Cactus Grid Computing: Review of Current Development Gabrielle Allen 1 , Werner Benger 1, 2 , Thomas Dramlitsch 1 , Tom Goodale 1 , Hans-Christian Hege 2 , Gerd Lanfermann 1 , Andr´ e Merzky 2 , Thomas Radke 1 , and Edward Seidel 1 , 3 1 Max-Planck-Institut f¨ ur Gravitationsphysik, Albert-Einstein-Institut, Golm (AEI) 2 Konrad-Zuse-Zentrum f¨ ur Informationstechnik, Berlin (ZIB) 3 National Center for Supercomputing Applications, Champaign, IL, (NCSA) Abstract. Cactus is an open source problem solving environment de- signed for scientists and engineers. Its modular structure facilitates par- allel computation across different architectures and collaborative code development between different groups. Here we detail some of the vari- ous Grid Tools which have been developed around Cactus, and describe Grid experiments which have been performed to test their application. 1 Introduction Cactus [1],[2] is an open source problem solving environment designed to provide a unified modular, parallel, portable and collaborative computational framework for physicists and engineers. The modularity and flexibility in Cactus is achieved by its design of a cen- tral core (or flesh) which connects to application or computational infrastructure modules (or thorns) through an extensible interface. Thorns provide all the func- tionality, from parallel drivers, I/O and checkpointing to physics applications such as black hole evolvers. Cactus is distributed with a Computational Toolkit containing thorns which provide, for example, coordinate systems, boundary conditions, parallel interpolation and reduction operations, elliptic solvers, and various parallel output methods using different data formats. Cactus contains a sophisticated and flexible make system, and is supported on a wide range of HPC architectures, including Origin 2000, Cray T3E, IBM SP, Hitachi SR8000, Compaq AlphaServer, as well as IA32/IA64 clusters running Windows or Linux. 2 Grid Computing with Cactus Cactus, and the physics codes from which it evolved, have been used as a Grid computing software laboratory since 1993. Simulations of Einstein’s equations, with live 3D visualizations shown in a CAVE, were distributed across multiple machines (CM-5’s and SGI Power Challenges) at SC’93 and SIGGRAPH 94. Large scale distributed computing across the vBNS was demonstrated at Su- percomputing 1995 (SC’95), with a direct ancestor of Cactus using the Cornell R. Sakellariou et al. (Eds.): Euro-Par 2001, LNCS 2150, pp. 817–824, 2001. c  Springer-Verlag Berlin Heidelberg 2001