Tokyo Tech News
December 27, 2016
A group of researchers from the RIKEN Advanced Institute for Computational Science (AICS) and Tokyo Tech have been given the "Best Paper" award at the SC16 supercomputing conference in Salt Lake City, Utah. The paper, entitled, "Daino: A High-Level Framework for Parallel and Efficient AMR on GPUs," by Naoya Maruyama and Mohamed Wahib Attia of AICS, along with Takayuki Aoki of Tokyo Tech, provides a framework for more efficient design of applications for use on supercomputers. The paper presents a new framework that allows the use of "Adaptive Mesh Refinement" — a process for more efficient simulations — on supercomputers. Essentially, AMR allows a computer performing simulations to adjust the detail so that important areas are calculated very precisely whereas non-important ones are calculated in less detail. Though the process is very efficient, it has proven difficult to implement on highly parallel supercomputers due to the complexity of programming, and the award-winning paper's framework greatly eases the task. The paper was selected from among 442 submissions for presentations at the conference.
According to SC16 Awards Chair Satoshi Matsuoka, "These awards are very important for the SC Conference Series. They celebrate the best and the brightest in high performance computing."
First author Wahib of AICS said, "SC is the top technical conference in the area of high performance computing. Each year, the SC Technical Papers Committee identifies one paper as the best paper from the conference's technical program. It is a great honor and motivation to be selected for this award. We hope that this award will give our work more visibility. We further hope that the work in this paper will be a useful contribution for the high performance computing community."