Parallel dense linear Algebra software in the multicore era

The recent emergence of multicore and hybrid microprocessor designs marks the beginning of a forced march toward an era of computing in which research applications must be able to exploit parallelism at an unprecedented scale. This chapter presents a new generation of dense linear algebra libraries that achieve the fastest possible time to an accurate solution on multicore systems by efficiently using all the processors that such systems will make available. To work within this design space and leverage the power of million way parallelism, it is necessary to combine new, highly parallelizable algorithms, a programming and execution model that can exploit massive task parallelism, and a flexible memory management facility that can help optimize data locality across a range of different platforms. The design space is also conditioned by the fact that, in order to support the broadest possible range of Computational Science, the resulting library frameworks must be able to scale both up and down, running at all levels of the platform development chain.