diff options
| author | Raghuveer Devulapalli <raghuveer.devulapalli@intel.com> | 2019-05-13 15:57:43 -0700 |
|---|---|---|
| committer | Raghuveer Devulapalli <raghuveer.devulapalli@intel.com> | 2019-05-17 18:05:20 -0700 |
| commit | a3e099495357b16489298bf2d40030b3415a14f0 (patch) | |
| tree | 6ab88f758c51c35438cc04fe30e494fe3f366b00 /numpy/core/numeric.py | |
| parent | 8a421d9374a278a1c0412cd35df20785c42cb080 (diff) | |
| download | numpy-a3e099495357b16489298bf2d40030b3415a14f0.tar.gz | |
ENH: use gather instruction in AVX exp/log instead of loadu
(1) AVX2 and AVX512F supports gather_ps instruction to load strided data
into a register. Rather than resort to scalar, these provide good
benefit when the input array is strided.
(2) Added some tests to validate AVX based algorithms for exp and log.
The tests compare the output of float32 against glibc's float64
implementation and verify maxulp error.
Diffstat (limited to 'numpy/core/numeric.py')
0 files changed, 0 insertions, 0 deletions