Saturday, May 11, 2013

Numpy Status Update

Hello Everyone,

I've started to work on NumPyPy since the end of April and here is a short update :

  • I implemented pickling support on ndarrays and dtypes, it will be compatible with numpy's pickling protocol when the "numpypy" module will be renamed to "numpy".
  • I am now working on subarrays.

I would also like to thank everyone who donated and allowed me to work on this.

Cheers,
Romain Guillebert

6 comments:

Anonymous said...

No, thank you! Cannot wait till the day PyPy fully supports NumPy.

Anonymous said...

I second the anonymous comment above. The day PyPy fully supports NumPy is the day I switch from CPython.

Paul Jaros said...

Aww... Anonymous.

@Romain Guillebert Thank you for the hard work you are putting into it. I will be testing my code with the current release.

Anonymous said...

This (and to a lesser extent Python 3 support) is the only thing holding me back from switching to PyPy for all of my python programming. Thank you very much for this fantastic project!

Paul Jaros said...

Results from running my own little Benchmark: Labyrinth Generator
Array Size: 77711x711:

C-Code:
4.45 Seconds, ~50M Memory Usage.

Pypy with standard List:
14.5 Seconds, ~750M Memory Usage.

Pypy with Numpypy:
11.0 Seconds, ~78M Memory Usage.

Pretty impressive if you ask me. Older Numpypy where about as fast as the standard List. Also Pypy is approaching C-Performance with bigger steps than I dared hoping for.

CPython Benchmark intentionally left out... it takes ages.

Robert Zaremba said...

It's great to see a progress in important libraries support.

Speed is important, but when we get acceptable speed then library support is what we need.