University of Torino (UniTo)


University of Torino
Computer Science Department
Corso Svizzera, 185
10149, Torino, Italy
torino
Research group:
Parallel Computing
Director:
Prof. Marco Aldinucci
Computer Science Department
Corso Svizzera, 185
10149, Torino, Italy
alpha

Institution Background

The University of Torino is one of the oldest and largest
universities in Italy (founded in 1404), and has more than 2000 lecturers and professors. The Computer Science department coordinates the
research and teaching activity in the area of Computer Science and Information Technology. It supports one bachelor program featuring three
different curricula, one master program (four curricula) and a Ph.D. school
in computer science. The Computer Science department was founded in
1971 and nowadays has 35 professors (12 full, 23 associate), 41 permanent researchers, 22 research fellows, and 34 PhD students (http://di.unito.it/people). Their scientific activity covers a wide range of areas, such as artificial intelligence, data base and information systems, computer networks, distributed systems, security, formal methods in computing, image processing and virtual reality, performance evaluation, concurrency, parallel, distributed and high performance computing, clouds, bio-informatics, agent-oriented and service-oriented computing.

 

Research Group Background 

The Parallel Computing (Alpha http://alpha.di.unito.it) research group, located at Computer Science Department and led by Prof. Marco Aldinucci, has been recently formed and currently consists of 2 postdocs and 3 PhD students. The group focuses on research in parallel and distributed computing systems, and in particular on parallel programming models. Alpha participates in the EU FP7-NoE HiPEAC, EU FP7-STREP Paraphrase, EU FP7-STREP REPARA, and the Italian IMPACT projects. It cooperates with well-known international research centres, among the others University of Pisa, University of Catania, University of Cambridge, Italian National Research Council, HLRS Supercomputing Centre Stuttgart, and Queens University Belfast. It also has had industrial cooperation with other partners, including IBM, NVidia, and A3Cube.

 

Prof. Marco Aldinucci is the leader of the research group on Parallel Computing (Alpha) and a Principal Investigator of the NVidia CUDA research centre at University of Torino. He has authored over 110 papers and participated in over 15 research projects concerning parallel computing, autonomic computing, grid and cloud topics, including the Grid.it, CoreGRID FP6 EC-NoE, GridComp FP6 EC-STREP, BEinGRID FP6 EC-IP. He is the recipient of a HPC Advisory Council award 2011 and delivered over 15 invited talks in international venues. He has edited conference proceedings (such as IEEE PDP 2014), and special issue for several journals (such as Sage IJHPCA). He is member of HPC Advisory Council and HPC500. He participated to the design of several frameworks for parallel programming including compilers, libraries and frameworks, both in industrial and academic teams. They include ASSIST, muskel, and FastFlow programming environments, the VirtuaLinux cloud platform, and the ETSI standard Grid Component Model (GCM). He is leading the work-package on Low-level virtualization of the EC-STREP FP7 Paraphrase project, and the University of Torino unit in the EU-NoE HiPEAC and the EU-STREP REPARA projects.

 

Major Interest in Action Scientific Topics

  • Parallel programming models suitable for ultrascale; programming run-times for adapting to failures and load variability; trade-offs among programmability, performance, scalability and energy-efficiency.
  • Dynamic replication of data and/or behavior, proactive actions based on efficient and reliable fault-prediction mechanisms.
  • Monitoring systems for heterogeneous ultrascale systems (low power servers, Graphics processing units (GPUs,…); models of energy consumption as basis for ultra scale infrastructures software components.
  • Identification of a set of key characteristics to determine a priori whether applications are amenable for ultrascale computing.