2009 Poster Sessions : DELITE: Simplifying Parallel Programming with DSLs

Student Name : Nathan Bronson, Hassan Chafi
Advisor : Oyekunle Olukotun
Research Areas: Computer Systems
Future increases in processor performance will come from multiple cores per chip. This means parallel programming will become much more widespread, as it is the only way to harness this extra potential.
Existing techniques such as threads+locks and explicit message passing dramatically increase the complexity and fragility of software, which is a problem.

The Pervasive Parallelism Lab is using Domain Specific Languages to harness parallelism while exposing a simple programming model. In addition, DSLs enable us to use high-level knowledge to perform optimizations that are not feasible to discover using traditional compiler or runtime techniques. DELITE is a framework that combines a run-time library and VM extensions. The framework supports embedding implicitly parallel DSLs in Scala with dynamic and static optimizations that take advantage of the particular domain expressed by a DSL. DELITE also provides tools for dynamic verification of data race freedom.

Hassan Chafi is a third year PhD candidate in the Computer Science dept at Stanford University. His research interests include computer architectures and parallel programming models.

Nathan Bronson is a third year PhD student in the Computer Science Department at Stanford. He is interested in the formalization and efficient implementation of high-level programming abstractions, especially in the areas of parallelism and concurrency. In his free time he enjoys rock climbing and mountaineering.