Section: Partnerships and Cooperations

International Initiatives

Inria Associate Teams not involved in an Inria International Labs

PROSPIEL

• Title: Profiling and specialization for locality

• International Partner (Institution - Laboratory - Researcher):

  • Universidade Federal de Minas Gerais (Brazil) - Dpt of Computer Science - Fernando Magno Quintao Pereira

• Start year: 2015

• See also: https://team.inria.fr/alf/prospiel/

• The PROSPIEL project aims at optimizing parallel applications for high performance on new throughput-oriented architectures: GPUs and many-core processors. Traditionally, code optimization is driven by a program analysis performed either statically at compile-time, or dynamically at run-time. Static program analysis is fully reliable but often over-conservative. Dynamic analysis provides more accurate data, but faces strong execution time constraints and does not provide any guarantee. By combining profiling-guided specialization of parallel programs with runtime checks for correctness, PROSPIEL seeks to capture the advantages of both static analysis and dynamic analysis. The project relies on the polytope model, a mathematical representation for parallel loops, as a theoretical foundation. It focuses on analyzing and optimizing performance aspects that become increasingly critical on modern parallel computer architectures: locality and regularity.

Inria International Partners

Informal International Partners

The ALF project-team has informal collaborations (visits, common publications) with University of Wisconsin at Madison (Pr Wood), University of Toronto (Pr Moshovos), University of Ghent (Dr Eyerman), University of Upsalla (Pr Hagersten), University of Cyprus (Pr Sazeides), the Egyptian-Japanese University of Science and Technology (Pr Ahmed El-Mahdy).

Participation In other International Programs

UFMG Chair (Brazil)

• Program: Cátedras Francesas UFMG

• Title: A language runtime with fault-resiliency for approximate computing

• Inria principal investigator: Sylvain Collange

• International Partner (Institution - Laboratory - Researcher):

  • Universidade Federal de Minas Gerais (UFMG) - Computer Science Department - Fernando Pereira

• Duration: Sep 2015 - Oct 2015

• In this project we propose to implement fault tolerance at the runtime level within a virtual machine for a managed language. Our approach consists in developing a just-in-time compiler analysis that identifies and extracts side-effect free computations, such as pure functions, within the code. For each of these computations, an approximate implementation will be generated in addition to the regular native code. When the computation is invoked during execution, the runtime will first execute the approximate implementation. In case the quality or accuracy of the result is not sufficient at the time it is needed, the runtime will transparently re-execute the computation in exact mode.