Polaris Objectives To advance the state of the art in automatic program parallelization. To create and maintain an optimizing compiler that represents this state of the art and that can be used as a solid infrastructure by compiler research and development groups.

Overview

The Polaris compiler will include all optimization techniques necessary to transform a given sequential program into a form that runs efficiently on the target machine. This includes techniques such as automatic detection of parallelism and distribution of data. The intended target machines are high-performance parallel computers with a global address space as well as traditional shared-memory multiprocessors.

Polaris is developed under support by the Advanced Research Projects Agency (ARPA). The project involves research groups at both Purdue University and the University of Illinois at Urbana-Champaign.

Contents of the Purdue Polaris Web Site

Purdue Polaris Download



Polaris Publications at Purdue



Using Polaris at ECN



Polaris Demonstrations



Netcare: Try out Polaris on the Web!

Members of the Purdue Polaris Research Group

Rudolf Eigenmann Ph.D.

Brian Armstrong
Seung-Jai Min
Zhelong Pan

---

For an overall description of the project and a list of additional projects see the Polaris home page at the University of Illinois.