numpy in multi-threaded application

Jan 21, 2013 at 1:55 PM

Hi,

I'm trying to use numpy library from IronPython embedded in a custom multi-threaded .Net application. When I try to execute a script that uses this library I get InsufficientMemoryException almost each time "import numpy" is called. The only situation it works correctly is when numpy is imported for the first time in application lifetime or is imported consecutively on the same thread as for the first import. Each attempt to import numpy on a new thread fails with the above exception.

Do you know if it's the expected behavior (numpy is expected to work only in single-threaded applications) or maybe there exists a solution to this problem? I'm using IronPython 2.7.3 and numpy 2.0.0-1

Coordinator
Jan 21, 2013 at 3:25 PM

You'll likely get a better answer from IronPython for this - see http://ironpython.net/support/ for the ways you can get in touch with that community.