[Ns-developers] GSOC - Parallel Simulations | First ideas

[Mathieu Lacage]

On Tue, 2008-03-25 at 14:08 +0100, Hagen Paul Pfeifer wrote:

> Right! To bring some light in: my idea is that the introduced
> 
> parallelization
> 
> algorithm should technical not so far away from a distributed approach. The
> 
> treading via POSIX threads IS suitable! Consider this as an pro for the
> 
> threading
> 
> approach.


Regarding which technology is most appropriate to implement parallel
code on a single machine:
  - restricting ourselves to use only openmp-enabled compilers feels
really bad.
  - the MPI API is pretty horrid and it is really an extra layer on top
of the system thread API if you do not use its distributed version

Mathieu