Difference between revisions of "Summer2009:Using cython"

From NAMIC Wiki
Jump to: navigation, search
 
Line 27: Line 27:
  
 
<h3>Progress</h3>
 
<h3>Progress</h3>
* Cython code for tensor and fiber tracking done
+
* Implemented Cython code for tensor and fiber tracking
 
* Performance depends mainly of the nature of the Python code - if numpy adequately used (vectorization...) - no significant difference with C performance  
 
* Performance depends mainly of the nature of the Python code - if numpy adequately used (vectorization...) - no significant difference with C performance  
  

Latest revision as of 21:17, 28 June 2009

Home < Summer2009:Using cython


Key Investigators

  • BWH: Julien de Siebenthal, Sylvain Bouix

Objective

We are investigating ways of accelerating the stochastic tractography code developed in python and part of the current version of Slicer 3.4.

Approach, Plan

Cython is used to develop a direct way of integrating native C code where python bytecode is too slow and hinders performance of the algorithm.

During the summer week, we will discuss improvements on the current implementation and possibly other approaches that could complement ours.

Progress

  • Implemented Cython code for tensor and fiber tracking
  • Performance depends mainly of the nature of the Python code - if numpy adequately used (vectorization...) - no significant difference with C performance


References