Intel is pushing the efforts on Exascale computing by bringing in more partners from all over the world
For the last few years, Intel has been making different moves towards the development of exascale computing: developing the new many-core Xeon Phi coprocessors, purchasing QLogic's InfiniBand, and now developing a number of Parallel Computing Centers.
Raj Hazra, General manager of Intel's Technical Computing Group and vice president in the company’s Datacenter and Connected Systems Group explains that exascale computing development will take more than new hardware such as processors and servers. It will also need to optimize applications in use today. New software will need to leverage the whole system and function at any level, from individual workstations to supercomputers.
Intel is hoping to use the Parallel Computing Centers to speed up the development of parallel applications that are open standard, scalable and portable. According to Hazra, this will be achieved by combining a range of different elements: hardware, computational science, programmer tools, libraries and compilers, with domain expertise and knowledge.
Intel, various research institutions, other vendors, different organizations and even the US Government are pushing hard to get to exascale level computing. It would mean a thousand times increase from petascale level. Exascale computing is expected to have huge impact on industries as oil, gas, biology, engineering and national security.
The first Parallel Computing Centers are going to be created in Italy, at Cineca in Italy, in Germany at Zuse Institut in Berlin, in the US at Purdue University, at the University of Texas Advanced Computing Center, and at the University of Tennessee.