tag:github.com,2008:https://github.com/SparseLinearAlgebra/cuBool/releases 2021-06-11T09:06:34Z tag:github.com,2008:Repository/300269874/v1.2.0 2021-06-12T08:31:21Z <p>cuBool project source code for the version 1.2.0.</p> <h3>Summary</h3> <p>Python-package update: support backend type and memory type selection within package by environment variables. Update docs, add performance comparison section.</p> <h3>New features</h3> <ul> <li>Backend type config in pycubool</li> <li>GPU memory type config in pycubool</li> </ul> EgorOrachyov tag:github.com,2008:Repository/300269874/v1.1.0 2021-05-08T14:23:51Z <p>cuBool project source code for the version 1.1.0.</p> <h3>Summary</h3> <p>Introduce sparse vector of boolean values support for Cuda and CPU computations. Added new primitive type as well as supported common operations, which involve matrix-vector operations and vector manipulation. Added new vector C API, exposed vector primitive into python-package.</p> <h3>New features</h3> <ul> <li>Sparse vector support</li> <li>Vector implementation for Cuda backend</li> <li>Vector implementation for Sequential backend</li> <li>Vector creation (empty, from data, with random data)</li> <li>Matrix-vector operations (matrix-vector and vector-matrix multiplication)</li> <li>Vector-vector operations (element-wise addition)</li> <li>Vector data extraction (as list of indices)</li> <li>Vector syntax sugar (pretty string printing, slicing, iterating through non-zero indices)</li> <li>Matrix operations (extract row or matrix column as sparse vector, reduce matrix (optionally transposed) to vector)</li> </ul> <h3>Deprecated</h3> <ul> <li>Matrix reduce to matrix (compatibility feature)</li> </ul> EgorOrachyov tag:github.com,2008:Repository/300269874/v1.0.0 2021-04-02T10:26:09Z <p>cuBool project source code for the version 1.0.0.</p> <h3>Summary</h3> <p>cuBool is a sparse linear Boolean algebra for Nvidia Cuda computations. Library provides C compatible API, the core of the library is written on C++ with CUDA C/C++ API and Thrust for Cuda related computations. Library also supports CPU backend for non-Cuda devices. Pycubool Python package is shipped with the library source code and it provides high-level safe and efficient access to the library within Python runtime.</p> <h3>Features</h3> <ul> <li>Cuda backend</li> <li>Sequential (CPU) backend</li> <li>Sparse matrix support</li> <li>Matrix creation (empty, from data, with random data)</li> <li>Matrix-matrix operations (multiplication, element-wise addition, kronecker product)</li> <li>Matrix operations (equality, transpose, reduce to vector, extract sub-matrix)</li> <li>Matrix data extraction (as lists, as list of pairs)</li> <li>Matrix syntax sugar (pretty string printing, slicing, iterating through non-zero values)</li> <li>IO (import/export matrix from/to <code>.mtx</code> file format)</li> <li>GraphViz (export single matrix or set of matrices as a graph with custom color and label settings)</li> <li>Debug (matrix string debug markers, logging, operations time profiling)</li> </ul> EgorOrachyov