Section: Partnerships and Cooperations

National Initiatives

ANR PetaQCD 01-2009/10-2012

Participants : Junjie Lai, André Seznec.

Simulation of Lattice QCD is a challenging computational problem that requires very high performance exceeding sustained Petaflops/s. The ANR PetaQCD project combines research groups from computer science, physics and two SMEs (CAPS Entreprise, Kerlabs) to address the challenges of the design of LQCD oriented supercomputer.

ANR W-SEPT

Participants : Hanbing Li, Isabelle Puaut, Erven Rohou.

Critical embedded systems are generally composed of repetitive tasks that must meet drastic timing constraints, such as termination deadlines. Providing an upper bound of the worst-case execution time (WCET) of such tasks at design time is thus necessary to prove the correctness of the system. Static WCET estimation methods, although safe, may produce largely over-estimated values. The objective of the project is to produce tighter WCET estimates by discovering and transforming flow information at all levels of the software design process, from high level-design models (e.g. Scade, Simulink) down to binary code. The ANR W-SEPT project partners are Verimag Grenoble, IRIT Toulouse, Inria Rennes. A case study is provided by Continental Toulouse.

Large Scale Initiative: Large scale multicore virtualization for performance scaling and portability

Participant : Erven Rohou.

An Inria Large Scale Initiative (Action d'Envergure) has been submitted and approved. It is entitled “Large scale multicore virtualization for performance scaling and portability”. Partner project-teams include: ALF, ALGORILLE, CAMUS, REGAL, RUNTIME, as well as DALI.

This project aims to build collaborative virtualization mechanisms that achieve essential tasks related to parallel execution and data management. We want to unify the analysis and transformation processes of programs and accompanying data into one unique virtual machine.

ADT PADRONE 2012-2014

Participants : Erven Rohou, Emmanuel Riou.

Computer science is driven by two major trends: on the one hand, the lifetime of applications is much larger than the lifetime of the hardware for which they are initially designed; on the other hand the diversity of computing hardware keeps increasing. The net result is that many applications are not optimized for their current executing environment. The objective of PADRONE is to design and develop a platform for re-optimization of binary executables at run-time. There are many advantages: actual hardware is known, the whole application is visible (including libraries), profiling can be collected, and source code is not necessary (interesting in the case of proprietary applications).