You need:

  • CMake (3.22+)
  • GNU make or Ninja
  • Fortran compiler which supports at least Fortran 2008 (including the TS 29113 when using the C-bindings)
  • BLAS+LAPACK implementation (reference, OpenBLAS and MKL have been tested. Note: DBCSR linked to OpenBLAS 0.3.6 gives wrong results on Power9 architectures.)
  • Python version installed (2.7 or 3.6+ have been tested)


  • LIBXSMM (1.10+, and pkg-config) for Small Matrix Multiplication acceleration
  • LAPACK implementation (reference, OpenBLAS-bundled and MKL have been tested), required when building the tests

To build DBCSR's GPU backend:

  • CUDA Toolkit (targets NVIDIA GPUs, minimal version required: 5.5) with cuBLAS
    • Host C++ compiler which supports at least C++11 standard
  • or HIP compiler (targets NVIDIA or AMD GPUs) and hipBLAS (ROCm 4.5.2 was tested)
    • Host C++ compiler which supports at least C++11 standard
  • or OpenCL, i.e., development headers (opencl-headers), generic loader "ocl-icd" (ocl-icd-opencl-dev),
    • Vendor specific OpenCL package, e.g., Intel Compute Runtime, or CUDA Toolkit (includes OpenCL)
    • Nvidia GPU mode shall be DEFAULT (nvidia-smi -c DEFAULT) if MPI puts multiple ranks onto a single GPU; MPS daemon with GPU mode EXCLUSIVE_PROCESS is not an option
    • For the OpenCL backend, a plain C compiler is sufficient (C90 standard)
    • Optionally, clinfo (can be useful to show available devices)

DBCSR is tested against GNU and Intel compilers on Linux systems, and GNU compiler on MacOS systems. See a list of supported compilers here.


Download either a release tarball or clone the latest version from Git using:

git clone --recursive


DBCSR can be compiled in four main variants: * Serial, i.e., no OpenMP and no MPI * OpenMP * MPI * OpenMP+MPI In addition, the variants can support accelerators.

Run inside the dbcsr directory:

mkdir build
cd build
cmake ..

The configuration flags for the CMake command are (default first):


When providing a build of LIBXSMM, make sure the lib directory is added to the PKG_CONFIG_PATH variable prior to running cmake. For example, if LIBXSMM was checked out using Git to your home folder:

export PKG_CONFIG_PATH="${PKG_CONFIG_PATH}:${HOME}/libxsmm/lib"

CMake Build Recipes

For build recipes on different platforms, make sure to also read the CMake Build Recipes.

Building with spack

DBCSR and its dependencies can be built with the spack package manager:

spack install dbcsr +openmp +mpi +cuda cuda_arch=70
spack install dbcsr +openmp +mpi +rocm amdgpu_target=gfx906
spack install dbcsr +openmp +mpi +opencl ^cuda

See spack info dbcsr for all supported variants.

C/C++ Interface

If MPI support is enabled (the default), the C API is automatically built.