Équipe Associée — PolyFlow — Associate Team
Polyhedral Compilation and Optimization for Data-Flow Programming Languages
Indian Institute of Science (IISc), Department of Computer Science and Automation (CSA), Bangalore, India
INRIA Paris - Rocquencourt and École Normale Supérieure, PARKAS project-team, Paris, France

Overview

Principal Investigators

Uday Kumar Reddy Bondhugula (IISc) and Albert Cohen (INRIA)

Acknowledgments

PolyFlow was first funded as an Inria Associate Team for 3 years (January 2013 to December 2015), and then renewed for 2016 to 2019. Complementary support is provided from the partner's research grants and by the hosting institutions (IISc, ENS).

Ambitions

The objective of the associate team is to foster collaborations on fundamental and applied research. It also supports training sessions, exchange of undergraduate and master students, and highlighting opportunities in the partners' research, education and economic environments.

Research Statement

Polyhedral techniques for program transformation are now used in several proprietary and open source compilers. However, most of the research on polyhedral compilation has focused on imperative languages, where computation is specified in terms of computational statements within nested loops and control structures. Graphical data-flow languages, where there is no notion of statements or a schedule specifying their relative execution order, have so far not been studied using a powerful transformation or optimization approach. These languages are extremely popular in the system analysis, modeling and design of embedded reactive control applications. They also underline the construction of domain-specific languages and compiler intermediate representations. The execution semantics of data-flow languages impose a different set of challenges for compilation and optimization. We are studying techniques enabling the extraction of a polyhedral representation from data-flow programs, to transform them with the goal of generating memory-efficient and high-performance code for modern architectures.

The research conducted in PolyFlow covers both fundamental and applied aspects. The partners also emphasize the development of solid research tools. The associate team will facilitate their dissemination as free software and their exploitation through industrial collaborations.

Recent and Ongoing Work

Several collaborations have been initiated.

Events

Publications

PolyFlow Joint Publications

Earlier Joint Publications