Concurrent and parallel programming are becoming indispensable for
exploiting modern hardware. Because possible speed increases of single
processors have reached their limit, increasing transistor count will
yield more, but not necessarily faster cores, and this for the
foreseeable future. This means that, from now on, parallelism in
software will have to double every 18 months to keep up with
hardware. This problem has been identified as the ``Popular Parallel
Programming'' grand challenge by the computer architecture community.

The proposed project will research new ways to solve this challenge.
We start with a set of domain-specific languages which naturally admit
a high degree of parallelism. The domain specific languages are
integrated in a common host language using polymorphic language
embeddings. Such embeddings provide a high degree of abstraction,
which gives considerable freedom in their actual representation and
implementation. The new direction taken by this proposal is to combine
polymorphic embeddings with optimizing rewritings in a staged
compilation process. This can lead to highly parallel and efficient
implementations on a variety of heterogeneous hardware.