Parallel sparse direct solvers are now able to solve efficiently
real-life three-dimensional problems with several millions of equations. 
The PaStiX solver provided a hybrid MPI-thread implementation that is
well suited for SMP nodes.
This technique allows to treat large 3D problems where the memory
overhead due to communication buffers was a bottleneck to the use of
direct solvers.
We introduce a simple way to schedule dynamically an application based
on a dependency tree to be more suitable for NUMA or multi-core
architectures.