[NumPy-Tickets] [NumPy] #2219: Basic array-like function like sin; cos ... are too slow with numpy-MKL

NumPy Trac numpy-tickets@scipy....
Fri Sep 28 03:15:50 CDT 2012


#2219: Basic array-like function like sin; cos ... are too slow with numpy-MKL
-------------------------+--------------------------------------------------
 Reporter:  bred         |       Owner:  somebody   
     Type:  enhancement  |      Status:  new        
 Priority:  normal       |   Milestone:  Unscheduled
Component:  numpy.core   |     Version:  1.6.2      
 Keywords:               |  
-------------------------+--------------------------------------------------
 If we perform a sin function on a very big array the computation is very
 slow respect to the real possibility offered by MKL.

 Example:

 {{{
 x = np.random.rand(1e7)
 y = np.sin(x) # is too slow !!! and it's not parallel!
 }}}

 I've seen that mkl offers the functions for solving these operation in a
 performant manner.

 Can you implement these functions in munpy in a better way?

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/2219>
NumPy <http://projects.scipy.org/numpy>
My example project


More information about the NumPy-Tickets mailing list