Sam Larsen

I'm a former graduate student in the EECS department at MIT. I worked under Professor Saman Amarasinghe in the commit group. Before coming to MIT, I was a student at the University of Utah in the CS department.

Research

My research focused on compilation for multimedia extensions. For my doctoral work, I developed selective vectorization, a technique that balances computation across a processor's scalar and vector resources. The approach leads to software pipelines with shorter schedules, and therefore, higher performance. Selective vectorization is applicable to any design that supports ILP and short-vector parallelism.

Publications

Samuel Larsen. Compilation Techniques for Short-Vector Instructions. PhD thesis, Massachusetts Institute of Technology, April 2006. (ps, pdf)

S. Larsen, R. Rabbah, and S. Amarasinghe. Exploiting Vector Parallelism in Software Pipelined Loops. In Proceedings of the 38th International Symposium on Microarchitecture, Barcelona, Spain, November 2005. (ps, pdf, ppt)

S. Larsen, E. Witchel, and S. Amarasinghe. Increasing and Detecting Memory Address Congruence. In Proceedings of the 11th International Conference on Parallel Architectures and Compilation Techniques, Charlottesville, VA, September 2002. (ps, pdf, ppt)

E. Witchel, S. Larsen, C.S. Ananian, and K. Asanovic. Direct Address Caches for Reduced Power Consumption. In Proceedings of the 34th Annual International Symposium on Microarchitecture, Austin, TX, December 2001. (pdf)

S. Larsen and S. Amarasinghe. Exploiting Superword Level Parallelism with Multimedia Instruction Sets. In Proceedings of the SIGPLAN '00 Conference on Programming Language Design and Implementation, Vancouver, B.C., June 2000. (ps, pdf, ppt)
errata (correct in online version)

Samuel Larsen. Exploiting Superword Level Parallelism with Multimedia Instruction Sets. Master's thesis, Massachusetts Institute of Technology, May 2000. (ps, pdf)

SUIF to Trimaran Converter

For my doctoral work I wrote a set of SUIF passes that convert SUIF IR to IMPACT Lcode, an IR used in the Trimaran compilation and simulation infrastructure. These passes, along with an updated version of SUIF, are now part of the Trimaran 4.0 release. You can download them from the Trimaran download page.