[mlpack-svn] [MLPACK] #179: Design parallelization framework
MLPACK Trac
trac at coffeetalk-1.cc.gatech.edu
Tue Dec 20 23:19:14 EST 2011
#179: Design parallelization framework
----------------------+-----------------------------------------------------
Reporter: rcurtin | Owner:
Type: wishlist | Status: new
Priority: major | Milestone: MLPACK 2.0.0
Component: MLPACK | Keywords: parallelization
Blocking: | Blocked By:
----------------------+-----------------------------------------------------
This is a master ticket designed to encompass the big 2.0 feature:
parallelization. Before this ticket is done:
* A framework for parallelization must be found
* This framework must be applied to each method
The framework we use should, ideally, make parallelization of the
algorithm easy and fairly transparent to the developer. They should not
have to write workers, threads, callbacks, so on and so forth; they should
be able to simply specify somehow that some particular loop could be
parallelized... or something.
I seem to understand OpenMP is a decent framework for something like this,
but I know so little about parallelization that I am not certain if what
we are looking for is really possible at the moment.
--
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/179>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed by the FASTLAB at Georgia Tech under Dr. Alex Gray.
More information about the mlpack-svn
mailing list