[Numpy-discussion] NumPy re-factoring project
Thu Jun 10 15:28:28 CDT 2010
Thu, 10 Jun 2010 18:48:04 +0200, Sturla Molden wrote:
> 5. Allow OpenMP pragmas in the core. If arrays are above a certain size,
> it should switch to multi-threading.
Some places where Openmp could probably help are in the inner ufunc
loops. However, improving the memory efficiency of the data access
pattern is another low-hanging fruit for multidimensional arrays.
I'm not sure how Openmp plays together with thread-safety; especially if
used in outer constructs. Whether this is possible to do easily is not so
clear, as Numpy uses e.g. iterator and loop objects, which cannot
probably be shared between different openmp threads. So one would have to
do some extra work to parallelize these parts manually.
But probably if multithreading is desired, openmp sounds like the way to
More information about the NumPy-Discussion