Re: Test issues ( No.1 ) |
- Date: 2020/11/01 20:06
- Name: Gilad Gani <giladg@campus.technion.ac.il>
- continuation of the results file:
Total elapsed time (s) 1625.78 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 13 input_example/Mol_MnO.dat Elapsed time(s)= 47.99 diff Utot= 0.000000000248 diff Force= 0.000000000095 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 13 input_example/Mol_MnO.dat Elapsed time(s)= 47.16 diff Utot=121.101364666716 diff Force=1000.010883822193 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 14 input_example/Ndia2.dat Elapsed time(s)= 19.41 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 1594.92 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 14 input_example/Ndia2.dat Elapsed time(s)= 22.57 diff Utot=11.571668430754 diff Force=1000.000000000001
Total elapsed time (s) 1579.54 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 14 input_example/Ndia2.dat Elapsed time(s)= 20.79 diff Utot=11.571668430754 diff Force=1000.000000000001
Total elapsed time (s) 1477.48 14 input_example/Ndia2.dat Elapsed time(s)= 14.67 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 1657.83 Invalid comparison due to the different number of grids. You may use a different radix for FFT. 14 input_example/Ndia2.dat Elapsed time(s)= 13.88 diff Utot=11.571668430754 diff Force=1000.000000000001
Total elapsed time (s) 1644.04
|
Re: Test issues ( No.2 ) |
- Date: 2020/11/01 23:53
- Name: Naoya Yamaguchi
- Dear Gilad,
It looks a problem of MPI. Can you try a "-runtest" calculation with one MPI process?
Regards, Naoya Yamaguchi
|
Re: Test issues ( No.3 ) |
- Date: 2020/11/02 18:06
- Name: Gilad Gani <giladg@campus.technion.ac.il>
- Dear Naoya,
I ran the command : LD_LIBRARY_PATH=/opt/intel/mkl/lib/intel64 mpirun -np 1 openmx -runtest
and got the following in the runtest file:
1 input_example/Benzene.dat Elapsed time(s)= 14.24 diff Utot= 0.000000000027 diff Force= 0.000000000016 2 input_example/C60.dat Elapsed time(s)= 112.73 diff Utot= 0.000000000004 diff Force= 0.000000000005 3 input_example/CO.dat Elapsed time(s)= 27.59 diff Utot= 0.000000000093 diff Force= 0.000000007049 4 input_example/Cr2.dat Elapsed time(s)= 22.00 diff Utot= 0.000000001192 diff Force= 0.000000000074 5 input_example/Crys-MnO.dat Elapsed time(s)= 173.49 diff Utot= 0.000000000005 diff Force= 0.000000000026 6 input_example/GaAs.dat Elapsed time(s)= 133.69 diff Utot= 0.000000000003 diff Force= 0.000000000001 7 input_example/Glycine.dat Elapsed time(s)= 21.30 diff Utot= 0.000000000001 diff Force= 0.000000000001 8 input_example/Graphite4.dat Elapsed time(s)= 27.11 diff Utot= 0.000000000017 diff Force= 0.000000000026 9 input_example/H2O-EF.dat Elapsed time(s)= 13.35 diff Utot= 0.000000000000 diff Force= 0.000000000004 10 input_example/H2O.dat Elapsed time(s)= 7.99 diff Utot= 0.000000000001 diff Force= 0.000000003221 11 input_example/HMn.dat Elapsed time(s)= 46.50 diff Utot= 0.000000000009 diff Force= 0.000000000022 12 input_example/Methane.dat Elapsed time(s)= 7.43 diff Utot= 0.000000000002 diff Force= 0.000000000001 13 input_example/Mol_MnO.dat Elapsed time(s)= 27.56 diff Utot= 0.000000000248 diff Force= 0.000000000095 14 input_example/Ndia2.dat Elapsed time(s)= 13.51 diff Utot= 0.000000000001 diff Force= 0.000000000000
Total elapsed time (s) 648.48
Thank you, Gilad
|
Re: Test issues ( No.4 ) |
- Date: 2020/11/02 18:22
- Name: Naoya Yamaguchi
- Dear Gilad,
Since the calculation with one process is done normally, the problem may be caused by the mismatch of MPI. Please check the combination of MPI you use and MPI OpenMX calls.
Regards, Naoya Yamaguchi
|
Re: Test issues ( No.5 ) |
- Date: 2020/11/02 19:15
- Name: Gilad Gani <giladg@campus.technion.ac.il>
- Dear Naoya,
Okay, I will check it. are there any major disadvantages to using one process? It is important to note that this installment will be used neither for any major nor time-consuming calculations, but only for test runs with very small models.
Thank you, Gilad
|
Re: Test issues ( No.6 ) |
- Date: 2020/11/02 19:36
- Name: Naoya Yamaguchi
- Dear Gilad,
>are there any major disadvantages to using one process? Without parallel calculations with MPI, OpenMX can use only one node that limits a total amount of used RAM and one process that does calculation in the same way as a PC with a single core CPU.
Regards, Naoya Yamaguchi
|