|
Scientific Computing
Sandia's computational sciences program
creates technologies for all of Sandia's strategic business units. The program
develops numerical methods for solving scientific and engineering problems, a software
infrastructure for parallel computing, and techniques for distributed computing. The
work in numerical methods focuses on massively parallel computing. A key challenge
is to develop algorithms that scale to tens of thousands of processors. Software
infrastructure research includes work in parallel operating systems, mesh generation,
static and dynamic load balancing, visualization, and user interfaces.
Sandia has played a leading role in massively parallel computing for more than a
decade. During the past five years, Sandia and Intel have set many world records for
computational speed. Sandia is the only institution in the world that twice has won
the Gordon Bell Prize, which is given annually for the most significant achievement in
parallel computing. We have won numerous other national awards and have received
patents for their associated inventions. Sandia's software is licensed worldwide.
Computational technologies are playing an important role in a wide range of Sandia
programs. Examples include simulating the performance of weapons, analyzing weapons
safety issues, and computation-based design and manufacturing. An important goal for
the future is the creation of an easy-to-use design analysis environment that seamlessly
integrates model development, analysis, and postprocessing.
Distributed computing technologies are critical to fulfilling our future computing
requirements. Sandia is pioneering new techniques that will leverage commercial
off-the-shelf hardware and software to create a rich computing infrastructure. The
Computational Plant Project is creating many of the needed technologies. A
significant Computational environment is being developed. |