I am using lammps-31Mar17 MPI version. With the script file listed below, I found mpirun -np 1 could produce good result, however, running with more threads like -np 20, the error message comes up like "*** Error in `./lmp_mpi': malloc(): memory corruption (fast): 0x0000000002167da0 ***".
My computer have two processors "Intel® Xeon(R) CPU E5-2698 v4 @ 2.20GHz × 16" with 40 cores, and the number of threads usually didn't effect the outcome without computing the cluster ID. So I guess the problem has something to do with the compute cluster/atom command.
I wonder if this error is due to a mistake on my side, or a problem in lammps. Please shed some light on this error. Testing the script file and posting your results would also be very helpful.
I am using lammps-31Mar17 MPI version. With the script file listed below,
I found mpirun -np 1 could produce good result, however, running with more
threads like -np 20, the error message comes up like "*** Error in
`./lmp_mpi': malloc(): memory corruption (fast): 0x0000000002167da0 ***".
My computer have two processors "Intel® Xeon(R) CPU E5-2698 v4 @ 2.20GHz ×
16" with 40 cores, and the number of threads usually didn't effect the
outcome without computing the cluster ID. So I guess the problem has
something to do with the compute cluster/atom command.
I wonder if this error is due to a mistake on my side, or a problem in
lammps. Please shed some light on this error. Testing the script file and
posting your results would also be very helpful.
sorry, i cannot reproduce the error you are seeing with the 19May2017
version of LAMMPS with gcc 6.3 and MPICH 3.2 using the unmodified mpi
makefile
please let us know what compiler, compiler version, compiler flags, MPI
library etc. you are using.