Wolfgang Joppich

Parallelizing the ECMWF's weather forecast program: the 2D case

Abstract GMD and ECMWF (European Centre for Medium-Range Weather Forecasts) joined forces some months ago in order to parallelize ECMWFs production code for medium-range weather forecasts, the Integrated Forecasting System (IFS). Meanwhile, the first milestone of this cooperation has been reached: The 2D model of the IFS, which contains already all relevant data structures and algorithmic components of the corresponding 3D models, has been parallelized and run successfully on quite a large variety of differential parallel machines. Performance measurements confirm the expected parallel efficiencies of up to 80% and more. This paper discusses the parallelization strategy employed and gives a survey on the performance results obtained on the parallel systems.

A comparison of parallel multigrid and a fast Fourier transform algorithm for the solution of the Helmholtz equation in numerical weather prediction

Abstract The efficient solution of Helmholtz equations is required in limited area weather forecasting models. The parallel performance of two different algorithms — FFT with Gauss elimination and multigrid — is compared.

Real applications on the new parallel system NEC Cenju-3

The new massively parallel computer Cenju-3 of NEC has entered the market recently. NEC has set up a 64-processor machine at GMD. After the implementation of the PARMACS programming interface, large applications have been ported to the system. Early benchmark results show the performance of the Cenju-3 in a variety of application areas.

Semiconductor Process Modeling

The sections in this article are 1 Physical Models 2 Discrete Models 3 Process Modeling and Simulation Tools 4 State of Technology and Future Trends Keywords: topography modeling; semiconductor structures; ion implantation; bulk particle transport; mechanical deformation models; front propagation techniques; grid generation; grid adaptation; discretization; process simulation tools; simulation

A parallel multigrid solver applied to the simulation of thermal oxidation and diffusion processes

The complexity of the future processing models and the need for three-dimensional simulation require both efficient numerical methods and fast computers. In this paper, a general parallel PDE solver is presented, which has been adapted to the solution of so-called coupled problems. The numerical solution strategy exploits the optimal behavior of multigrid methods and the grid partitioning for an easy parallelization. As the typical thermal processes in VLSI process simulation can be considered as a coupled problem, the program has been applied to this important class of applications. The flow of the silicon dioxide is modeled as viscous flow. The impurity diffusion takes oxidation enhancement into account.

Medium-range weather forecast on parallel systems

GMD and ECMWF (European Centre for Medium-Range Weather Forecasts) joined forces in autumn 1992 in order to parallelize the Integrated Forecasting System (IFS), that is run operationally each day for the European medium-range weather prediction. The parallelization is performed in a portable way with all relevant parallel systems as target machines. Meanwhile, the parallelization of the 3D weather model has been achieved: the code runs with excellent parallel efficiencies of 90% and more on different parallel systems such as the SP1 (IBM) and the CM5 (Thinking Machines). First optimizations promise significant further improvements.<<ETX>>

First results with a parallelized 3D weather prediction code

Abstract A survey on the parallelization of the 2D model of the Integrated Forecasting System (IFS), the weather prediction code of the ECMWF, is given in this issue [4]. In this short note, we present first results of the parallelization of the corresponding 3D version.

An Environment for the Parallel Solution of Coupled Problems on Structured Grids

The solution of coupled problems is one of the grand challenges in scientific computing. Many problems can only be solved partially. Not all dependencies can be taken into account, e.g. only subproblems can be solved completely, even when optimal numerical methods (multigrid methods) are used. Therefore, parallel simulation is inevitable.

Empirical performance modeling for parallel weather prediction codes

Abstract Performance modeling for large industrial or scientific codes is of value for program tuning or for selection of new machines when benchmarking is not yet possible. We discuss an empirical method of estimating runtime for certain large parallel programs where computational work is estimated by regression functions based on measurements and time cost of communication is modeled by program analysis and benchmarks for communication primitives. The method is demonstrated with the local weather model (LM) of the German Weather Service (DWD) on SP-2, T3E, and SX-4. The method is an economic way of developing performance models because only a moderate number of measurements is required. The resulting model is sufficiently accurate even for very large test cases.

Procedures for Adaptive Multigrid Simulation of Evolution Processes

This chapter presents procedures for the adaptive multigrid simulation of evolution processes.