| Error Message | Probable Cause | Solution | | :--- | :--- | :--- | | Fatal error: Cannot open file 'prec.inc' | Missing preprocessing step. | Run make veryclean; make again. Ensure -DMPI is set. | | undefined reference to 'cheev_' | LAPACK/BLAS not linked correctly. | Add -mkl (Intel) or -lblas -llapack (GNU) to LLIBS . | | forrtl: severe (174): SIGSEGV | Stack size too small. | Set ulimit -s unlimited in your job script. | | Error: Type mismatch in argument at compilation | Old compiler incompatibility. | Use ifort version ≥ 18, or add -fallow-argument-mismatch for GCC≥10. |
cd testsuite/ make test This runs over 50 short calculations. Look for PASS results. Create a directory with testsuite/Si/ . Run mpirun -np 4 vasp_std . Compare your total energy and forces with the reference OUTCAR provided. For version 5.4.4, a silicon primitive cell should yield exactly -10.66142203 eV/atom . Part 7: Common Troubleshooting for vasp.5.4.4.tar.gz Even with the mature 5.4.4, errors are common. Here is a log of frequent issues. vasp.5.4.4.tar.gz
VASP 5.4.4 is sensitive to the order of libraries in the makefile.include . The safest order is: -lfftw3 -lmkl_scalapack_lp64 -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lpthread . Part 8: Why Not Newer? Comparing 5.4.4 vs. VASP 6.x You might ask: If VASP 6.x is available, why use vasp.5.4.4.tar.gz ? | Error Message | Probable Cause | Solution
# Precompiler options CPP_OPTIONS = -DMPI -D openmp -Duse_collective -DscaLAPACK -DCACHE_SIZE=4000 FC = ifort FCL = mpiifort -mkl FFTW FFTW_ROOT ?= /path/to/fftw3 LLIBS += -L$(FFTW_ROOT)/lib -lfftw3 BLAS/LAPACK via Intel MKL LLIBS += -mkl=sequential | | undefined reference to 'cheev_' | LAPACK/BLAS