CCA is applicable to a broad range of parallel applications. We highlight a few of these endeavors.
Combustion Modeling. One of the most sophisticated implementations of the CCA paradigm to date is in combustion modeling. The endeavor, which started in 2001 at the Computational Facility for Reacting Flow Science (CFRFS) project,10 seeks to create a facility for the high fidelity simulation of flames involving realistic physical models, nonlinear PDEs, and a spectrum of time and length scales. Given the complexity of the problem and the multiplicity of physical and mathematical models required for the task, a component based approach was a natural fit. CCA was chosen primarily for its high performance and simplicity. General purpose components, implementing a particular numerical or physical functionality, are reused in various code assemblies.
Fusion. The standardization of fusion codes has become of paramount importance with the beginning of the fusion integrated modeling. In 2006, two SciDAC projects, SWIM (Center for Simulation of RF Wave Interactions with Magnetohydrodynamics)11 and CPES (the Center for Plasma Edge Simulation),12 were funded. In 2007, another integrating SciDAC project FACETS (the Framework Application for Core-Edge Transport Simulations)13 started. Each of these three projects addresses a different area of integration, and some of them will possibly unite in the upcoming Fusion Simulation Project sometime in 2008. All projects are looking at components technologies to assist them in defining and composing participating codes. FACETS uses the CCA language interoperability tool, Babel, to wrap F90 modules for the use in its C++ framework.
Quantum Chemistry. In response to the strong need for a community software base in the quantum chemistry community, members of the Quantum Chemistry Scientific Application Partnership (QCSAP) have leveraged the software engineering practices and tools developed by the CCA Forum to create a community-based architecture for chemistry simulation. By designing flexible interfaces by which quantum chemistry capabilities can be shared, the scaling of human effort is drastically improved, allowing the exploration of new algorithms and hardware in a fraction of the time required by traditional approaches. A screenshot of such an application is shown in Figure 4. For the quantum chemistry packages adopting this new design paradigm [GAMESS (Ames Laboratory), NWChem (Pacific Northwest National Laboratory) and MPQC (Sandia National Laboratories) thus far], many advantages have already been realized, including the abilities to leverage more efficient optimization components written by domain experts, transparently share low-level integral evaluation routines, more efficiently utilize high performance computers, and automatically tune applications for specific molecular systems and hardware environments.
![Figure 4](../../../../../figures/12/parker-figure4.gif)