[mlpack-git] [mlpack/mlpack] Design parallelization framework (#173)

[Ryan Curtin]

I think that OpenMP is really the best way to go here, so one way to do this is just "pick a method and OpenMP-ize it".  If you're interested in finding other parallelism solutions, you can explore libraries and propose solutions, but one of the biggest keys is that the code remains easy to maintain and easy to understand.  i.e., MPI code can be really ugly so we should really try to avoid that since it would make the algorithms very hard to read, for instance.  (The same is true of stuff like OpenCL and CUDA unfortunately.)

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/173#issuecomment-253206021
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20161012/a503b0bd/attachment.html>