Hi all, I noticed the following thing:
When I compile the siesta the FFLAGS setting like this: FFLAGS= -O2 -I$(MKLROOT)/include/intel64/lp64/ -I$(MKLROOT)/include Then, when I run the siesta after the compiling process finished, I see the following information on the stdout: --------------- werner@debian:~/software/hpc/siesta/siesta-trunk-449/Obj$ ./siesta Siesta Version: siesta-trunk-449 Architecture : ifort-from-composer_xe_2011_sp1.13.367+(mkl,impi)-from-l_ics_2013.1.039_intell64 Compiler flags: mpifc -fc=ifort -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/llp64/ -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include -O2 PP flags : -DFC_HAVE_FLUSH -DFC_HAVE_ABORT -DGRID_DP -DPHI_GRID_SP -DBROYDEN_DP -DMPI --------------- Please note the following line appeared in the above message: ------------ -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/llp64/ -------------- Though for my case, this will not do any harm due to the correct settings given by intel's mklvars.sh. But it's obviously that this is a bug of the siesta code when it pass the include line for compiling. The version I tested is siesta-trunk-449, as for other version, I don't know whether they has this issue or not. Regards -- Hongyi Zhao <[email protected]> Xinjiang Technical Institute of Physics and Chemistry Chinese Academy of Sciences GnuPG DSA: 0xD108493 On Sat, 2014-01-11 at 15:22 +0800, Hongyi Zhao wrote: > Please note the following line appeared in the above message: > > ------------ > -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/llp64/ > -------------- > > Though for my case, this will not do any harm due to the correct > settings given by intel's mklvars.sh. But it's obviously that this is a > bug of the siesta code when it pass the include line for compiling. Well, this is not so obvious to me. Can you please explain why you think it is a bug? Cheers, Bartek Dear Bartek Szyja, Perhaps I don't describe it so clearly. I mean in the arch.make, I use the following line: FFLAGS= -O2 -I$(MKLROOT)/include/intel64/lp64/ But, when I compiled it, the siesta give the following information: -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/llp64/ You can see the the path is expanded to a wrong one: -I$(MKLROOT)/include/intel64/lp64/ should be expanded to -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/lp64/ instead of -I/opt/intel/composer_xe_2013_sp1.0.080/mkl/include/intel64/llp64/ Regards hs.zhao Dear Hongyi Zhao, Can you please attach the arch.make file you are using? Also you have to be sure that the arch.make is at the correct directory. Regards, GBokas Sent from my iPad Dear GBokas, See the attachment. Then mkl link line setting used by me is obtain by intel's mkl_link_tool, which is a local version of the intel link advisor. I also use the suggested comipling directory by siesta group. i.e., the Obj directory. Regards hs.zhao In my case, the MKLROOT path varaible has been defined and loaded into the environment by the intel's script for mkl setting, i.e., the mklvars.sh. See the following for detail: werner@debian:~$ echo ${MKLROOT} /opt/intel/composer_xe_2013_sp1.0.080/mkl Regards hs Dear all siesta users I installed siesta package in serial version very well that header of siesta output is in below: “Siesta Version: siesta-3.2 Architecture : x86_64-unknown-linux-gnu--unknown Compiler flags: f95 -g -O2 SERIAL version * Running in serial mode >> Start of run: 8-JAN-2014 2:44:56” However, I cannot install siesta package in parallel version in fedora 17 linux (Processors: Intel core i7, operating system: 64bit). I installed MPICH2 and intel fortran compilers and MKL liberties by composer_xe_2011 version that they were addressed in bash in the following: export PATH_MPI=$PATH_MPI:/opt/mpich2-1.0.6/bin source /opt/intel/composer_xe_2011_sp1.9.293/bin/compilervars.sh intel64 source /opt/intel/composer_xe_2011_sp1.10.319/bin/compilervars.sh intel64 source /opt/intel/composer_xe_2011_sp1.9.293/mkl/bin/intel64/mklvars_intel64.sh source /opt/intel/composer_xe_2013.2.146/mkl/bin/intel64/mklvars_intel64.sh I installed siesta.3.1 . I run benzene example in Test folder by command “mpirun –np 8 siesta<benzene.fdf>benzene.out “. The output was not good that is in below: …. siesta: Atomic coordinates (Bohr) and species siesta: 4.69829 6.95419 13.04522 2 1 siesta: 6.03717 9.27319 13.04522 2 2 siesta: 8.71491 9.27319 13.04522 2 3 siesta: 10.05378 6.95419 13.04522 2 4 siesta: 8.71491 4.63520 13.04522 2 5 siesta: 6.03717 4.63520 13.04522 2 6 siesta: 2.59692 6.95419 13.04522 1 7 siesta: 4.98648 11.09303 13.04522 1 8 siesta: 9.76560 11.09303 13.04522 1 9 siesta: 12.15516 6.95419 13.04522 1 10 siesta: 9.76560 2.81536 13.04522 1 11 siesta: 4.98648 2.81536 13.04522 1 12 siesta: Automatic unit cell vectors (Ang): siesta: 11.776283 0.000000 0.000000 siesta: 0.000000 11.030875 0.000000 siesta: 0.000000 0.000000 6.266317 rank 7 in job 2 root_41367 caused collective abort of all ranks exit status of rank 7: return code 174 rank 6 in job 2 root_41367 caused collective abort of all ranks exit status of rank 6: killed by signal 9 rank 5 in job 2 root_41367 caused collective abort of all ranks exit status of rank 5: return code 174 rank 4 in job 2 root_41367 caused collective abort of all ranks exit status of rank 4: return code 174 rank 3 in job 2 root_41367 caused collective abort of all ranks exit status of rank 3: return code 174 rank 2 in job 2 root_41367 caused collective abort of all ranks exit status of rank 2: return code 174 rank 1 in job 2 root_41367 caused collective abort of all ranks exit status of rank 1: return code 174 What is the best arch.make file for my pc by above details? You can find my arch.make file in attachment. What is my problem? How can I solve it? Thanks lots. 1/13/2014 -- Esmaeil Abdoli Nashaljei, PhD of Solid state physics, Kashan, IRAN http://www.ssphysics.co.nr/
arch.make
Description: Binary data
