Welcome on the forums. Please post in English or French.
You are not logged in. You won't be able to post until you register and log in using the top-right bar.
I have been trying to install Code Aster on a powerpc on which I only have user permissions.
The admin confirmed that
gcc-c++.ppc64 (gcc c++ support)
gcc-gfortran.ppc64 (gcc fortran support)
are installed but I cannot sort out some lapack related errors like:
Checking for FFLAGS from setup.cfg... -fno-tree-dse
<DBG> Checking for initial value for FFLAGS... -fno-tree-dse
Checking for F90FLAGS from setup.cfg... -fno-tree-dse
<DBG> Checking for initial value for F90FLAGS... -fno-tree-dse
<DBG> Checking for initial value for CFLAGS_DBG... empty
<DBG> Checking for initial value for FFLAGS_DBG... empty
<DBG> Checking for initial value for F90FLAGS_DBG... empty
<DBG> Checking for initial value for LDFLAGS... empty
<DBG> get_cmd_compil returns : gcc -o main.o -DLINUX64 -O2 -fno-stack-protector -fPIC -c main.c
<DBG> get_cmd_compil returns : /usr/bin/gfortran -o test.o -DLINUX64 -fno-tree-dse -fdefault-double-8 -fdefault-integer-8 -fdefault-real-8 -c test.f
<DBG> get_cmd_compil returns : /usr/bin/gfortran -o a.out main.o test.o -L/usr/lib/gcc/ppc64-redhat-linux/4.4.4 -lstdc++ -L/home/pschmitt/aster-full-src-10.7.0/liblinks -llapack -L/home/pschmitt/aster-full-src-10.7.0/liblinks -lblas -L/usr/lib64 -lpthread -L/usr/lib64 -lz
<DBG> ERROR : iret = 4 +++ STANDARD OUTPUT: RESULTS : 10 / '01/01/2010' / 1 / 0.000000 / 5.000000
<DBG> +++ STANDARD ERROR: +++ END
<DBG> working directory is : /tmp/check_compilers.68172/run_test
Checking for C/fortran program using blas/lapack... no
---------- ERROR MESSAGE ----------
lapack and lablas are in /usr/lib64, I linked them to another folder as shown here
[pschmitt@localhost aster-full-src-10.7.0]$ ls -rtl liblinks
lrwxrwxrwx. 1 pschmitt pschmitt 29 Jul 5 12:43 liblapack.so -> /usr/lib64/liblapack.so.3.2.1
lrwxrwxrwx. 1 pschmitt pschmitt 27 Jul 6 12:05 libblas.so -> /usr/lib64/libblas.so.3.2.1
and then use
MATHLIB = '-L/home/pschmitt/aster-full-src-10.7.0/liblinks -llapack -L/home/pschmitt/aster-full-src-10.7.0/liblinks -lblas'
to specify the directory but with no success.
Thanks for any ideas on how to fix this,
Last edited by wavemaster (2012-07-16 14:41:53)
I get a similar error (approximately) when I use the install script and point to the threaded Intel maths libraries instead of the default sequential ones (Ubuntu 12.04). When the script asks whether I want to proceed despite the error I enter "yes" and the compilation continues successfully.
I'm not sure if this is any help...
I had a look at the setup.log files you submitted and it seems to me everything went OK although there's an error about BLAS/Lapack at the configure stage.
Indeed everything installed OK except :
- grace but this is because you're missing the motif library. Moreover grace is optional (it is used to plot some graphs).
- homard : this is a mesh refinement software. The aster package contains an x86 binary which therefore will not work for ppc64. It is also optional so no worries.
Now you should try to go to the install location and try to launch a simple test :
$ cd /home/pschmitt/aster/testing $ /home/pschmitt/aster/bin/as_run forma01a.export (or any export in this directory)
This will launch a simple test. If the diagnostic is OK then your install works.
However in your case it should fail. This is expected. Indeed Aster makes some internal assumptions, one of which is that we're running a little endian processor. Since PowerPC is big endian, you'll need to fix one function in the source code.
This is explained here : http://www.code-aster.org/forum2/viewto … 731#p11731
After fixing envima.c, you need to rebuild the Code_Aster binary :
$ /home/pschmitt/aster/bin/as_run --make
Last edited by Thomas DE SOZA (2012-07-19 17:38:31)
Thanks you very much for your comments.
When i run the forma1.export test case I still get errors, which is different from the one in the post
http://www.code-aster.org/forum2/post.p … ;qid=11714
. I will try to apply the patch and update you on how it goes...