Toolchains: Difference between revisions
No edit summary |
No edit summary |
||
Line 44: | Line 44: | ||
** [https://developer.nvidia.com/cuda-downloads CUDA 10.0] | ** [https://developer.nvidia.com/cuda-downloads CUDA 10.0] | ||
== Selected Makefiles == | |||
*[[A Debian based installation of VASP]] | *[[A Debian based installation of VASP]] | ||
*[[A Ubuntu based installation of VASP]] | *[[A Ubuntu based installation of VASP]] |
Revision as of 14:53, 31 January 2020
Apart from a C and FORTRAN compiler, the parallel versions of VASP "vasp_std, vasp_gam, vasp_ncl" depend on four additional libraries:
To compile VASP successfully, all versions of the chosen toolchain must be compatible with each other. This can sometimes be challenging, so that we provide a list of four compatible toolchains that have been used successfully for building and validating VASP 6.1.0 on Ubuntu Server 18.04.
- Intel
- Intel Parallel Studio 2020.0.166 (includes all libraries)
- PGI (contains LAPACK, OpenMPI and ScaLAPACK)
- PGI 19.10
- QD 2.3.17 (quadruple precision software library)
- FFTW 3.3.6 compiled with gcc 7.3.0
The CUDA ports "vasp_gpu, vasp_gpu_ncl" have been successfully build with following toolchains:
- PGI (contains LAPACK, OpenMPI)
- PGI 19.10
- QD 2.3.17 (quadruple precision software library)
- CUDA 10.0
- FFTW 3.3.6 compiled with gcc 7.3.0
- Intel (contains LAPACK and FFTW)
Selected Makefiles
- A Debian based installation of VASP
- A Ubuntu based installation of VASP
- A Fedora based installation of VASP
- A CentOS based installation of VASP
Related Sections
Installing VASP, Precompiler flags, GPU port of VASP, Validation tests