My Community

Dear users,
I ran the same system in two clusters, and in one of them, it looked very well. but in another, the calculation can not reach convergence within default 60 cycles. and the vasp.4.lib includes files as follows:

crayerrf.F derrf_t3d.c drdatab.o lapack_single_1.1.f makefile.fujitsu makefile.linux_ifc_P4 makefile.t3d README.lapack timing.fujitsu.F
dclock_.c diolib.f errf.F lapack_single.f makefile.hp makefile.linux_pg makefile.t3e sclock_cray.F timing_.o
dclock.c diolib.F fujitsu.F libdmy.a makefile.hpux_itanium makefile.linux_pgi_opt makefile.vpp sclock_nec.F
dclock_ds20.c diolib.o itmtv.s linpack_double.f makefile.linux makefile.nec preclib.f sclock_t3d.F
dclock_.o dlexlib.f lapack_atlas.f linpack_double.o makefile.linux_abs makefile.rs6000 preclib.F serrf.c
dclock_simple.c dlexlib.F lapack_atlas.o linpack_single.f makefile.linux_alpha makefile.rs6000_p1 preclib.mod stripnr
derrf_.c dlexlib.o lapack_double_1.1.f Makefile makefile.linux_efc_itanium makefile.sgi preclib.o timing_.c
derrf.c drdatab.f lapack_double.f makefile.cray makefile.linux_ifc_ath makefile.sp2 ptimers.f timing.c
derrf_.o drdatab.F lapack_double.o makefile.dec makefile.linux_ifc_opt makefile.sun ptimers.h timing_ds20.c

the current .lib is suitble for parallel calculations.


thanks

the information you give here does not help at all to find out, please check where the calculations start to differ.
are there any hints or warnings in the job.stdout file?
Do you use the same compiler on both clusters? if not: Try in the calculation converges if you compile VASP at lowest optimization.
If it does not: Maybe the installation of the libraries (BLAS) is not correct on one of the clusters?