View Position
| Project Title | Intern in Acoustic Simulation |
| Summary | The intern will work on developing parallel and scalable algorithms for solving the acoustic wave equation. It builds on recent research at UNC Chapel Hill on a new wave solver based on adaptive decomposition and parallelizing on a distributed memory architecture. The goal is to develop a massively parallel solver and evaluate its performance on indoor and outdoor space. |
| Job Description | We request an intern and computing cycles on Blue Waters computing system to perform peta-flop acoustic simulation. This work builds on the work on developing novel wave-based solvers for acoustic simulation at UNC Chapel Hill for 6 years and recent work on developing massively parallel algorithm. My graduate students have developed a new solver for homogeneous environments using adaptive rectangular decomposition and also parallelized on a single many-core GPUs, as well as distributed memory architectures. The specific goals of this project include: 1. Evaluate the scalability of our distributed memory algorithm on large number of cores: 10,000 â 100K cores. 2. Evaluate the approach on high frequency benchmarks, which result in very small and irregular subdivisions. Improve the load balancing and scalability on such benchmarks. 3. Extend our approach to improve the parallel performance and thereby enable performing accurate acoustic simulation on very large indoor and outdoor acoustic spaces. If successful, this project can result in major advances in developing good parallel solvers for acoustic simulation, a major problem in scientific and engineering applications. The undergraduate intern will interact closely with my graduate (Ph.D.) students at UNC and get exposed to research in parallel computing and acoustic simulation. |
| Conditions/Qualifications | The intern should have sufficient background in programming and parallel computing to run these simulations. |
| Start Date | 05/15/2014 |
| End Date | 05/31/2015 |
| Location | The student can work at UNC Chapel Hill or a remote location with access to a massively parallel computing system. |
| Interns | Vivek Chavda
|