Towards Optimal Petscale Simulations (TOPS)
Multiscale, multirate scientific and engineering applications in the SciDAC portfolio possess resolution requirements that are practically inexhaustible and demand execution on the highest-capability computers available, which will soon reach the petascale. While the variety of applications is enormous, their needs for mathematical software infrastructure are surprisingly coincident; moreover the chief bottleneck is often the solver. At their current scalability limits, many applications spend a vast majority of their operations in solvers, due to solver algorithmic complexity that is superlinear in the problem size, whereas other phases scale linearly. Furthermore, the solver may be the phase of the simulation with the poorest parallel scalability, due to intrinsic global dependencies. This proposal brings together the providers of some of the world's most widely distributed, freely available, scalable solver software and focuses them on relieving this bottleneck for many specific applications within SciDAC, which are representative of many others outside.
Solver software directly supported under TOPS includes: hypre, PETSc, SUNDIALS, SuperLU, TAO, and Trilinos. Transparent access is also provided to other solver software through the TOPS interface. The primary goals of TOPS are the development, testing, and dissemination of solver software, especially for systems governed by PDEs. Upon discretization, these systems possess mathematical structure that must be exploited for optimal scalability; therefore, application-targeted algorithmic research is included. TOPS software development includes attention to high performance as well as interoperability among the solver components. Support for integration of TOPS solvers into SciDAC applications is also directly supported by this proposal.
The TOPS consortium will conduct algorithmic research on optimal solution methods for simulations based on partial differential equations, create software, and provide consulting on its use throughout the DOE national laboratory complex.
Established as part of DOE's 2001 line-item initiative "Scientific Discovery through Advanced Computing," the "Integrated Software Infrastructure Center" (ISIC) is one of only seven created around the country. The Center focuses on developing and implementing algorithms and supporting scientific investigations performed by the DOE. Simulations of importance often involve the solution of partial differential equations on terascale computers - those capable of performing more than a trillion calculations per second. The TOPS Center will research, develop, and deploy an integrated toolkit of open-source, optimal complexity solvers for the nonlinear partial differential equations that arise in many DOE application areas, including fusion, accelerator design, global climate change, and reactive chemistry. The algorithms created as part of this project will aim to reduce current computational bottlenecks by orders of magnitude on terascale computers, enabling scientific simulation on a scale heretofore impossible.
In many areas of science, physical experimentation is impossible, such as with cosmology; dangerous, as with manipulating the climate; or simply expensive, as with fusion reactor design. It is hoped that large-scale simulation will give scientists insight and confirmation of existing theories in such areas, without benefit of full experimental verification. The codes used for such simulations may be checked against experiment in a variety of well understood laboratory contexts to validate them. Along with usability, robustness, and algorithmic efficiency, an important goal of this ISIC will be to attain the highest possible computational performance in its implementations by accommodating to the memory bandwidth limitations of hierarchical memory architectures. Today's high-end computers, such as the world's most powerful unclassified 4-Teraflop machine at Lawrence Berkeley National Laboratory, designed by IBM, are one-of-a-kind, and come without all of the scientific software libraries that scientists expect to find on desktop workstations.
Extraordinary advances in computing technology in the past decade have set the stage for a major advance in scientific computing. Within the next five to ten years, computers another thousand times faster than today's computers will become available. These advances herald a new era in scientific computing - if they can be harnessed with scalable algorithms and software. To exploit this opportunity, these computing advances must be translated into corresponding increases in the performance of the scientific codes used to model physical, chemical, and biological systems. This is a daunting problem. Current advances in computing technology are being driven by market forces in the commercial sector, not by scientific computing. Harnessing commercial computing technology for scientific research poses problems unlike those encountered in previous supercomputers, in magnitude as well as in kind. These problems will be solved only with increased investments in computer software - in research and development on scientific simulation codes as well as on the mathematical and systems software that underlie these codes.
TOPS Philosophy links you to a 4-page pdf file that provides a more technical overview of the need for and philosophy behind the TOPS project.