Managing Multicore with OpenMP

摘要

High end distributed and distributed shared memory platforms with many thousands of cores will be deployed in the coming years to solve the toughest technical problems. Their individual nodes will be heterogeneous multithreading, multi-core systems, capable of executing many threads of control, and with a deep memory hierarchy. For example, the petascale architecture to be put in production at the US National Center for Supercomputing Applications (NCSA) in 2011 is based on the IBM Power7 chip which uses multicore processor technology. Thousands of compute nodes with over 200,000 cores are envisioned. The Roadrunner system that will be deployed at the Los Alamos National Laboratory (LANL) is expected to have heterogneous nodes, with both AMD Opterons and IBM Cells configured, and a similar number of cores.