Dear VASP users
I am doing MD simulation of 100 surface with 15 Angs. vacuum. My problem is that my run stops automatically after 2 iteration (I gave NSW=5000). I have changed KPOINTS, NPAR value etc. but still problem is not resolved.
With all same INCAR tags, I have run bulk calculation which completed successfully. I do not know where is the problem in surface case. I have attached here my INCAR file and POSCAR.
Can you please help me resolving this problem by your experience. Where is the problem ?
Thank You
Shilendra
Dear Shilendra,
Did you try to change the ALGO tag to Fast or Normal? Sometimes VeryFast gives some optimization problems.
Regards
Dear Sinhué López,
I have tried ALGO= Normal also but still same problem. After 2 iteration it does not give any value but job runs for long time until and unless we do not kill that job forcefully.
I asked another researchgate colleague, he is saying with my input in his cluster job is running after 2 iteration also. In my cluster other jobs are running without problem. Even same job for DFT calculation without MD is also running well. Only for MD, it is creating problem.
Thank You
Shilendra
From my small experience, I have seen problems with parallel computing. You can check this type of error if you run your job on a single node and single core. Another possible error can be if you have less number of bands. It will be better if you let us know what type of error message you are getting.
Thank You Maitrayee for your kind reply !
It does not kill the job but stops giving energy values after 2 iteration and job running time increases continously. After killing job forcefully it shows message like
Fatal error in PMPI_Allgatherv: Invalid buffer pointer, error stack: PMPI_Allgatherv(1471): MPI_Allgatherv(sbuf=0xb8623c0, scount=13600, MPI_DOUBLE_COMPLEX, rbuf=0xb8623c0, rcounts=0x93c9db0, displs=0x93b81a0, MPI_DOUBLE_COMPLEX, comm=0x84000003) failed PMPI_Allgatherv(1406): Buffers must not be aliased
In my colleague cluster it is running well. It does not stop after 2 iteration with same input whatever I am using in my cluster and without MD same DFT job is running successfully.
What is the solution of this problem ? Since MPT is not creating any problem for DFT calculation when I am doing MD then it is creating problem. This is surface calculation for bulk MD system in same cluster job is running without any error.
Thank You
Shilendra
Thank you dear Girinath Pillai for your kind reply !
If I will have any issue, I will email you.
Thank You
Shilendra
Dear Shilendra,
Have you solved this problem, I meet the same problem, MD calculation stoped after 2 iteration.
- Badges
- Science topic