PFI-TT: A Parallel Computing Engine for Simulation of Complex Multi-Scale Systems

点击翻译按钮获取中文摘要

The broader impact/commercial potential of this PFI project stems from the development of an inexpensive, highly parallel computing platform to solve scientific and engineering simulations while taking advantage of the front end and post-processing tools of commercially available finite-element software. The breakthrough scientific and commercial opportunity stems from the use of parallel computing hardware developed in recent years in support of the computer gaming industry, that has not been leveraged yet for solution of simulation problems currently addressed using finite element methods. This can lead to a breakthrough in the simulation industry by embracing an approach to the solution of boundary value problems that does not require inversion of large matrices. The proposed approach will benefit several scientific and engineering applications where large dimensions or multi-scale geometries makes meshing computationally prohibitive, and addresses critical bottlenecks in engineering design and simulation, in particular in transient analysis and design optimization. The potential business opportunity is significant: the successful adoption of the proposed approach would lead to a major breakthrough in the performance of software for engineering and scientific simulation, while exploiting newly available parallel computing resources (gaming graphics engines).The proposed project supports the development of an inexpensive, highly parallel simulation platform that can be integrated with pre and post processing tools of existing finite-element software. The proposed module will be implemented using gaming graphics processors, enabling a breakthrough speed-up in simulation time compared to conventional finite element software. The final deliverables will be presented to potential partners or investors for commercialization and future development. Interviews with engineers and scientists under an NSF I-Corps study revealed a significant need and interest for simulation tools that address the bottlenecks of existing finite-element tools, in particular in terms of prohibitively long simulation times. The proposed project is transformative for the following reasons: (i) A novel approach for the simulation of complex systems using highly-parallel computing is proposed, (ii) the proposed approach can lead to a breakthrough reduction in simulation time compared to available software in large scale problems and transient analyses, and (iii) the proposed approach will lead to a simulation engine that can be integrated with pre and post processing tools from existing finite element method (FEM) software, enabling its use within well developed products that already have a solid customer base.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.