Research Activities

Research

Research activities focus on complex computing systems (typically multiprocessors) and on the development of efficient methodologies for the solution of application problems requiring high-performance computing. Currently, investigations pursue the following main research lines.


1. Models and algorithmic methodologies for the development of portable software on high-performance computing platforms with applications to autonomic computing (the IBM launched initiative aimed at reducing human intervention in the management of computing systems). Special attention is devoted to the performance optimization of parallel systems with hierarchical memory.


2. Design and development of optimized applications for science and engineering (e.g., high-energy physics, astronomy, structural engineering). In particular, the group features high expertise in the field of finite elements simulations as witnessed by the release of the software package FEMS. The group is also responsible for the development of ?psort?, the fastest sorting software for PC machines (winner of the Pennysort benchmark for years 2008 and 2009).


3. Development of support methodologies for global computing (computing on distributed heterogeneous large-scale platforms, typically internet-based). The group concentrates on the development of efficient protocols for enforcing connectivity in ad-hoc and mobile networks and on the deployment of p2p systems, in particular, with respect to microbenchmarking techniques, synchronization protocols, data structures for efficient search and credit reputation mechanisms for optimal resource allocations.


4. Extraction of significant information (e.g., unexpected or overrepresented patterns) from large unstructured datasets with applications to business intelligence (market-basket analysis) and genomics (motif search in DNA sequences).