dbcsr ☆

« Back to VersTracker

Description:
Distributed Block Compressed Sparse Row matrix library

Type: Formula | Tracked Since: Jan 28, 2026

Links: Homepage | formulae.brew.sh

Category: Developer tools

Tags: hpc sparse-matrix quantum-chemistry linear-algebra cp2k mpi

Install: brew install dbcsr

About:
DBCSR (Distributed Block Compressed Sparse Row) is a high-performance library for sparse matrix-matrix multiplication, specifically optimized for linear scaling electronic structure calculations. It provides efficient handling of block-sparse matrices by distributing data across MPI processes and leveraging GPU acceleration. Its main value is enabling large-scale quantum chemistry and material science simulations in software like CP2K.

Key Features:

Distributed, MPI-parallel sparse matrix operations
Optimized for block-sparse matrices common in quantum chemistry
GPU acceleration support (CUDA, HIP)
Efficient communication patterns for scalability
Core linear algebra backend for the CP2K materials science suite

Use Cases:

Large-scale electronic structure calculations (e.g., with CP2K)
High-performance computing research in materials science
Developing scalable scientific applications requiring sparse linear algebra

Alternatives:

PETSc – More general-purpose numerical toolkit; DBCSR is specialized for block-sparse quantum chemistry.
Intel MKL – Optimized dense and sparse BLAS for shared memory; DBCSR is MPI-distributed and block-sparse focused.

Version History

Detected	Version	Rev	Change	Commit
Jan 28, 2026 11:35pm		0	META	02944691