Skip to content

Releases: tttapa/batmat

0.0.21

15 Apr 12:31

Choose a tag to compare

  • Elementwise linear algebra operations and reductions (axpy, dot products, norms, etc.)
  • SIMD select and popcount abstractions

Full Changelog: 0.0.20...0.0.21

0.0.20

14 Apr 20:02

Choose a tag to compare

  • batmat::matrix::View improvements
  • Make thread pool movable.

Full Changelog: 0.0.19...0.0.20

0.0.19

25 Mar 15:49

Choose a tag to compare

  • Add gemm_diag_sub, syrk_diag, syrk_diag_sub.

Full Changelog: 0.0.18...0.0.19

0.0.18

03 Mar 20:54

Choose a tag to compare

Full Changelog: 0.0.17...0.0.18

0.0.17

03 Mar 09:43

Choose a tag to compare

  • Support for shared builds.

Full Changelog: 0.0.16...0.0.17

0.0.16

20 Feb 18:09

Choose a tag to compare

  • Improved documentation
  • Fully support double and single precision, both using GCC's simd implementation and GSI-HPC/simd
  • Metaprogramming utilities for enabled dtypes and vector lengths (useful in test matrix generation)

Full Changelog: 0.0.15...0.0.16

0.0.15

19 Feb 13:04

Choose a tag to compare

  • Improved batmat::matrix::View and batmat::matrix::Matrix APIs
  • Replace gemm/gemv shift_B by rotate_B

Full Changelog: 0.0.14...0.0.15

0.0.14

18 Feb 09:52

Choose a tag to compare

  • Add syrk_diag_add_potrf function.

Full Changelog: 0.0.13...0.0.14

0.0.13

17 Feb 10:31

Choose a tag to compare

  • Breaking: batmat::matrix::View::data is now a member function instead of a variable, for consistency with batmat::matrix::Matrix.
  • Free functions data(X), depth(X), rows(X), cols(X), outer_stride(X) for various matrix and view types.
  • The default number of columns of a view initialized with zero rows is now zero instead of one.
  • Improved support for simdify() with multiple batches.
  • Added documentation (https://tttapa.github.io/batmat).
  • Deprecated some old APIs using the global thread pool.

Full Changelog: 0.0.12...0.0.13

0.0.12

03 Feb 12:35

Choose a tag to compare

  • Instantiate more gemv variants without masks

Full Changelog: 0.0.11...0.0.12