MIT EECS - Landsman Undergraduate Research and Innovation Scholar
Saman P. Amarasinghe
Physics simulations are widely used in areas including science research, mechanical design, and video games production. These applications are usually computationally intensive in nature. Although modern day compilers can perform several optimizations on common code patterns to increase software performance, they are usually unaware of data structures, cache size, and other factors, providing limited optimizations to more complicated computations. The compiler group at CSAIL has developed a programmable system that recognizes specific computation processes. With this system, programmers can develop fast physics simulation software. Specifically, I will work on code generation and optimizations for this system.
I have worked at CSAIL Computational Fabrication Group on computation intensive algorithms for the graphical functionalities of 3D printers, and in fact, they are some important applications of the system being developed in this Super UROP. I also interned at Treyarch and my tasks included performance engineering and working with a parser. With years of experience using C/C++. I have solid understanding of the language down to the assembly level.