GRAND-LARGE - 2012 - Annual activity report

GRAND-LARGE

GRAND-LARGE - 2012

Project-Team Grand-large

Members

Overall Objectives

Grand-Large General Objectives

Scientific Foundations

Large Scale Distributed Systems (LSDS)
Volatility and Reliability Processing
Parallel Programming on Peer-to-Peer Platforms (P5)
Methodology for Large Scale Distributed Systems
High Performance Scientific Computing

Software

APMC-CA
YML
The Scientific Programming InterNet (SPIN)
V-DS
PVC: Private Virtual Cluster
OpenWP
Parallel solvers for solving linear systems of equations
OpenScop
Clay
CALU for multicore architectures
MIDAPACK for CMB data analysis
Fast linear system solvers in public domain libraries
cTuning: Repository and Tools for Collective Characterization and Optimization of Computing Systems

New Results

Communication avoiding algorithms for linear algebra
Preconditioning techniques for solving large systems of equations
MIcrowave Data Analysis for petaScale computers
Innovative linear system solvers for hybrid multicore/GPU architectures
MILEPOST GCC: machine learning enabled self-tuning compiler
Loop Transformations: Convexity, Pruning and Optimization
Non-self-stabilizing and self-stabilizing gathering in networks of mobile agents–the notion of speed
Making Population Protocols Self-stabilizing
Self-stabilizing synchronization in population protocols with cover times
Impossibility of consensus for population protocol with cover times
Routing and synchronization in large scale networks of very cheap mobile sensors
Self-Stabilizing Control Infrastructure for HPC
Large Scale Peer to Peer Performance Evaluations
High Performance Linear Algebra on the Grid
Emulation of Volatile Systems
Exascale Systems

Partnerships and Cooperations

Regional Initiatives
International Initiatives
European Initiatives
International Research Visitors

Dissemination

Scientific Animation
Teaching - Supervision - Juries
Popularization

Bibliography

Major publications
Publications of the year
References in notes

Previous |

Home | Next next

Section: Software

CALU for multicore architectures

Participant : Laura GRIGORI [correspondant] .

The communication avoiding algorithms are implemented in the form of a portable library. In its current form, this library is designed for multicore architectures and uses a hybrid scheduling technique that exploits well the data locality and can adapt to dynamic changes in the machine. The library will be publicly available since February 2012.

Version: 1.0

Previous |

Home | Next next