**Installation of OpenMX Ver. 3.9 on Ubuntu 18.04.4 LTS on WSL on Windows 10 (64bit)** |
- Date: 2021/08/24 21:14
- Name:
**T. Ozaki**
- Dear all,
I would like to share my experience of installation of OpenMX Ver. 3.9 with patch 3.9.6 on Ubuntu 18.04.4 LTS on WSL1 and WSL2 on Windows 10 (64bit) as below.
Best regards,
Taisuke Ozaki
-----------------------
1. Installation of Intel oneAPI Base Toolkit + HPC Toolkit
Intel oneAPI Toolkits are now free for all developers at https://software.intel.com/content/www/us/en/develop/articles/free-intel-software-developer-tools.html which includes intel compilers, MKL, intel MPI, blacs, ScaLapack, and FFTW: all are required for the OpenMX installation.
Via the website, download l_BaseKit_p_2021.3.0.3219_offline.sh with the following specification: Operating System: Linux Distribution: Web&Local (recommended) Installer Type: Local
and do the following on command line of WSL: sudo bash l_BaseKit_p_2021.3.0.3219_offline.sh
Also, via the wesite, download l_HPCKit_p_2021.3.0.3230_offline.sh with the following specification: Operating System: Linux Distribution: Web&Local (recommended) Installer Type: Local
and do the following on command line of WSL: sudo bash l_HPCKit_p_2021.3.0.3230_offline.sh
After the installation of oneAPI Base Toolkit and HPC Toolkit, perform the following on command line: source /opt/intel/oneapi/setvars.sh
Also, it would be better to add source /opt/intel/oneapi/setvars.sh to .bashrc Then, you can use Intel oneAPI Base Toolkit and HPC Toolkit on Ubuntu 18.04.4 LTS on WSL on Windows 10 64bit.
2. Installation of OpenMX Ver. 3.9 with patch 3.9.6
In the makefile of OpenMX Ver. 3.9 with patch 3.9.6, specify MKLROOT, CC, FC, and LIB as follows: (As for the installation of OpenMX Ver. 3.9 with patch 3.9.6, please also refer http://www.openmx-square.org/bugfixed/21Aug21/README.txt )
MKLROOT = /opt/intel/oneapi/mkl/2021.3.0/ CC = mpiicc -O3 -xHOST -ip -no-prec-div -qopenmp -I${MKLROOT}/include -I${MKLROOT}/include/fftw FC = mpiifort -O3 -xHOST -ip -no-prec-div -qopenmp LIB= -L${MKLROOT}/lib/intel64 -lmkl_scalapack_lp64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lifcore -lmkl_blacs_intelmpi_lp64 -liomp5 -lpthread -lm -ldl
Then, for the installation you can perform as make all make install
After the installation, you will get the executable file: openmx in the work directory. Using 4 cores of Intel(R) Core(TM) i7-8550U CPU@1.8GHz with memory of 15.9 GB, which is my laptop computer, the runtest results are as follows:
WSL1: mpirun -np 4 ./openmx -runtest -nt 1
1 input_example/Benzene.dat Elapsed time(s)= 10.35 diff Utot= 0.000000000031 diff Force= 0.000000000003 2 input_example/C60.dat Elapsed time(s)= 48.24 diff Utot= 0.000000000000 diff Force= 0.000000000001 3 input_example/CO.dat Elapsed time(s)= 23.61 diff Utot= 0.000000000096 diff Force= 0.000000000262 4 input_example/Cr2.dat Elapsed time(s)= 21.58 diff Utot= 0.000000000944 diff Force= 0.000000000294 5 input_example/Crys-MnO.dat Elapsed time(s)= 84.86 diff Utot= 0.000000000003 diff Force= 0.000000000004 6 input_example/GaAs.dat Elapsed time(s)= 99.17 diff Utot= 0.000000000000 diff Force= 0.000000000000 7 input_example/Glycine.dat Elapsed time(s)= 11.67 diff Utot= 0.000000000001 diff Force= 0.000000000000 8 input_example/Graphite4.dat Elapsed time(s)= 7.99 diff Utot= 0.000000000006 diff Force= 0.000000000127 9 input_example/H2O-EF.dat Elapsed time(s)= 8.70 diff Utot= 0.000000000000 diff Force= 0.000000000002 10 input_example/H2O.dat Elapsed time(s)= 8.13 diff Utot= 0.000000000000 diff Force= 0.000000000011 11 input_example/HMn.dat Elapsed time(s)= 29.00 diff Utot= 0.000000000131 diff Force= 0.000000000021 12 input_example/Methane.dat Elapsed time(s)= 6.54 diff Utot= 0.000000000016 diff Force= 0.000000000001 13 input_example/Mol_MnO.dat Elapsed time(s)= 17.34 diff Utot= 0.000000000370 diff Force= 0.000000000118 14 input_example/Ndia2.dat Elapsed time(s)= 8.71 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 385.90
WSL1: mpirun -np 2 ./openmx -runtest -nt 2
1 input_example/Benzene.dat Elapsed time(s)= 42.82 diff Utot= 0.000000000014 diff Force= 0.000000000000 2 input_example/C60.dat Elapsed time(s)= 111.33 diff Utot= 0.000000000004 diff Force= 0.000000000000 3 input_example/CO.dat Elapsed time(s)= 90.98 diff Utot= 0.000000000095 diff Force= 0.000000003127 4 input_example/Cr2.dat Elapsed time(s)= 136.40 diff Utot= 0.000000000981 diff Force= 0.000000000100 5 input_example/Crys-MnO.dat Elapsed time(s)= 186.73 diff Utot= 0.000000000000 diff Force= 0.000000000092 6 input_example/GaAs.dat Elapsed time(s)= 239.03 diff Utot= 0.000000000011 diff Force= 0.000000000001 7 input_example/Glycine.dat Elapsed time(s)= 110.98 diff Utot= 0.000000000001 diff Force= 0.000000000000 8 input_example/Graphite4.dat Elapsed time(s)= 100.46 diff Utot= 0.000000000009 diff Force= 0.000000000042 9 input_example/H2O-EF.dat Elapsed time(s)= 82.04 diff Utot= 0.000000000002 diff Force= 0.000000000002 10 input_example/H2O.dat Elapsed time(s)= 105.53 diff Utot= 0.000000000001 diff Force= 0.000000000386 11 input_example/HMn.dat Elapsed time(s)= 186.35 diff Utot= 0.000000000151 diff Force= 0.000000000022 12 input_example/Methane.dat Elapsed time(s)= 95.88 diff Utot= 0.000000000000 diff Force= 0.000000000002 13 input_example/Mol_MnO.dat Elapsed time(s)= 194.06 diff Utot= 0.000000000316 diff Force= 0.000000000082 14 input_example/Ndia2.dat Elapsed time(s)= 123.05 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 1805.65
Though the speed is not so fast, one can confirm that OpenMX runs normally in both the MPI and MPI/OpenMP hybrid modes.
I have also tested "runtest" on WSL2 on the same machine, and obtained the following results:
WSL2: mpirun -np 4 ./openmx -runtest -nt 1
1 input_example/Benzene.dat Elapsed time(s)= 8.26 diff Utot= 0.000000000031 diff Force= 0.000000000003 2 input_example/C60.dat Elapsed time(s)= 39.83 diff Utot= 0.000000000000 diff Force= 0.000000000000 3 input_example/CO.dat Elapsed time(s)= 17.06 diff Utot= 0.000000000096 diff Force= 0.000000000272 4 input_example/Cr2.dat Elapsed time(s)= 13.07 diff Utot= 0.000000000943 diff Force= 0.000000000292 5 input_example/Crys-MnO.dat Elapsed time(s)= 42.87 diff Utot= 0.000000000003 diff Force= 0.000000000007 6 input_example/GaAs.dat Elapsed time(s)= 73.80 diff Utot= 0.000000000000 diff Force= 0.000000000000 7 input_example/Glycine.dat Elapsed time(s)= 10.12 diff Utot= 0.000000000001 diff Force= 0.000000000000 8 input_example/Graphite4.dat Elapsed time(s)= 6.38 diff Utot= 0.000000000002 diff Force= 0.000000000053 9 input_example/H2O-EF.dat Elapsed time(s)= 7.05 diff Utot= 0.000000000000 diff Force= 0.000000000002 10 input_example/H2O.dat Elapsed time(s)= 5.54 diff Utot= 0.000000000000 diff Force= 0.000000000011 11 input_example/HMn.dat Elapsed time(s)= 24.60 diff Utot= 0.000000000131 diff Force= 0.000000000021 12 input_example/Methane.dat Elapsed time(s)= 5.25 diff Utot= 0.000000000016 diff Force= 0.000000000001 13 input_example/Mol_MnO.dat Elapsed time(s)= 13.90 diff Utot= 0.000000000370 diff Force= 0.000000000118 14 input_example/Ndia2.dat Elapsed time(s)= 7.57 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 275.30
WSL2: mpirun -np 2 ./openmx -runtest -nt 2
1 input_example/Benzene.dat Elapsed time(s)= 7.58 diff Utot= 0.000000000016 diff Force= 0.000000000003 2 input_example/C60.dat Elapsed time(s)= 44.08 diff Utot= 0.000000000005 diff Force= 0.000000000001 3 input_example/CO.dat Elapsed time(s)= 15.27 diff Utot= 0.000000000095 diff Force= 0.000000003113 4 input_example/Cr2.dat Elapsed time(s)= 10.46 diff Utot= 0.000000000485 diff Force= 0.000000000013 5 input_example/Crys-MnO.dat Elapsed time(s)= 48.64 diff Utot= 0.000000000001 diff Force= 0.000000000029 6 input_example/GaAs.dat Elapsed time(s)= 75.99 diff Utot= 0.000000000011 diff Force= 0.000000000000 7 input_example/Glycine.dat Elapsed time(s)= 9.36 diff Utot= 0.000000000001 diff Force= 0.000000000001 8 input_example/Graphite4.dat Elapsed time(s)= 5.46 diff Utot= 0.000000000000 diff Force= 0.000000000062 9 input_example/H2O-EF.dat Elapsed time(s)= 7.33 diff Utot= 0.000000000000 diff Force= 0.000000000003 10 input_example/H2O.dat Elapsed time(s)= 4.78 diff Utot= 0.000000000000 diff Force= 0.000000003251 11 input_example/HMn.dat Elapsed time(s)= 23.01 diff Utot= 0.000000000151 diff Force= 0.000000000022 12 input_example/Methane.dat Elapsed time(s)= 4.70 diff Utot= 0.000000000001 diff Force= 0.000000000001 13 input_example/Mol_MnO.dat Elapsed time(s)= 13.68 diff Utot= 0.000000000150 diff Force= 0.000000000001 14 input_example/Ndia2.dat Elapsed time(s)= 5.21 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 275.55
WSL2: mpirun -np 1 ./openmx -runtest -nt 4
1 input_example/Benzene.dat Elapsed time(s)= 8.44 diff Utot= 0.000000000034 diff Force= 0.000000000007 2 input_example/C60.dat Elapsed time(s)= 41.34 diff Utot= 0.000000000008 diff Force= 0.000000000001 3 input_example/CO.dat Elapsed time(s)= 14.01 diff Utot= 0.000000000026 diff Force= 0.000000007116 4 input_example/Cr2.dat Elapsed time(s)= 9.90 diff Utot= 0.000000000076 diff Force= 0.000000000126 5 input_example/Crys-MnO.dat Elapsed time(s)= 64.26 diff Utot= 0.000000000002 diff Force= 0.000000000020 6 input_example/GaAs.dat Elapsed time(s)= 103.12 diff Utot= 0.000000000011 diff Force= 0.000000000000 7 input_example/Glycine.dat Elapsed time(s)= 10.61 diff Utot= 0.000000000001 diff Force= 0.000000000001 8 input_example/Graphite4.dat Elapsed time(s)= 4.95 diff Utot= 0.000000000032 diff Force= 0.000000000100 9 input_example/H2O-EF.dat Elapsed time(s)= 7.12 diff Utot= 0.000000000001 diff Force= 0.000000000002 10 input_example/H2O.dat Elapsed time(s)= 5.12 diff Utot= 0.000000000001 diff Force= 0.000000003260 11 input_example/HMn.dat Elapsed time(s)= 26.83 diff Utot= 0.000000000150 diff Force= 0.000000000021 12 input_example/Methane.dat Elapsed time(s)= 4.21 diff Utot= 0.000000000000 diff Force= 0.000000000001 13 input_example/Mol_MnO.dat Elapsed time(s)= 15.21 diff Utot= 0.000000000087 diff Force= 0.000000000092 14 input_example/Ndia2.dat Elapsed time(s)= 5.74 diff Utot= 0.000000000000 diff Force= 0.000000000001
Total elapsed time (s) 320.86
In summary, we see the following results:
WSL1, 4 MPI procs. & 1 OMP threads: 385.90 sec. WSL1, 2 MPI procs. & 2 OMP threads:1805.65 sec. WSL2, 4 MPI procs. & 1 OMP threads: 275.30 sec. WSL2, 2 MPI procs. & 2 OMP threads: 275.55 sec. WSL2, 1 MPI procs. & 4 OMP threads: 320.86 sec.
The results suggest that WSL2 is faster for the execution of OpenMX Ver. 3.9.6.
| |