Geek Answers Handbook

Odorless assimilate with numpy / scipy

Has anyone used a bottomless swallow with numpy / scipy for numerical / scientific applications? Is it significantly faster in your experience? Any opinions would be wonderful.

+7
optimization python numpy scipy unladen-swallow
source share
3 answers

No one has extensive experience working with Unladen Swallow (other than developers), so it will be difficult to find many people who can discuss this. Also, with the talk of merging Unladen Swallow (which is built using LLVM ) with CPython runtime, everything will be a bit of a moving target until everything becomes more stable.

There are tags for Unladen Swallow, but numpy and scipy are not included. As the developers themselves explain : "... the performance of expansion modules such as numpy is uninteresting because C has numpy core functions.

In short, if you write good code for numpy and scipy , your code will not run “much faster” under Unladen Swallow, since it already runs under the virtual machine level. If you write bad code for numpy and scipy , you need to fix your code and then return to the first sentence.

+5
source share

It should be faster. I did not test it myself, but I just returned from pycon, and they talked about an unemployed swallow in which they mentioned performance improvements with numpy and other packages. You can watch the conversation here .

+1
source share

There is no answer in the question:

 Total runtime = python + numpy + interface, cpython/unladenswallow + mostlyC + interface.

Without real data on how these 3 splits are 20 70 10, 40 40 20? and that for> 1 test,
there is no way to say which way up.

+1
source share

More articles:


All Articles