diff options
| author | Nathaniel J. Smith <njs@pobox.com> | 2016-01-06 23:28:32 +0000 |
|---|---|---|
| committer | Nathaniel J. Smith <njs@pobox.com> | 2016-01-06 23:28:32 +0000 |
| commit | 25c8d1c2d3590c63645d5267807b15be9abd00e4 (patch) | |
| tree | e2c0ae8bba755913f67d868cc26f452d85c709dd /doc | |
| parent | a7377d881469bb45ca4710df2a46c719aa9a80e3 (diff) | |
| parent | 8d8a74d8b2f86d2548a04565744ab122537a4f62 (diff) | |
| download | numpy-25c8d1c2d3590c63645d5267807b15be9abd00e4.tar.gz | |
Merge pull request #6932 from jakirkham/opt_dot_trans
ENH: Use `syrk` to compute certain dot products more quickly and accurately
Diffstat (limited to 'doc')
| -rw-r--r-- | doc/release/1.11.0-notes.rst | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/doc/release/1.11.0-notes.rst b/doc/release/1.11.0-notes.rst index 54b6874cc..099ce3976 100644 --- a/doc/release/1.11.0-notes.rst +++ b/doc/release/1.11.0-notes.rst @@ -130,6 +130,14 @@ useless computations when printing a masked array. The function now uses the fallocate system call to reserve sufficient diskspace on filesystems that support it. +``np.dot`` optimized for operations of the form ``A.T @ A`` and ``A @ A.T`` +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Previously, ``gemm`` BLAS operations were used for all matrix products. Now, +if the matrix product is between a matrix and its transpose, it will use +``syrk`` BLAS operations for a performance boost. + +**Note:** Requires the transposed and non-transposed matrices to share data. + Changes ======= |