Code Saturne-2.0.0 fails to compile...

Hi Forum,

I have tried to build CS-2.0.0 in my Debian Lenny 5.0.0 amd64.

First, i built bft-1.1.1, then fvm-0.14.0, then mei-1.0.0 and at last ecs-2.0.0

All they build cleanly, the only hack i have to perform is --disable-python-bindings in configuring mei, so that it does not complain about some missing header... so far so good.

Compiling ncs-2.0.0 :

alex@iskandhar:~/Desktop/CDSat/ncs-2.0.0-beta2$ ./configure --prefix=/home/alex/Desktop/Code_Saturne --with-bft=/home/alex/Desktop/Code_Saturne --with-bft-exec=/home/alex/Desktop/Code_Saturne/bin --with-bft-include=/home/alex/Desktop/Code_Saturne/include --with-bft-lib=/home/alex/Desktop/Code_Saturne/lib --with-mei-include=/home/alex/Desktop/Code_Saturne/include --with-mei-lib=/home/alex/Desktop/Code_Saturne/lib --with-mei=/home/alex/Desktop/Code_Saturne --with-fvm=/home/alex/Desktop/Code_Saturne --with-fvm-exec=/home/alex/Desktop/Code_Saturne/bin --with-fvm-include=/home/alex/Desktop/Code_Saturne/include --with-fvm-lib=/home/alex/Desktop/Code_Saturne/lib

Yelds this result... :

Configuration options:
use debugging code: false
MPI (Message Passing Interface) support: no
OpenMP support: no
BLAS (Basic Linear Algebra Subprograms) support: yes
Libxml2 (XML Reader) support: yes
MEI (Mathematical Expressions Interpreter) support: yes
SYRTHES 3 coupling support: yes
IP socket support (for SYRTHES 3 or CFD_Proxy): yes
Dynamic loader support (for YACS): yes

The package has been configured. Type:
make
make install

To generate and install the Code_Saturne kernel

alex@iskandhar:~/Desktop/CDSat/ncs-2.0.0-beta2$


so why 'make' produces this error...? :


Making all in apps
make[3]: Entering directory `/home/alex/Desktop/CDSat/ncs-2.0.0-beta2/src/apps'
/bin/sh ../../libtool --tag=CC --mode=link gcc -funroll-loops -O2 -Wuninitialized -std=c99 -funsigned-char -pedantic -W -Wall -Wshadow -Wpointer-arith -Wcast-qual -Wcast-align -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations -Wnested-externs -Wunused -Wfloat-equal -rdynamic -O -o cs_solver libsaturne.la -L/home/alex/Desktop/Code_Saturne/lib -lmei -L/home/alex/Desktop/Code_Saturne/lib -lfvm -lmedC -lhdf5 -L/home/alex/Desktop/Code_Saturne/lib -lbft -lz -lxml2 -lcblas -latlas -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2 -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2/../../../../lib -L/lib/../lib -L/usr/lib/../lib -L/opt/intel/Compiler/11.1/046/lib/intel64 -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2/../../.. -lgfortranbegin -lgfortran -lm -ldl
libtool: link: gcc -funroll-loops -O2 -Wuninitialized -std=c99 -funsigned-char -pedantic -W -Wall -Wshadow -Wpointer-arith -Wcast-qual -Wcast-align -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations -Wnested-externs -Wunused -Wfloat-equal -rdynamic -O -o .libs/cs_solver ./.libs/libsaturne.so -L/home/alex/Desktop/Code_Saturne/lib /home/alex/Desktop/Code_Saturne/lib/libmei.so /home/alex/Desktop/Code_Saturne/lib/libfvm.so /usr/lib/libmedC.so -L/usr/lib /usr/lib/libhdf5.so -lpthread /home/alex/Desktop/Code_Saturne/lib/libbft.so -lz /usr/lib/libxml2.so -lcblas -latlas -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2 -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2/../../../../lib -L/lib/../lib -L/usr/lib/../lib -L/opt/intel/Compiler/11.1/046/lib/intel64 -L/usr/lib/gcc/x86_64-linux-gnu/4.3.2/../../.. -lgfortranbegin -lgfortran -lm -ldl -Wl,-rpath -Wl,/home/alex/Desktop/Code_Saturne/lib
./.libs/libsaturne.so: undefined reference to `_create_join_sync'
./.libs/libsaturne.so: undefined reference to `_destroy_join_sync'
collect2: ld returned 1 exit status
make[3]: *** [cs_solver] Error 1
make[3]: Leaving directory `/home/alex/Desktop/CDSat/ncs-2.0.0-beta2/src/apps'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/home/alex/Desktop/CDSat/ncs-2.0.0-beta2/src'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/alex/Desktop/CDSat/ncs-2.0.0-beta2'
make: *** [all] Error 2
alex@iskandhar:~/Desktop/CDSat/ncs-2.0.0-beta2$


Can somebody pls help me in this...?


Best regards


Alex

Hello Alex,

This is our fault... We added a last-minute patch to the 2.0-beta2 version but forgot to test that it compiled without MPI support. And it does not Sorry about that!
So, two solutions are possible at the moment:

• Wait a couple of days so that I can provide you with a patch (I have got one around)
• Compile Code_Saturne with MPI support (FVM and the kernel)

As a side note, you do not need to be that precise in the configure options for the dependencies, i.e. [code:1]--with-bft=/home/alex/Desktop/Code_Saturne[/code:1]is enough (you do not need to give separately headers, libraries and binaries). And if you want to have MEI library fully available (it is used for checking mathematical expressions in the graphical interface), you should install SWIG packages (development and standard packages) thanks to your package manager.

If you need more information, let me know.

Kind regards,
David

Hi David

Thanks for your reply,

So now CS can, as default, make thermo-hydraulic coupling... ( through syrthes ) ? nice...

I have compiled Syrthes with icc/ifort/mkl, is it possible to do the same with Code Saturne...? or is it better to stick to gcc/gfortran...?

I ask this because I have already compiled my Code aster 10.* with Intel dev. tools... as well as OpenFOAM-1.6.x...

In relation to Code coupling... in case I want to perform a Fluid-Thermal-Structural coupling ( resorting to a FEA analysis code such as ASTER ) do I have to specify it in Code saturne, or must it be done in Code Aster *.comm file... ( in other words, can Aster launch Code Saturne so as to find the pressure field in the bounding structure of a flow domain, and calculate its deformation in order to re-calculate the flow quatities in the "deformed" flow domain... and so on... till both solutions (structural and flow ) stabilize in a time stationary analysis...)...


It would be interesting to use Code Saturne-Syrthes-Code Aster to analyze a Stirling Engine, for example...

Best regards


Alex

Hi David...

Good news and not so good news...

I Have compiled completely the Code Saturne applications, did it as user. In /home/alex/Code_Saturne/bin if i launched ./cs gui the interface would appear... so far so good,

So i mv /home/alex/Code_Saturne /usr/local

In my ~/.bashrc i added the following lines :

#####################################################################################################
#### Code Saturne 2.2.0
export PATH=/usr/local/Code_Saturne/bin:/usr/local/Code_Saturne/share/ncs:$PATH
#####################################################################################################

so why do when I open a new shell, so that the environment is set, if i type cs create, a strange message pops up... :

alex@iskandhar:~$ cs
Traceback (most recent call last):
File "/usr/local/Code_Saturne/bin/cs", line 40, in <module>
import cs_config
ImportError: No module named cs_config
alex@iskandhar:~$


There was no reference to cs_config when i did the same thing in /home/alex/Code_Saturne/bin... something is different now, I guess something is not correctly configured, but what...?

This may sound like a n00b question.. but i am clueless...

Thanks in advance

Alex

So now CS can, as default, make thermo-hydraulic coupling... ( through syrthes ) ? nice...

Actually, Code_saturne has been able to do that for quite a long time, but SYRTHES has only been released open-source recently. We do not have yet a tutorial for the coupling setup but we will have a simple "how-to" soon.

I have compiled Syrthes with icc/ifort/mkl, is it possible to do the same with Code Saturne...? or is it better to stick to gcc/gfortran...?

No problem! Intel compilers usually yield better performance (up to 20% in some cases); Code_Saturne can be compiled with icc/ifort and should automatically find MKL library for BLAS support. To use Intel compilers, do something like:[code:1]./configure --prefix=... CC=icc FC=ifort[/code:1].

In relation to Code coupling... in case I want to perform a Fluid-Thermal-Structural coupling ( resorting to a FEA analysis code such as ASTER ) do I have to specify it in Code saturne, or must it be done in Code Aster *.comm file... ( in other words, can Aster launch Code Saturne so as to find the pressure field in the bounding structure of a flow domain, and calculate its deformation in order to re-calculate the flow quatities in the "deformed" flow domain... and so on... till both solutions (structural and flow ) stabilize in a time stationary analysis...)...

Actually, Code_Saturne/Code_Aster coupling is still under testing and will be done through the SALOME platform (YACS module to be precise). We still need to sort out a couple of things. At the moment, and depending on the simulation, you can run a fluid simulation, post-process a boundary part (in MED format) with only the pressure activated and then you run a Code_Aster simulation with this MED file as boundary conditions. This is not a coupled calculation but a chained one. To assess convergence, you can iterate the process.

It would be interesting to use Code Saturne-Syrthes-Code Aster to analyze a Stirling Engine, for example...

Indeed <br /><br />Post edited by: David Monfort, at: 2009/08/16 18:12

You are close to make it work.

What you cannot do is moving your installation in another place because some paths are &quot;hard-coded&quot; during the installation and furthermore the dynamic libraries will problably not be loaded because of the same &quot;hard-coded&quot; paths.

What you should do if you want to install Code_Saturne in /usr/local/Code_Saturne is:[code:1]./configure --prefix=/usr/local/Code_Saturne
make
sudo make install[/code:1]
Or really install it in /home/alex/Code_Saturne and leave it in place.

Let me know if it works.