17517

Dynamic Parallelism in GPU Optimized Barnes Hut Trees for Molecular Dynamics Simulations

Melisa Carranza Zuniga
Winston-Salem, North Carolina
Wake Forest University, 2017

@phdthesis{zuniga2017dynamic,

   title={Dynamic Parallelism in GPU Optimized Barnes Hut Trees for Molecular Dynamics Simulations},

   author={Zuniga, Melisa Carranza},

   year={2017},

   school={Wake Forest University}

}

Download Download (PDF)   View View   Source Source   

2542

views

Since the beginning of the modern computing era, high performance computing has been pushing the boundaries of the types of problems that can be solved in many different disciplines. One of the leading fields is computational biophysics where molecular dynamics (MD) simulations provide microscopic resolution details of how biomolecules move, fold, and assemble into intricate complexes that perform biological functions. However, it still remains a challenge to accurately perform MD simulations of biologically relevant complexes at timescales that can be directly compared with experiments. While the fundamental features of biomolecular dynamics, folding, and assembly are very interesting, their misfolding or misassembly can lead to deleterious repercussions that lead to diseases such as Parkinson’s and Alzheimer’s. MD simulations have played key roles in successes so far in directing experiments that lead to therapies, but advances in high performance computing hardware and algorithms will expand the scope of the problems that can be solved. For the past decade, graphics processing units (GPUs), which were originally designed for rendering images, have been successfully repurposed for scientific computing largely due to its parallel architecture that allow many cores to quickly implement parallel algorithms. Continuous advances in the capabilities of GPUs have introduced novel parallel algorithms that are specifically optimized for the GPU architecture. One of the key new capabilities of GPUs is dyanamic parallelism that allows a natural implementation of recursive algorithms, which have been difficult to implement except through intricate workarounds. In this thesis is presented the first application of the Barnes-Hut algorithm for Self-Organized Polymer (SOP) Model MD simulations on the GPU. To our knowledge, Barnes-Hut algorithm, which has been used extensively in other types of simulations in other fields, most notably astrophysics, has not been implemented for MD simulations of biological systems.
Rating: 4.0/5. From 4 votes.
Please wait...

* * *

* * *

HGPU group © 2010-2024 hgpu.org

All rights belong to the respective authors

Contact us: