High Level Parallel Programming and Parallelization

Selected Papers | Software
The problems involved in developing efficient and correct parallel programs have proved much harder than those in developing efficient sequential ones, both for programmers and for compilers. We have made the first attempt to construct a calculational framework for parallelization of sequential programs, including a novel inductive synthesis lemma and a calculation algorithm for parallelization of general recursive functions. Being constructive, this work has proved to be useful both in parallelizing compiler and in parallel programming.

We are now applying the technique to design and implementation of skeletal parallel programming environment called SkeTo, which supports efficient parallel programming with a fixed set of parallel skeletons. It can deal with a large class of problems on lists, 2-dimensional arrays, and irregular trees. Here is video (in Japanese) about part of the project.

We are also collaborating with Dr. Guy Steele and his team at Sun Microsystems Laboratories to develop a library for skeletal parallel programming in Fortress (a programming language designed by Sun for high-performance computing). Here contains more information about the joint project.


Selected Papers

More papers are available here.


Software


Maintained by Zhenjiang Hu. Last modified in 2006.