2011 Poster Sessions : A Domain Specific Approach to Heterogeneous Parallel Computing

Student Name : Arvind Sujeeth
Advisor : Oyekunle Olukotun
Research Areas: Computer Systems
As heterogeneous parallel systems become dominant, application developers are being forced to turn to an incompatible mix of low level programming models (e.g. OpenMP, MPI, CUDA, OpenCL). However, these models do little to shield developers from the difficult problems of parallelization, data decomposition and machine specific details. Ordinary programmers are having a difficult time using these programming models effectively. To provide a programming model that addresses the productivity and performance requirements for the average programmer, we explore a domain-specific approach to heterogeneous parallel programming.

Our approach uses the new principle of language virtualization to embed domain-specific languages (DSLs) inside a host language in a form that is amenable to optimization and code generation. Our embedded DSLs are built on top of a parallel framework and runtime, called Delite, which manages data movement and parallel execution across different compute devices. We demonstrate how this organization enables simple, sequential application code to be executed in parallel on CPUs and GPUs through OptiML, a DSL for machine learning.

Arvind Sujeeth is a Ph.D. candidate working for Prof. Kunle Olukotun. He is interested in leveraging domain specific languages as a parallel programming platform. His research focuses on building reusable infrastructure for embedded parallel DSLs and applying those concepts to the development of a DSL for machine learning.