MIT SoE | Quest Undergraduate Research and Innovation Scholar
Optimizing Performance on Graph Applications With A High-Performance Graph DSL
- Software Design and Programming Languages
Saman P. Amarasinghe
This project focuses on the use, case implementation, and evaluation of GraphIt, a new high-performance graph DSL. The first phase of my SuperUROP project will be to implement several larger algorithms in the GraphIt DSL to test if additional operators are needed for GraphIt. Currently, we have planned to implement the following algorithms: community detection, delta-stepping, and Stochastic Approach for Link-Structure Analysis(SALSA). The second phase of my project will be to evaluate the algorithms’ performance when implemented with GraphIt and compare it to the performance of existing frameworks, including Ligra, GraphMat, and GraphMarl. The evaluation will be based on the runtime performance as well as scalability measures with regards to both data size and processor count.
“I’ m participating in SuperUROP because I want to gain more advanced research experience in computer science. I have learned during my previous UROP what to expect in SuperUROP, and I have gained a lot of programming experience from my previous summer internships, which will help me do my SuperUROP project better. I want to learn about both performance programming and the research process, and I hope to publish a paper by the end of my SuperUROP experience.”