diff options
| author | Matti Picus <matti.picus@gmail.com> | 2020-04-23 08:14:09 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2020-04-23 08:14:09 +0300 |
| commit | 96c21f037679c0766142fa8dc3fb143b5927439f (patch) | |
| tree | b6185e2a51a673bf8bebb68038b962ccf3688824 /doc | |
| parent | 517f53dc91164f4249de9dbaf31be65df02ffde7 (diff) | |
| parent | 6bd0cb90a607f46c6c45fbc36d280dd25e23b4a3 (diff) | |
| download | numpy-96c21f037679c0766142fa8dc3fb143b5927439f.tar.gz | |
Merge pull request #15648 from xiegengxin/avx512-exp-float64
MAINT: AVX512 implementation with intrinsic for float64 input np.exp()
Diffstat (limited to 'doc')
| -rw-r--r-- | doc/release/upcoming_changes/15648.improvement.rst | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/doc/release/upcoming_changes/15648.improvement.rst b/doc/release/upcoming_changes/15648.improvement.rst new file mode 100644 index 000000000..2dbfa9f99 --- /dev/null +++ b/doc/release/upcoming_changes/15648.improvement.rst @@ -0,0 +1,7 @@ +Use AVX512 intrinsic to implement ``np.exp`` when input is ``np.float64`` +-------------------------------------------------------------------------- +Use AVX512 intrinsic to implement ``np.exp`` when input is ``np.float64``, +which can improve the performance of ``np.exp`` with ``np.float64`` input 5-7x +faster than before. The _multiarray_umath.so module has grown about 63 KB on +linux64. + |