MM5 Model on SGI
Complier Options for SGI's New OS and Compiler
The following compiler option may replace what is currently in
configure.user file.
#-----------------------------------------------------------------------------
# 3b.1 Hardware: SGI/CRAY O2, Octane, Origin200, Origin2000, Onyx2
# Indy, Indigo2, (Power)Challenge, (Power)Onyx
# Operating System: IRIX/IRIX64 6.2,6.3,6.4
# Compiler: 7.1, 7.2
# Operating System
#> uname -a
#IRIX64 gs2 6.4 11191505 IP27
# Compiler Version
#> versions -b | grep ftn_dev
#I ftn_dev 01/15/97 Fortran, 7.1
# For R4000 or R4400
# 1) replace -mips4 with -mips3
# 2) link with -lfpe (floating point exception handler) - if you
have one
# 3) runtime environment varable '> setenv TRAP_FPE "UNDERFL=FLUSH_ZERO"'
# Hardware Inventory
#> hinv -t cpu
#CPU: MIPS R4400 Processor Chip Revision: 5.0
# For R5000
# 1) link with -lfpe (floating point exception handler) - if you
have one
# 2) runtime environment varable '>setenv TRAP_FPE "UNDERFL=FLUSH_ZERO"'
# use the second LDOPTIONS if compiling Burk-Thompson PBL scheme
#
# For multiprocessor shared memory systems (i.e. Origin200, Octane
and Onyx2)
# 1) set runtime environment variable '>setenv MP_SET_NUMTHREADS
#'
# where # is the number of processors you wish to use
# 2) set runtime environment variable '>setenv _DSM_ROUND_ROBIN'
# to spread memory across availble nodes.
# In General
# use the second LDOPTIONS if compiling Burk-Thompson PBL
#
# For R10000 and Origin series, use -lfastm only if the recommended
patches
# listed below are installed.
#
#-----------------------------------------------------------------------------
FC = f77
FCFLAGS = -I$(LIBINCLUDE) -n32 -mips4 -mp -O3 -OPT:roundoff=3:IEEE_arithmetic=3
CFLAGS = -n32 -mips4
CPP = /usr/lib/cpp
CPPFLAGS = -I$(LIBINCLUDE) -C -P
LDOPTIONS = -n32 -mips4 -mp
#LDOPTIONS = -n32 -mips4 -mp -Wl,-Xlocal,bt1_,-Xlocal,blk1_,-Xlocal,blk2_
LOCAL_LIBRARIES = -lfastm
#LOCAL_LIBRARIES = -lfastm -lfpe
MAKE = make -i -r
Thanks to Wesley Jones of SGI
|