUBLEPRECISION while using -r8 flag seems generate compiling errors.
after changing it to MPI_REAL (MPI_REAL8 won't work) (Digital FORTRAN
forces real to real*8 when using -r8 flag) the compile went through well,
however, a float exception happened in runtime during initialization of the
model, though I'm not sure at this moment if it's caused by the stacksize
problem (I used a single cpu machine to do dmpirun -npx ...). I haven't had
time to play with it and wondering if someone has experience in running
SPMD on DEC alpha.
2. I was trying to use SOM instead of DOM but then was told by the model
that I need initial data set for it. After browsing the NCOM page (new one)
and downloaded the code, I still cannot figure out where I can find the
data. I understand from the document that the som is not supported by the
ccm3 but would appreciate if someone can enlighten me on this issue.
Best regards!
===========================================
Chien Wang
Massachusetts Institute of Technology, Room E40-269
Cambridge, MA 02139
Phone: (617)253-5432
Fax: (617)253-9845
Email: wangc@mit.edu
===========================================
This archive was generated by hypermail 2b27 : Thu Jan 04 2001 - 10:01:23 MST