Overview

The objective of this project is develop abstractions, algorithms and languages for parallelism and dynamic parallelism with applications to problems on large data sets.

Project pages

pdfs

A work-efficient algorithm for parallel unordered depth-first search

pasl

Parallel-algorithm scheduling library

chunkedseq

Chunked-sequence data structure

pbench

Parallel-program benchmarking kit

oracular scheduling

Granularity-control technique for parallel programming

Publications

See our publications on the following topics:

• parallelism,
• dynamic parallelism and on
• dynamic unstructured data sets.

People

• Umut Acar
• Vitaly Aksenov
• Arthur Charguéraud
• Mike Rainey
• Filip Sieczkowski

Collaborators

• Pramod Bhatotia
• Anna Malova
• Stefan Muller
• Ram Raghunathan
• Rodrigo Rodrigues
• Kanat Tangwongsan

Research environment

The ERC project is associated with the Inria Rocquencourt team Gallium, whose permanent members are Xavier Leroy, Luc Maranget, François Pottier, Didier Rémy, and Damien Doligez.

Funding

This project takes place in INRIA-Rocquencourt and is made possible by a generous grant from the European Union (ERC-2012-StG 308246—DeepSea).