My Phd. thesis is about ASIP processor instruction set extension. I'm mostly interested by following topics:
Graph theory algorithms such as pattern generation and graph covering. In my work, covering is expressed as a Constraint Programming problem that models both selection and scheduling under architectural constraints into a single optimization space.
Automatic Parallelization. I work on the extraction of custom instructions that can be parallelized at a loop level. I'm using the powerful polyhedral abstraction to find a good and legal transformation of the iteration domain.
Parallel Architectures.The target of our ASIP flow is a processor coupled to an extension. It provides a level of potential parallelism between the core/extension but also in the extension itself that can contains multiple parallel datapaths.
Constraint Programming (CP).The CP solvers offers a high level of expressivity and an effective and customizable solution space exploration through advanced filtering techniques (called consistency) adapted to each global constraints.
Model Driven Engineering (MDE). Most of my developed softwares are based on Model Driven Engineering (specifically EMF). It provides access to all facilities offered by EMF (such as Model to Model transformations, Model To Text and Text to Model) that are especially useful in the context of a fast optimization compiler design used to validate experimental approaches of an ASIP flow.