diff options
author | Victor Rodriguez <victor.rodriguez.bahena@intel.com> | 2017-12-21 17:21:14 +0000 |
---|---|---|
committer | Victor Rodriguez <victor.rodriguez.bahena@intel.com> | 2017-12-23 15:37:28 +0000 |
commit | bcf949bc3dfb873c09a37dbcfaaf018124d5517d (patch) | |
tree | 1bd1045a42e7acd91759ec426fd640044f09aeb7 /doc | |
parent | 0225771c13e2cf54dc3dd1ff4d39af6a78fde9d8 (diff) | |
download | python-numpy-bcf949bc3dfb873c09a37dbcfaaf018124d5517d.tar.gz python-numpy-bcf949bc3dfb873c09a37dbcfaaf018124d5517d.tar.bz2 python-numpy-bcf949bc3dfb873c09a37dbcfaaf018124d5517d.zip |
ENH: Enable AVX2/AVX512 support to numpy #10251
This patch enables AVX2/AVX-512F instructions and distuils flags to
maximise the use of IA technology such as Haswell and Skylake platformns
on math functions of numpy
Signed-off-by: Arjan van de Ven arjan@linux.intel.com
Signed-off-by: William Douglas william.douglas@intel.com
Signed-off-by: Victor Rodriguez victor.rodriguez.bahena@intel.com
Diffstat (limited to 'doc')
-rw-r--r-- | doc/release/1.15.0-notes.rst | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/doc/release/1.15.0-notes.rst b/doc/release/1.15.0-notes.rst index 8f2de1fda..efcf9ac8d 100644 --- a/doc/release/1.15.0-notes.rst +++ b/doc/release/1.15.0-notes.rst @@ -60,6 +60,11 @@ Improvements This means it takes all the same arguments, making more code written for ndarray work for masked array too. +Enable AVX2/AVX512 at compile time +------------------------------------------------------------- +Change to simd.inc.src to use AVX2 or AVX512 at compile time. Solving the gap +that if compile numpy for avx2 (or 512) with -march=native, still get the SSE +code for the simd functions even though rest of the code gets AVX2. Changes ======= |