Skip to content

Latest commit

 

History

History
67 lines (48 loc) · 2.59 KB

CHANGELOG.md

File metadata and controls

67 lines (48 loc) · 2.59 KB
Raw

Changelog

All notable changes to this project are documented in this file. Changes format: version name and date, minor notes, new features, fixes, what removed.

Releases are available at link.

v1.2.0 - June 11, 2021

Summary

Python-package update: support backend type and memory type selection within package by environment variables. Update docs, add performance comparison section.

New features

  • Backend type config in pycubool
  • GPU memory type config in pycubool

v1.1.0 - May 8, 2021

Summary

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.

New features

  • Sparse vector support
  • Vector implementation for Cuda backend
  • Vector implementation for Sequential backend
  • Vector creation (empty, from data, with random data)
  • Matrix-vector operations (matrix-vector and vector-matrix multiplication)
  • Vector-vector operations (element-wise addition)
  • Vector data extraction (as list of indices)
  • Vector syntax sugar (pretty string printing, slicing, iterating through non-zero indices)
  • Matrix operations (extract row or matrix column as sparse vector, reduce matrix (optionally transposed) to vector)

Deprecated

  • Matrix reduce to matrix (compatibility feature)

v1.0.0 - Apr 2, 2021

Summary

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.

New features

  • Cuda backend
  • Sequential (CPU) backend
  • Sparse matrix support
  • Matrix creation (empty, from data, with random data)
  • Matrix-matrix operations (multiplication, element-wise addition, kronecker product)
  • Matrix operations (equality, transpose, reduce to vector, extract sub-matrix)
  • Matrix data extraction (as lists, as list of pairs)
  • Matrix syntax sugar (pretty string printing, slicing, iterating through non-zero values)
  • IO (import/export matrix from/to .mtx file format)
  • GraphViz (export single matrix or set of matrices as a graph with custom color and label settings)
  • Debug (matrix string debug markers, logging, operations time profiling)