again edited the makefile, no there is more flexibility in defining the libraries (FFTW and ACML)
added the source of fftw-3.3.tar.gz to fulfill the GPL set_python_env.py was forgotten during last commit (for testing)
This commit is contained in:
Martin Diehl
parent
48aa5eb163
commit
0f34d14bee
Binary file not shown.
104
code/makefile
104
code/makefile
|
|
@ -1,28 +1,32 @@
|
|||
########################################################################################
|
||||
# Makefile to compile the Material subroutine for BVP solution using spectral method
|
||||
########################################################################################
|
||||
# Be sure to remove all files compiled with different options by using "make clean"
|
||||
#
|
||||
# use switch on make to determine PRECISION, e.g make PRECISION=single
|
||||
# default is PRECISION=double
|
||||
# be sure to remove all librarys with different PRECISION (make clean)
|
||||
#
|
||||
# Uses openmp to parallelise the material subroutines (set number of cores with "export DAMASK_NUM_THREADS=n" to n)
|
||||
# Uses linux threads to parallelise fftw3 (should also be possible with openmp)
|
||||
# Uses OpenMP to parallelise the material subroutines (set number of cores with "export DAMASK_NUM_THREADS=n" to n)
|
||||
# Uses linux threads to parallelise fftw3
|
||||
# Besides of the f90 files written at MPIE, the two library files of fftw3 "libfftw3_threads.a" "libfftw3.a" are also needed
|
||||
# Install fftw3 (v3.3 is tested) with "./configure --enable-threads --enable-sse2" and "make"; "make install" is not needed
|
||||
# as long as the two library files "libfftw3_threads.a" "libfftw3.a" are copied to the code/include directory.
|
||||
|
||||
# Need the AMD Core Math Library to be installed (v 4.4 is tested)
|
||||
########################################################################################
|
||||
# OPTIONS = standard (alternative): meaning
|
||||
#-------------------------------------------------------------
|
||||
# PRECISION = double (single): floating point precision
|
||||
# F90 = ifort (gfortran): compiler, choose Intel or GNU
|
||||
# PORTABLE = TRUE (FALSE): decision, if executable is optimized for the machine on which it was built. Until now only for ifort
|
||||
# OPTIMIZATION = DEFENSIVE (OFF,AGGRESSIVE,ULTRA): Optimization mode, O0, O2, O3
|
||||
# PORTABLE = TRUE (FALSE): decision, if executable is optimized for the machine on which it was built.
|
||||
# OPTIMIZATION = DEFENSIVE (OFF,AGGRESSIVE,ULTRA): Optimization mode: O2, O0, O3 + further options for most files, 03 + further options for all files
|
||||
# OPENMP = TRUE (FALSE): OpenMP multiprocessor support
|
||||
# PREFIX: specifie an arbitrary prefix
|
||||
# SUFFIX: specife an arbitrary suffix, e.g
|
||||
# ACMLPATH =/opt/acml4.4.0/ifort64/lib (...) Path to ACML Library, choose according to your system
|
||||
# ACMLPATH =/opt/acml4.4.0/ifort64_mp/lib (...) Path to ACML Library with multicore support, choose according to your system
|
||||
# FFTWOPTIONS =include/libfftw3.a include/libfftw3_threads.a -lpthread (...) Path to FFTW library files with Linux threads (multicore) support
|
||||
# FFTWOPTIONS =include/libfftw3.a (...) Path to FFTW library files without Linux threads (multicore) support
|
||||
# FFTWOPTIONS is different for single and double precision. Choose the options to use OpenMP instead of pthreads support or change the directory
|
||||
# PREFIX: specify an arbitrary prefix
|
||||
# SUFFIX: specify an arbitrary suffix
|
||||
# COMPILERNAME = overwrite name of Compiler, e.g. using mpich-g90 instead of ifort
|
||||
|
||||
# Here are some usefull debugging switches. Switch on by uncommenting last line:
|
||||
#--------------------------------------------------------------------------------
|
||||
########################################################################################
|
||||
# Here are some usefull debugging switches. Switch on by uncommenting the #SUFFIX line at the end of this section:
|
||||
# information on http://software.intel.com/en-us/articles/determining-root-cause-of-sigsegv-or-sigbus-errors/
|
||||
# check if an array index is too small (<1) or too large!
|
||||
DEBUG1 =-check bounds -g
|
||||
|
|
@ -32,56 +36,77 @@ DEBUG2 =-check arg_temp_created
|
|||
DEBUG3 =-fp-stack-check -g -traceback -gen-interfaces -warn interfaces
|
||||
#should not be done for OpenMP, but set "ulimit -s unlimited" on shell. Problably it helps also to unlimit other limits
|
||||
DEBUG4 =-heap-arrays
|
||||
|
||||
#checks for standard
|
||||
DEBUG5 =stand std03/std95
|
||||
|
||||
#SUFFIX =$(DEBUG1) $(DEBUG2) $(DEBUG3)
|
||||
########################################################################################
|
||||
|
||||
#BLAS for OPENMP=OFF
|
||||
BLAS_ifort =-L /opt/acml4.4.0/ifort64/lib -lacml
|
||||
BLAS_gfortran =
|
||||
|
||||
ifeq ($(F90), )
|
||||
ifndef F90
|
||||
F90 =ifort
|
||||
endif
|
||||
|
||||
|
||||
ifeq ($(OPTIMIZATION), )
|
||||
ifndef OPTIMIZATION
|
||||
OPTIMIZATION =DEFENSIVE
|
||||
endif
|
||||
MAXOPTI =$(OPTIMIZATION)
|
||||
|
||||
ifeq ($(OPTIMIZATION),ULTRA)
|
||||
MAXOPTI=AGGRESSIVE
|
||||
endif
|
||||
|
||||
ifeq ($(OPTIMIZATION),AGGRESSIVE)
|
||||
MAXOPTI=DEFENSIVE
|
||||
endif
|
||||
|
||||
ifeq ($(OPTIMIZATION),ULTRA)
|
||||
MAXOPTI=AGGRESSIVE
|
||||
OPTIMIZATION=AGGRESSIVE
|
||||
endif
|
||||
|
||||
ifeq ($(PORTABLE),FALSE)
|
||||
PORTABLE_SWITCH =-msse3
|
||||
endif
|
||||
|
||||
ifneq ($(OPENMP), OFF)
|
||||
ifndef OPENMP
|
||||
OPENMP=ON
|
||||
endif
|
||||
|
||||
ifeq ($(OPENMP),ON)
|
||||
OPENMP_FLAG_ifort =-openmp -openmp-report0 -parallel
|
||||
OPENMP_FLAG_gfortran =-fopenmp
|
||||
BLAS_ifort =-L /opt/acml4.4.0/ifort64_mp/lib -lacml_mp
|
||||
BLAS_gfortran =
|
||||
OPENMP =ON
|
||||
ifndef ACMLPATH
|
||||
ACMLPATH =/opt/acml4.4.0/ifort64_mp/lib
|
||||
endif
|
||||
ifndef FFTWOPTIONS
|
||||
ifeq ($(PRECISION),single)
|
||||
FFTWOPTIONS =include/libfftw3f_threads.a include/libfftw3f.a -lpthread
|
||||
else
|
||||
FFTWOPTIONS =include/libfftw3_threads.a include/libfftw3.a -lpthread
|
||||
endif
|
||||
endif
|
||||
BLAS_ifort =-L $(ACMLPATH) -lacml_mp
|
||||
BLAS_gfortran =
|
||||
else
|
||||
ifndef ACMLPATH
|
||||
ACMLPATH =/opt/acml4.4.0/ifort64/lib
|
||||
endif
|
||||
ifndef FFTWOPTIONS
|
||||
ifeq ($(PRECISION),single)
|
||||
FFTWOPTIONS =include/libfftw3f.a
|
||||
else
|
||||
FFTWOPTIONS =include/libfftw3.a
|
||||
endif
|
||||
endif
|
||||
BLAS_ifort =-L $(ACMLPATH) -lacml
|
||||
BLAS_gfortran =
|
||||
endif
|
||||
|
||||
|
||||
OPTIMIZATION_OFF_ifort =-O0
|
||||
OPTIMIZATION_OFF_gfortran =-O0
|
||||
OPTIMIZATION_DEFENSIVE_ifort =-O2
|
||||
OPTIMIZATION_DEFENSIVE_gfortran =-O2
|
||||
OPTIMIZATION_AGGRESSIVE_ifort =-O3 $(PORTABLE_SWITCH) -ip -static -fp-model fast=2 -no-prec-div
|
||||
OPTIMIZATION_ULTRA_ifort =-O3 $(PORTABLE_SWITCH) -ip -static -fp-model fast=2 -no-prec-div
|
||||
OPTIMIZATION_AGGRESSIVE_gfortran =-O3 -march=opteron -ffast-math -funroll-loops -ftree-vectorize -ftree-loop-linear $(PORTABLE_SWITCH)
|
||||
OPTIMIZATION_AGGRESSIVE_gfortran =-O3 $(PORTABLE_SWITCH) -march=opteron -ffast-math -funroll-loops -ftree-vectorize -ftree-loop-linear
|
||||
|
||||
COMPILE_OPTIONS_ifort =-fpp -diag-disable 8291,8290
|
||||
COMPILE_OPTIONS_gfortran =-xf95-cpp-input -ffree-line-length-none
|
||||
COMPILE_OPTIONS_gfortran =-xf95-cpp-input -ffree-line-length-none -fno-range-check
|
||||
|
||||
COMPILE =${OPENMP_FLAG_${F90}} ${COMPILE_OPTIONS_${F90}} ${OPTIMIZATION_${OPTIMIZATION}_${F90}} -c
|
||||
COMPILE_MAXOPTI =${OPENMP_FLAG_${F90}} ${COMPILE_OPTIONS_${F90}} ${OPTIMIZATION_${MAXOPTI}_${F90}} -c
|
||||
|
|
@ -92,14 +117,14 @@ endif
|
|||
|
||||
ifeq ($(PRECISION),single)
|
||||
DAMASK_spectral_single.exe: DAMASK_spectral_single.o CPFEM.a
|
||||
$(PREFIX) $(COMPILERNAME) ${OPENMP_FLAG_${F90}} -o DAMASK_spectral_single.exe DAMASK_spectral_single.o CPFEM.a include/libfftw3f_threads.a include/libfftw3f.a\
|
||||
constitutive.a advanced.a basics.a -lpthread ${BLAS_${F90}}
|
||||
$(PREFIX) $(COMPILERNAME) ${OPENMP_FLAG_${F90}} -o DAMASK_spectral_single.exe DAMASK_spectral_single.o CPFEM.a $(FFTWOPTIONS)\
|
||||
constitutive.a advanced.a basics.a ${BLAS_${F90}}
|
||||
DAMASK_spectral_single.o: DAMASK_spectral_single.f90 CPFEM.o
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE_MAXOPTI) DAMASK_spectral_single.f90 $(SUFFIX)
|
||||
else
|
||||
DAMASK_spectral.exe: DAMASK_spectral.o CPFEM.a
|
||||
$(PREFIX) $(COMPILERNAME) ${OPENMP_FLAG_${F90}} -o DAMASK_spectral.exe DAMASK_spectral.o CPFEM.a include/libfftw3_threads.a include/libfftw3.a\
|
||||
constitutive.a advanced.a basics.a -lpthread ${BLAS_${F90}}
|
||||
$(PREFIX) $(COMPILERNAME) ${OPENMP_FLAG_${F90}} -o DAMASK_spectral.exe DAMASK_spectral.o CPFEM.a $(FFTWOPTIONS)\
|
||||
constitutive.a advanced.a basics.a ${BLAS_${F90}}
|
||||
DAMASK_spectral.o: DAMASK_spectral.f90 CPFEM.o
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE_MAXOPTI) DAMASK_spectral.f90 $(SUFFIX)
|
||||
endif
|
||||
|
|
@ -168,7 +193,6 @@ debug.o: debug.f90 numerics.o
|
|||
$(PREFIX) $(COMPILERNAME) $(COMPILE) debug.f90 $(SUFFIX)
|
||||
math.o: math.f90 numerics.o
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE) math.f90 $(SUFFIX)
|
||||
|
||||
numerics.o: numerics.f90 IO.o
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE) numerics.f90 $(SUFFIX)
|
||||
IO.o: IO.f90 DAMASK_spectral_interface.o
|
||||
|
|
@ -183,8 +207,6 @@ else
|
|||
DAMASK_spectral_interface.o: DAMASK_spectral_interface.f90 prec.o
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE) DAMASK_spectral_interface.f90 $(SUFFIX)
|
||||
prec.o: prec.f90
|
||||
@echo $(OPTIMIZATION)
|
||||
@echo $(MAXOPTI)
|
||||
$(PREFIX) $(COMPILERNAME) $(COMPILE) prec.f90 $(SUFFIX)
|
||||
endif
|
||||
|
||||
|
|
|
|||
|
|
@ -0,0 +1,21 @@
|
|||
#!/usr/bin/env python
|
||||
import os,site,sys
|
||||
|
||||
# More elegant was to determine DAMASK install dir?
|
||||
#p=os.path.abspath(__file__)
|
||||
#l=p.split('/')
|
||||
#p=p[:-len(l[-1])]
|
||||
#print p
|
||||
#site.addsitedir(p) # adds the paths in the *.pth file to the python search path
|
||||
|
||||
basepath=os.path.expanduser('~')+'/DAMASK/'
|
||||
|
||||
|
||||
sys.path.insert(0,basepath+'processing')
|
||||
sys.path.insert(0,basepath+'processing/post')
|
||||
sys.path.insert(0,basepath+'processing/pre')
|
||||
sys.path.insert(0,basepath+'processing/setup')
|
||||
#print sys.path
|
||||
|
||||
|
||||
|
||||
Loading…
Reference in New Issue