A Common Component Architecture (CCA) based design and implementation for distributed parallel magnetotellurice forward model

The computing of Magnetotelluric forward and inversion model based on finite element method has always been carried out on stand-alone computers in geophysical resources exploration. There is however a large amount of inefficient operations due to the huge amount of calculations involved. It became imperative to perform these calculations on distributed high-performance environments. This paper presents a distributed high-performance Magnetotelluric forward model MT based on large-scale scientific Common Component Architecture (CCA) which was recently proposed by the U.S. Department of Energy in conjunction with the Indiana University and the University of Utah. This design not only makes the model compute in high performance, but also loosely couples the model with the underlying parallelism. Thus, the model has high degree of reusability and scalability, and the underlying parallelism is transparent to users resulting meaning the user could fully concentrate on physics development, without attending to the underlying computing details. Two experiments was carried out on a Windows Server 2003 (x64) workstation, and the result demonstrates desirable performance.