Dear, M. Dolfi I have installed alps-2.2.b4 using openmpi successfully on Suse 11.0. There are not any errors during make, make test, and make install. After installation I can running successfully mps_optim serially or 1-4 CPUs. I have encountered some errors while running mps_optim parallelly using 4 or more CPUs.
...... Running task 1. Running task 1. This binary contains symmetries: nu1 This binary contains symmetries: nu1 MAQUIS repo r4143 MAQUIS repo r4143 Temporary storage enabled in /home/mlj/bin/TempFiles/n3SquareLatticeSite4.64/storage_temp_fc5a128613e7/ Temporary storage enabled in /home/mlj/bin/TempFiles/n3SquareLatticeSite4.64/storage_temp_5a9cbccacee8/ Error destroying HDF5 context of file '/home/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5.tmp.4' boost::filesystem::rename: No such file or directory: "/home/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5.tmp.4", "/home/mlj/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5" -------------------------------------------------------------------------- mpirun noticed that process rank 3 with PID 5962 on node linux exited on signal 6 (Aborted). -------------------------------------------------------------------------- FILE ./n3SquareLatticeSite4.task1.out.h5DOES NOT EXIST!
How to solve this problem?
Thank you for your help.
L.J. Meng
Xiangtan University,Hunan Pro. China
The MPS applications don’t support MPI parallelization. What you are effectively doing is launching the mps_optim 4 times working on the same input and output files, i.e. there will be conflicts when writing the output.
What you can use is OpenMP, which is enabled during the cmake configuration process by passing the option -DALPS_ENABLE_OPENMP=ON. Then you control the number of threads with the OMP_NUM_THREADS environment variable.
Best regards, Michele
-- ETH Zurich Michele Dolfi Institute for Theoretical Physics HIT G 32.4 Wolfgang-Pauli-Str. 27 8093 Zurich Switzerland
dolfim@phys.ethz.ch www.itp.phys.ethz.ch
+41 44 633 78 56 phone +41 44 633 11 15 fax
On Mar 14, 2016, at 3:35 AM, Forwordom mljphy@qq.com wrote:
Dear, M. Dolfi I have installed alps-2.2.b4 using openmpi successfully on Suse 11.0. There are not any errors during make, make test, and make install. After installation I can running successfully mps_optim serially or 1-4 CPUs. I have encountered some errors while running mps_optim parallelly using 4 or more CPUs.
...... Running task 1. Running task 1. This binary contains symmetries: nu1 This binary contains symmetries: nu1 MAQUIS repo r4143 MAQUIS repo r4143 Temporary storage enabled in /home/mlj/bin/TempFiles/n3SquareLatticeSite4.64/storage_temp_fc5a128613e7/ Temporary storage enabled in /home/mlj/bin/TempFiles/n3SquareLatticeSite4.64/storage_temp_5a9cbccacee8/ Error destroying HDF5 context of file '/home/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5.tmp.4' boost::filesystem::rename: No such file or directory: "/home/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5.tmp.4", "/home/mlj/mlj/Test/n3SquareLatticeSite4.task1.out.0.h5"
mpirun noticed that process rank 3 with PID 5962 on node linux exited on signal 6 (Aborted).
FILE ./n3SquareLatticeSite4.task1.out.h5DOES NOT EXIST!
How to solve this problem?
Thank you for your help.
L.J. Meng Xiangtan University,Hunan Pro. China
comp-phys-alps-users@lists.phys.ethz.ch