If you have Intel’s MKL installed, and it is found by CMake, performance will be helpedīut by how much idepends on which architecture-specific implementation MKL picks at runtime. In Eigen is automatically enabled by injection of -march=native. We recommend using g++ or clang rather than an Intel compiler on AMD. DCMAKE_CXX_FLAGS=-mtune=native (at probably lower performance), or perform node-typeĪs a rough estimate, runtimes with vectorization and gcc/clang are 30% shorter than without This may cause the executables to not run. If you compile VOTCA on a heterogeneous cluster with different instruction sets, If g++ or clang are used, the compiler option -march=native is automatically injected No special flags have to be supplied to CMake. G++, clang, and ipcx from the Intel OneAPI basekit give similar performance
Below are some recommendationsįor different architectures: Intel Processors ¶ Performance can be gained by enabling vectorization and/or use of Intel’s MKLĪs backend, which is automatically detected by CMake. VOTCA-XTP relies on the Eigen library for vector-matrix operations, and a lot of Package HDF5 (ON/OFF, Default OFF relevant only for the
Cmake command not found git bash install#
INSTALL_CSGAPPS - Install the extra csg applications repo (ON/OFF,īUILD_XTP - Build the xtp repo (ON/OFF, Default OFF)ĬMAKE_INSTALL_PREFIX - where to install the votca executablesīUILD_TESTING - compile tests (ON/OFF, Default ON)ĬMAKE_DISABLE_FIND_PACKAGE_ - Disable using an optionalĬMAKE_DISABLE_FIND_PACKAGE_HDF5 - Disable using the optional Related tweaks when setting ENABLE_RPATH_INJECT to ON. VOTCA has enabled this option and a couple of other rpath Which can be enabled by setting CMAKE_INSTALL_RPATH_USE_LINK_PATH to Alternatively,ĬMake has options to remember where libraries came from at link time, Running an executable from the VOTCA installation afterwards (which canīe automated via the user’s login profile, e.g. Note that LD_LIBRARY_PATH also needs to be set every time when For example, in the case of a locally installed version ofĮxport LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/gromacs/lib:$HOME/anaconda/lib Will have to rerun the cmake command with the relevant flag(s)Īdded.
D flag is necessary to specify the path to a missed package. (most often a shared or dynamically loaded library, e.g.Įrror messages produced by CMake usually provide instructive suggestionsįor resolving dependency issues. Need to find the ‘non-standard’ location for each missing dependency Resolving the ‘not found’ dependency errors ¶Ĭorrectly installed, one or more might still appear with the ‘not found’ status uponĬonfiguring with the cmake command (see above). So in this case run this orĭownload the package from here. On Ubuntu 20.10 and older you will not get version 1.0.5 of the libecpint package. If you have an older version of a distribution, cmake can be run with the -DBUILD_OWN_GROMACS=ON or -DBUILD_OWN_LIBINT=ON flags to automatically install the correct GROMACS and libint version. Version=master # or 'stable' or 'v2023-rc.1'Ĭmake -B builddir -S votca -DBUILD_XTP=ON -DCMAKE_INSTALL_PREFIX=$
0 kommentar(er)
