问题
I am trying to build NAS benchmarks using Intel MPI and below is the makefile that I am using.
#---------------------------------------------------------------------------
#
# SITE- AND/OR PLATFORM-SPECIFIC DEFINITIONS.
#
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
# Items in this file will need to be changed for each platform.
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
# Parallel Fortran:
#
# For CG, EP, FT, MG, LU, SP and BT, which are in Fortran, the following must
# be defined:
#
# MPIF77 - Fortran compiler
# FFLAGS - Fortran compilation arguments
# FMPI_INC - any -I arguments required for compiling MPI/Fortran
# FLINK - Fortran linker
# FLINKFLAGS - Fortran linker arguments
# FMPI_LIB - any -L and -l arguments required for linking MPI/Fortran
#
# compilations are done with $(MPIF77) $(FMPI_INC) $(FFLAGS) or
# $(MPIF77) $(FFLAGS)
# linking is done with $(FLINK) $(FMPI_LIB) $(FLINKFLAGS)
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
# This is the fortran compiler used for MPI programs
#---------------------------------------------------------------------------
MPIF77 = gfortran
# This links MPI fortran programs; usually the same as ${MPIF77}
FLINK = $(MPIF77)
#---------------------------------------------------------------------------
# These macros are passed to the linker to help link with MPI correctly
#---------------------------------------------------------------------------
FMPI_LIB = -L/share/apps/intel/impi/5.0.2.044/intel64/lib -lmpi
#---------------------------------------------------------------------------
# These macros are passed to the compiler to help find 'mpif.h'
#---------------------------------------------------------------------------
FMPI_INC = -I/share/apps/intel/impi/5.0.2.044/intel64/include
#---------------------------------------------------------------------------
# Global *compile time* flags for Fortran programs
#---------------------------------------------------------------------------
FFLAGS = -O
#---------------------------------------------------------------------------
# Global *link time* flags. Flags for increasing maximum executable
# size usually go here.
#---------------------------------------------------------------------------
FLINKFLAGS = -O
#---------------------------------------------------------------------------
# Parallel C:
#
# For IS, which is in C, the following must be defined:
#
# MPICC - C compiler
# CFLAGS - C compilation arguments
# CMPI_INC - any -I arguments required for compiling MPI/C
# CLINK - C linker
# CLINKFLAGS - C linker flags
# CMPI_LIB - any -L and -l arguments required for linking MPI/C
#
# compilations are done with $(MPICC) $(CMPI_INC) $(CFLAGS) or
# $(MPICC) $(CFLAGS)
# linking is done with $(CLINK) $(CMPI_LIB) $(CLINKFLAGS)
#---------------------------------------------------------------------------
#---------------------------------------------------------------------------
# This is the C compiler used for MPI programs
#---------------------------------------------------------------------------
MPICC = cc
# This links MPI C programs; usually the same as ${MPICC}
CLINK = $(MPICC)
#---------------------------------------------------------------------------
# These macros are passed to the linker to help link with MPI correctly
#---------------------------------------------------------------------------
CMPI_LIB = -L/usr/local/lib -lmpi
#---------------------------------------------------------------------------
# These macros are passed to the compiler to help find 'mpi.h'
#---------------------------------------------------------------------------
CMPI_INC = -I/usr/local/include
#---------------------------------------------------------------------------
# Global *compile time* flags for C programs
#---------------------------------------------------------------------------
CFLAGS = -O
#---------------------------------------------------------------------------
# Global *link time* flags. Flags for increasing maximum executable
# size usually go here.
#---------------------------------------------------------------------------
CLINKFLAGS = -O
#---------------------------------------------------------------------------
# MPI dummy library:
#
# Uncomment if you want to use the MPI dummy library supplied by NAS instead
# of the true message-passing library. The include file redefines several of
# the above macros. It also invokes make in subdirectory MPI_dummy. Make
# sure that no spaces or tabs precede include.
#---------------------------------------------------------------------------
# include ../config/make.dummy
#---------------------------------------------------------------------------
# Utilities C:
#
# This is the C compiler used to compile C utilities. Flags required by
# this compiler go here also; typically there are few flags required; hence
# there are no separate macros provided for such flags.
#---------------------------------------------------------------------------
CC = cc -g
#---------------------------------------------------------------------------
# Destination of executables, relative to subdirs of the main directory. .
#---------------------------------------------------------------------------
BINDIR = ../bin
#---------------------------------------------------------------------------
# Some machines (e.g. Crays) have 128-bit DOUBLE PRECISION numbers, which
# is twice the precision required for the NPB suite. A compiler flag
# (e.g. -dp) can usually be used to change DOUBLE PRECISION variables to
# 64 bits, but the MPI library may continue to send 128 bits. Short of
# recompiling MPI, the solution is to use MPI_REAL to send these 64-bit
# numbers, and MPI_COMPLEX to send their complex counterparts. Uncomment
# the following line to enable this substitution.
#
# NOTE: IF THE I/O BENCHMARK IS BEING BUILT, WE USE CONVERTFLAG TO
# SPECIFIY THE FORTRAN RECORD LENGTH UNIT. IT IS A SYSTEM-SPECIFIC
# VALUE (USUALLY 1 OR 4). UNCOMMENT THE SECOND LINE AND SUBSTITUTE
# THE CORRECT VALUE FOR "length".
# IF BOTH 128-BIT DOUBLE PRECISION NUMBERS AND I/O ARE TO BE ENABLED,
# UNCOMMENT THE THIRD LINE AND SUBSTITUTE THE CORRECT VALUE FOR
# "length"
#---------------------------------------------------------------------------
# CONVERTFLAG = -DCONVERTDOUBLE
# CONVERTFLAG = -DFORTRAN_REC_SIZE=length
# CONVERTFLAG = -DCONVERTDOUBLE -DFORTRAN_REC_SIZE=length
#---------------------------------------------------------------------------
# The variable RAND controls which random number generator
# is used. It is described in detail in README.install.
# Use "randi8" unless there is a reason to use another one.
# Other allowed values are "randi8_safe", "randdp" and "randdpvec"
#---------------------------------------------------------------------------
RAND = randi8
# The following is highly reliable but may be slow:
# RAND = randdp
When I try to build let's say EP benchmark make NPROCS=8 EP CLASS=C, I get an error like
=========================================
= NAS Parallel Benchmarks 3.3 =
= MPI/F77/C =
=========================================
cd EP; make NPROCS=8 CLASS=C
make[1]: Entering directory `/home/vaibhavs/NPB3.3/NPB3.3-MPI/EP'
make[2]: Entering directory `/home/vaibhavs/NPB3.3/NPB3.3-MPI/sys'
cc -g -o setparams setparams.c
make[2]: Leaving directory `/home/vaibhavs/NPB3.3/NPB3.3-MPI/sys'
../sys/setparams ep 8 C
make.def modified. Rebuilding npbparams.h just in case
rm -f npbparams.h
../sys/setparams ep 8 C
gfortran -c -I/share/apps/intel/impi/5.0.2.044/intel64/include -O ep.f
gfortran -O -o ../bin/ep.C.8 ep.o ../common/print_results.o ../common/randi8.o ../common/timers.o -L/share/app s/intel/impi/5.0.2.044/intel64/lib -lmpi
ep.o: In function `MAIN__':
ep.f:(.text+0x31): undefined reference to `mpi_init_'
ep.f:(.text+0x4b): undefined reference to `mpi_comm_rank_'
ep.f:(.text+0x65): undefined reference to `mpi_comm_size_'
ep.f:(.text+0x347): undefined reference to `mpi_abort_'
ep.f:(.text+0x3df): undefined reference to `mpi_barrier_'
ep.f:(.text+0x803): undefined reference to `mpi_allreduce_'
ep.f:(.text+0x842): undefined reference to `mpi_allreduce_'
ep.f:(.text+0x879): undefined reference to `mpi_allreduce_'
ep.f:(.text+0x935): undefined reference to `mpi_allreduce_'
ep.f:(.text+0xc70): undefined reference to `mpi_finalize_'
../common/timers.o: In function `timer_stop_':
timers.f:(.text+0x27): undefined reference to `mpi_wtime_'
../common/timers.o: In function `timer_start_':
timers.f:(.text+0x57): undefined reference to `mpi_wtime_'
collect2: ld returned 1 exit status
make[1]: *** [../bin/ep.C.8] Error 1
make[1]: Leaving directory `/home/vaibhavs/NPB3.3/NPB3.3-MPI/EP'
make: *** [ep] Error 2
Can anybody please help me in resolving this issue?
Thanks
回答1:
If your MPI library is Intel MPI, and was built by the intel fortran compiler, then you're going to be sad when you try to build your MPI application with gfortran.
For a long time there was no standard way to call fortran symbols from C. Some compilers put one, two, or zero underscores on the symbol.
You should build NAS with ifort
You might be able to make things work with the -fno-underscoring option , but heed well this message from the gfortran man page:
Note that just because the names match does not mean that the interface implemented by GNU Fortran for an external name matches the interface implemented by some other language for that same name. That is, getting code produced by GNU Fortran to link to code produced by some other compiler using this or any other method can be only a small part of the overall solution---getting the code generated by both compilers to agree on issues other than naming can require significant effort, and, unlike naming disagreements, linkers normally cannot detect disagreements in these other areas.
回答2:
It seems that you the mpi library doesn't start. You can try other mpi library like openmpi.
来源：https://stackoverflow.com/questions/30604776/error-while-building-nas-benchmarks