- 學位論文
高速圖形顯示卡於自適性網格天文物理電腦模擬之應用與效能
A Novel Adaptive Mesh Refinement Code with Graphic-Processing-Unit Acceleration and Its Applications to Astrophysics
並列摘要
We present the newly developed code, GAMER (GPU-accelerated Adaptive MEsh Refinement code), which has adopted a novel approach to improve the performance of adaptive mesh refinement (AMR) astrophysical simulations by a large factor with the use of the graphic processing units (GPU). The AMR implementation is based on constructing a hierarchy of grid patches with an octree data structure. A hybrid CPU/GPU model is adopted, in which the time-consuming PDE solvers are implemented into GPUs and the complicated AMR data structure is manipulated by CPUs. The code is fully parallelized using either the rectangular domain decomposition or the Hilbert space-filling curve method. Several performance optimization strategies have been implemented in order to fully exploit the computing power in heterogeneous CPU/GPU clusters. To demonstrate the extraordinary capability of GAMER, we have implemented two astrophysical applications into the code, namely, the hydrodynamic system with self-gravity and the extremely light bosonic dark matter (ELBDM) model. In both cases, we first verify the accuracy of the code by performing several test problems with analytical solutions and estimate the numerical errors. We then demonstrate the high performance of GAMER by comparing the performances with and without GPU acceleration, in which maximum performance speed-ups of 68 and 35 are achieved in the hydrodynamic simulations and the ELBDM simulations, respectively.
參考文獻
延伸閱讀
- Goh, S. K. (2011). 多重解析度互動桌面系統之顯卡加速的軟體開發套件 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2011.01038
- Cheng, H. C. (2020). 應用於自動多視點三維顯示器之即時全高清光場分解晶片架構設計 [master's thesis, National Tsing Hua University]. Airiti Library. https://www.airitilibrary.com/Article/Detail?DocID=U0016-0209202114072724
- Lin, C. Y. (2009). 以圖形處理器為基礎之高效能樣式比對演算法 [master's thesis, National Tsing Hua University]. Airiti Library. https://doi.org/10.6843/NTHU.2009.00143
- Lin, Z. Y. (2016). Hardware-Friendly Algorithm and Architecture Design of Improved Dense Trajectories for Real-time Action Recognition [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU201702226
- Jian, H. M. (2021). Acoustic Modal Analysis and Applications from a Perspective of State-space Balanced Realization [master's thesis, National Tsing Hua University]. Airiti Library. https://www.airitilibrary.com/Article/Detail?DocID=U0016-2412202116174693