Blaze is an open-source, high-performance C++ math library for dense and sparse arithmetic. With its state-of-the-art Smart Expression Template implementation Blaze combines the elegance and ease of use of a domain-specific language with HPC-grade performance, making it one of the most intuitive and fastest C++ math libraries available.
The Blaze library offers ...
- ... high performance through the integration of BLAS libraries and manually tuned HPC math kernels
- ... vectorization by SSE, SSE2, SSE3, SSSE3, SSE4, AVX, AVX2, AVX-512, FMA, and SVML
- ... parallel execution by OpenMP, C++11 threads and Boost threads
- ... the intuitive and easy to use API of a domain specific language
- ... unified arithmetic with dense and sparse vectors and matrices
- ... thoroughly tested matrix and vector arithmetic
- ... completely portable, high quality C++ source code
BlazeIterative: A collection of iterative solvers (CG, BiCGSTAB, ...) for the Blaze library (Tyler Olsen)
15.9.2017: Attention early adopters: We have spent the last weeks to update all available views. One highlight of this refactoring is that it is now possible to provide the arguments of views as template arguments! As an example, consider the following two code snippets that demonstrate the setup of subvectors and submatrices, respectively:
blaze::DynamicVector<double,blaze::rowVector> x; // ... Resizing and initialization // Create a subvector from index 4 with a size of 12 (i.e. in the range [4..15]) (compile time arguments) auto sv1 = subvector<4UL,12UL>( x ); // Create a subvector from index 8 with a size of 16 (i.e. in the range [8..23]) (runtime arguments) auto sv2 = subvector( x, 8UL, 16UL );
blaze::DynamicMatrix<double,blaze::rowMajor> A; // ... Resizing and initialization // Creating a dense submatrix of size 4x8, starting in row 3 and column 0 (compile time arguments) auto sm1 = submatrix<3UL,0UL,4UL,8UL>( A ); // Creating a dense submatrix of size 8x16, starting in row 0 and column 4 (runtime arguments) auto sm2 = submatrix( A, 0UL, 4UL, 8UL, 16UL );
The same works with rows, columns and bands. We are happy to share this new feature with you and welcome any kind of feedback you might have at this time.
18.8.2017: Today, after nearly six month of hard work, we officially release Blaze 3.2! This version is dedicated to several of the most anticipated features: Blaze finally provides CMake support and an advanced configuration system, which allows you to configure each single detail of Blaze from the command line. Additionally, Blaze finally provides complete support of AVX-512 and introduces the
IdentityMatrix class. Furthermore, Blaze finally features binary custom operations and the componentwise matrix multiplication (Schur Product). Of course we have also spent time on a lot of smaller features and tweaked countless little details. We hope you enjoy this new release and the ton of new features.
We don't want to miss the opportunity to thank our many contributors: Thanks a lot for your efforts to make Blaze a better library!
3.7.2017: We are proud to announce the first of hopefully many Blaze projects: BlazeIterative. Check out this collection of iterative solves, which neatly integrate with the Blaze library.
20.2.2017: We are happy to announce that there is now a port of the Blaze library for the R language available: RcppBlaze.
Wiki: Table of Contents
- Configuration and Installation
- Getting Started
- Arithmetic Operations
- Shared-Memory Parallelization
- BLAS Functions
- LAPACK Functions
- Block Vectors and Matrices
- Intra-Statement Optimization
- Blazemark: The Blaze Benchmark Suite
- Benchmarks/Performance Results
The Blaze library is licensed under the New (Revised) BSD license. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
- Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
- Neither the names of the Blaze development group nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Blaze supports the C++14 standard and is compatible with a wide range of C++ compilers. In fact, Blaze is constantly tested with the GNU compiler collection (version 4.9 through 7.1), the Intel C++ compiler (16.0), the Clang compiler (version 3.7 through 4.0), and Visual C++ 2015 and 2017 (Win64 only). Other compilers are not explicitly tested, but might work with a high probability.
If you are looking for a C++98 compatible math library you might consider using an older release of Blaze. Until the release 2.6 Blaze was written in C++-98 and constantly tested with the GNU compiler collection (version 4.5 through 5.0), the Intel C++ compiler (12.1, 13.1, 14.0, 15.0), the Clang compiler (version 3.4 through 3.7), and Visual C++ 2010, 2012, 2013, and 2015 (Win64 only).
- K. Iglberger, G. Hager, J. Treibig, and U. Rüde: Expression Templates Revisited: A Performance Analysis of Current Methodologies (Download). SIAM Journal on Scientific Computing, 34(2): C42--C69, 2012
- K. Iglberger, G. Hager, J. Treibig, and U. Rüde: High Performance Smart Expression Template Math Libraries (Download). Proceedings of the 2nd International Workshop on New Algorithms and Programming Models for the Manycore Era (APMM 2012) at HPCS 2012
Klaus Iglberger -- Project initiator and main developer
Georg Hager -- Performance analysis and optimization
Christian Godenschwager -- Visual Studio 2010/2012/2013/2015 bug fixes and testing
Tobias Scharpff -- Sparse matrix multiplication algorithms
byzhang -- Bug fixes
Emerson Ferreira -- Bug fixes
Fabien Péan -- CMake support
Denis Demidov -- Export CMake package configuration
Jannik Schürg -- AVX-512 support and cache size detection for macOS in CMake
Marcin Copik -- CMake fixes
Hartmut Kaiser -- HPX backend