Research Interests

1. enabling compilers to treat MPI as a language extension, rather than a library, so that optimizations can be performed on the MPI calls themselves, as well as code with dependencies to them.
2. exposing more infomation about the data exchange to the application layer, so that additional sources of overhead (such as protocol handshakes and memory registration costs) can be further optimized.

Selected Publications

• A. Danalis, L. Pollock, M. Swany, and J. Cavazos, "MPI-aware Compiler Optimizations for Improving Communication-Computation Overlap", In ICS '09: Proceedings of the 23rd annual International Conference on Supercomputing, Yorktown Heights, NY, USA, June 2009. [ Full text: pdf ]
• A. Danalis, L. Pollock and M. Swany, "Gravel: a communication library to fast path MPI", In EuroPVM/MPI, Dublin, Ireland, September 2008. [ Full text: pdf ]
• A. Danalis, L. Pollock, M. Swany and J. Cavazos, "Implementing an Open64-based Tool for Improving the Performance of MPI Programs", In The Open64 Workshop, in conjunction with IEEE/ACM International Symposium on Code Generation and Optimization (CGO) 2008, Boston, MA, USA [ Full text: pdf ]
• A. Danalis, L. Pollock and M. Swany, "Automatic MPI application transformation with ASPhALT", In Workshop on Performance Optimization for High-Level Languages and Libraries (POHLL 2007), in conjunction with the 21th International Parallel and Distributed Processing Symposium (IPDPS 2007), Long Beach, California USA, Mar. 2007. [ Full text: pdf ]
• A. Danalis, K. Y. Kim, L. Pollock and M. Swany, "Transformations to Parallel Codes for Communication-Computation Overlap", In Proceedings of the ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis (SC05), Seattle, WA, November, 2005. [ Full text: pdf ]