Synthetic models of distributed-memory parallel programs

Synthesizing imperative distributed-memory implementations from functional data-parallel programs

FACULTY OF PHYSICAL SCIENCES AND ENGINEERING DEPARTMENT OF ELECTRONICS AND COMPUTER SCIENCE Doctor of Philosophy by Tristan Aubrey-Jones Distributed memory architectures such as Linux clusters have become increasingly common but remain difficult to program. We target this problem and present a novel technique to automatically generate data distribution plans, and subsequently MPI implementation...

Generating Efficient Parallel Programs for Distributed Memory Systems

Leveraging the performance of distributed and shared memory clusters in scientific computing is challenging in terms of programmability and efficiency. The dimensions of the problem are data distribution, computation distribution, efficient communications and the ease of programming. To address those dimensions in a balanced manner, we present a directive-based programming model for hybrid dist...

Procedure Call Models for Distributed Parameters in Data Parallel Programs

When a computer program invokes a procedure, both the caller and the callee must agree on how to pass the parameters into and out of the procedure. In this paper, this agreement is referred to as the procedure call model. In data parallel languages like High Performance Fortran (HPF), the procedure call model for distributed parameters can have an impact on procedure call overhead. This paper i...

Structuring parallel and distributed programs

The paper presents a structuring language, Darwin, which allows distributed and parallel programs to be structured in terms of groups of process instances which communicate by message passing. In addition to expressing static structure, Darwin can be used to express structures which change dynamically as execution progresses. The paper presents a set of examples illustrating the use of Darwin i...

Generation of Distributed Parallel Java Programs