Horst Simon on the slowdown of HPC, Moore’s Law and Parallelism

On Feb 12, Horst Simon, Deputy Director of the Lawrence Berkeley National Laboratory and a maintainer of the Top500 List gave a talk arguing that Moore’s law and parallelism can no longer be counted on to provide the exceptional growth that has been driving high-performance computing for over 60 years.

He stated that there is a need for new architectures and technologies  such as non-von Neumann systems like quantum computing and neuromorphic (brain-like) computing. Simon provided measurable evidence using trends built using Top500 data and other sources.

He even brought into question the current projections of reaching exascale computing by 2020. Reasons for this include lack of funding, engagement from vendors and technical challenges. However, the limitation of Moore’s law is seen as the prime offender.

Nonetheless, exascale can, and will happen. He said: “the only thing standing between us and an exascale machine is a lot of money – billion of dollars of investment and maybe a power bill of $50-100 million a year.”

HPC Wire have the full article and 1+ hour video here.

This entry was posted in HPC Articles:. Bookmark the permalink.