GPU Implementation of Iterative Solvers in Numerical Weather Predicting Models
نویسنده
چکیده
Numerical weather predicting models often require solving a 3-D Helmholtz problem which derived from the governing equation of dynamical core in Met Office Unified Model, by preconditioned iterative solvers. In this dissertation, a GPU implementation of preconditioned conjugate gradient (CG) iterative method will be focused on. A given serial code has been ported on GPU. According to the porting, performance analysis and optimisation will be carried out. At last, this dissertation defines the main performance bottleneck of the GPU implementation of preconditioned CG. According to these analyses, an initial optimisation and further optimisation suggestions will be shown in this dissertation.
منابع مشابه
Matrix-free multi-GPU Implementation of Elliptic Solvers for strongly anisotropic PDEs
Manycore chip architectures such as Graphics Processing Units (GPUs) have been shown to be highly e cient both in terms of total performance and small power consumption in many areas of Scientific Computing. However, implementing memory-bound applications such as elliptic solvers for large partial di↵erential equations (PDEs) remains a challenge. We consider iterative solvers for strongly aniso...
متن کاملMatrix-free GPU implementation of a preconditioned conjugate gradient solver for anisotropic elliptic PDEs
Many problems in geophysical and atmospheric modelling require the fast solution of elliptic partial differential equations (PDEs) in “flat” three dimensional geometries. In particular, an anisotropic elliptic PDE for the pressure correction has to be solved at every time step in the dynamical core of many numerical weather prediction (NWP) models, and equations of a very similar structure aris...
متن کاملPetascale elliptic solvers for anisotropic PDEs on GPU clusters
Memory bound applications such as solvers for large sparse systems of equations remain a challenge for GPUs. Fast solvers should be based on numerically efficient algorithms and implemented such that global memory access is minimised. To solve systems with up to one trillion (10) unknowns the code has to make efficient use of several million individual processor cores on large GPU clusters. We ...
متن کاملViscous Models Comparison in Water Impact of Twin 2D Falling Wedges Simulation by Different Numerical Solvers
In this paper, symmetric water entry of twin wedges is investigated for deadrise angle of 30 degree. Three numerical simulation of a symmetric impact, considering rigid body dynamic equations of motion in two-phase flow is presented. The two-phase flow around the wedges is solved by Finite Element based on Finite Volume method (FEM-FVM) which is used in conjunction with Volume of Fluid (VOF) sc...
متن کاملGPU Based Acceleration of WRF Model: A Review
The Weather Research and Forecasting model (WRF) is a simulating system developed for atmospheric weather prediction. WRF model is used for both operational as well as research purposes. The need for accurate weather and climate simulation to be carried out in shorter time is increasing day by day, which leads to the acceleration of existing Numerical Weather Prediction (NWP) system. This paper...
متن کامل