EU Regional School - Laird Seminar Part2
Prof. Dr. Carl D. Laird - Parallel Nonlinear Optimization on Emerging High-Performance Architectures
Texas A&M University
The size and complexity of nonlinear optimization problems of interest in academia and industry continues to grow, often outstripping the capability of a single CPU workstation. Furthermore, computer chip manufacturers are no longer focusing on increasing clock speeds and instruction throughput, but rather on multi-core architectures and hyper-threading. This means that the “free” performance improvements that we have enjoyed as a result of advances in computing hardware will no longer be possible unless we develop algorithms that are capable of utilizing modern concurrent architectures efficiently. All computing architectures are not created equal, and we will discuss the key differences in parallel architectures available for scientific computing, including distributed clusters, general multicore systems, and graphics processing units (GPU). This lecture will build on the theory of large-scale nonlinear programming and specifically address the potential for parallel computation. In particular, we will discuss both problem level decomposition techniques and internal decomposition techniques for structured problems like those arising in optimal design and optimal control.