[pythran] Re: Moving to xsimd

  • From: Neal Becker <ndbecker2@xxxxxxxxx>
  • To: pythran@xxxxxxxxxxxxx
  • Date: Fri, 2 Nov 2018 18:43:32 -0400

Do you really want to find euclidean_distance as hypot(x)**2?  Isn't it
faster to do real(x)**2 + imag(x)**2?
Might not be standard conforming though.

On Fri, Nov 2, 2018 at 5:33 PM serge guelton <
serge.guelton@xxxxxxxxxxxxxxxxxxx> wrote:

Hi pythraners,

a blogpost to describe the move to xsimd as vectorization engine:


http://serge-sans-paille.github.io/pythran-stories/bye-bye-boostsimd-welcome-xsimd.html

happy reading!


Other related posts: