. RSL is a parallel runtime system library developed at Argonne National Laboratory that is tailored to regular-grid atmospheric models with mesh refinement in the form of two-way interacting nested grids. RSL provides high-level stencil and interdomain communication, irregular domain decomposition, automatic local/global index translation, distributed I/O, and dynamic load balancing. RSL was used with Fortran90 to parallelize a well-known and widely used regional weather model, the Penn State/NCAR Mesoscale Model. Key words. Weather modeling, parallel computing, mesh refinement, dynamic load balancing. 1. Introduction. Models of the earth's atmosphere were among the first applications for supercomputers and continue to push the limits of available resources today [3]. Dynamic models of the atmosphere are used for forecasting and climate prediction. Such models may be categorized as global and regional. Global models provide relatively low-resolution predictive capabilities and are c...

Because large scientific codes are rarely static objects, developers are often faced with the tedious task of accounting for discrepancies between new and old versions. This paper describes a new technique called relative debugging that addresses this problem by automating the process of comparing a modified code against a correct reference code. The paper examines the utility of the relative debugging technique by applying a relative debugger called Guard to a range of debugging problems in a large atmospheric circulation model. The experience confirms the effectiveness of the approach. Using Guard, it was possible to validate a new sequential version of the atmospheric model, and to identify the source of a significant discrepancy in a parallel version in a short period of time. 1. Introduction Large scientific codes are constantly evolving. Refinements in understanding of physical phenomena result in changes to physics, improved numerical methods result in changes to solution tech...

Because large scientific codes are rarely static objects, developers are often faced with the tedious task of accounting for discrepancies between new and old versions. In this paper, we describe a new technique called relative debugging that addresses this problem by automating the process of comparing a modified code against a correct reference code. We examine the utility of the relative debugging technique by applying a relative debugger called Guard to a range of debugging problems in a large atmospheric circulation model. Our experience confirms the effectiveness of the approach. Using Guard, we are able to validate a new sequential version of the atmospheric model, and to identify the source of a significant discrepancy in a parallel version in a short period of time.

This paper describes a tool for debugging programs which develop faults after they have been modified or are ported to other computer systems. The tool enhances the traditional debugging approach by automating the comparison of data structures between two running programs. Using this technique, it is possible to use early versions of a program which are known to operate correctly to generate values for comparison with the new program under development. The tool allows the reference code and the program being developed to execute on different computer systems by using open distributed systems techniques. A data visualisation facility allows the user to view the differences in data structures. By using the data flow of the code, it is possible to locate faulty sections of code rapidly. An evaluation is performed by using three case studies to illustrate the power of the technique. 2 1. INTRODUCTION The high cost of software development, in combination with advances in software engineer...

This presentation discusses the NASA data assimilation project at the Data Assimilation Office at the NASA/Goddard Space Flight Center. The goal is to produce accurate gridded datasets of atmospheric fields by assimilating a range of observations along with physically consistent model forecasts. This work produces datasets that are used by the climate research community. The data come from conventional sources that are used for weather forecasts (e.g., radiosondes, earth-surface measurements, and satellite temperature retrievals), as well as new sources such as satellites that will be launched under the Mission To Planet Earth Enterprise. An end-to-end Goddard Earth Observing System (GEOS) Data Assimilation System (DAS) currently supports stratospheric flight missions and reanalysis projects for NASA. The current Core of this system (Model, and Analysis) is a multitasking algorithm that runs on Cray J90 and C90 computers at Goddard and NASA Ames Research Center. Future Core computing w...

With the March 1998 release of the Penn State University/NCAR Mesoscale Model (MM5), the official version of the model (MM5v2 Release 8) now runs on distributed memory (DM) message-passing platforms. Under an IBM-funded effort, source translation and runtime library support minimize the impact of parallelization on the original model source code with the result that the majority of code is line-for-line identical with the original version. Parallel performance and scaling are equivalent to earlier, hand-parallelized versions; the modifications have no effect when the code is compiled and run without the DM option. Supported computers include the IBM SP2, Cray T3E, and Fujitsu VPP. The approach is compatible with sharedmemory parallelism, allowing DM/SM hybrid parallelization on distributed memory clusters of SMP. Preliminary results show that scalability on distributed shared memory computers such as the SGI Origin 2000 also benefits from a distributed memory programming mode...

This proposal addresses the difficulties in attaining high performance in climate models when ported to new computer systems or new configurations of existing systems. The short-term goal is to evaluate the current and emerging scientific supercomputers in the 100 GFlop to 1 TFlop range and identify how they can be used most efficiently for production scale computing of coupled atmosphere-ocean general circulation models (GCMs). Primary performance goals are achieving good scalability on large numbers of processors and good per-node performance on the individual processors; but, of critical importance, the performance gains must be portable. Massive restructuring of a code for efficiency on one platform that impairs efficiency on others is ultimately counter-productive. Thus, the long-term goal is to develop and demonstrate performance portability techniques that allow what appears to be a single source code to the computational scientist to be easily ported and optimized on different ...

This paper describes MM90, a parallel regional weather model based

Until recently, debates concerning humans and their environment tended to favor either cultural or environmental mechanisms as the driving force behind change, but the most persuasive studies now focus on the interface between human/cultural systems as well as interactions between them. We have proposed that early urban settlements in the Near