[NumPy-Tickets] [NumPy] #2065: Poor ndarray.take performance on Fortran order arrays

NumPy Trac numpy-tickets@scipy....
Mon Mar 12 13:13:24 CDT 2012


#2065: Poor ndarray.take performance on Fortran order arrays
--------------------+-------------------------------------------------------
 Reporter:  wesm    |       Owner:  somebody   
     Type:  defect  |      Status:  new        
 Priority:  normal  |   Milestone:  Unscheduled
Component:  Other   |     Version:  1.6.1      
 Keywords:          |  
--------------------+-------------------------------------------------------

Comment(by mwiebe):

 Confirmed. ndarray.take needs to be rewritten to be CPU cache-aware. In
 1.6, element-wise ufuncs got this treatment. In 1.7, the ufunc reduce
 method has received this treatment. Making NumPy use more algorithms that
 respect the CPU cache should continue.

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/2065#comment:1>
NumPy <http://projects.scipy.org/numpy>
My example project


More information about the NumPy-Tickets mailing list