Intel announces Xeon Phi processors for supercomputers to 10nm can be produced. The third generation of Xeon Phi will integrate a new version of the Omni-Path Architecture: a platform for optical interconnects that would be quicker than infiniBand.
During the Supercomputing Conference in the U.s. city of New Orleans Intel has the successor of the next Xeon Phi generation, code name Knights Landing published. This sequel is going to Knight Hill called and should be on 10nm are produced, while Knights Landing on 14nm produced Silvermont cores is equipped.
When Knights Hill should appear is not known, but in any case, after Knights Landing, which in the second half of 2015 appears. Intel announced Knights Landing earlier this year. The Xeon Phi processors are accelerators for high performance computing.
Intel also announced that upcoming Xeon Phi’s support for Omni-Path Architecture. This is a platform for optical interconnects with a throughput of 100Gbit/s and a switch latency, 56 percent, would be lower than that of the current InfiniBand communication lines.
The Omni-Path Architecture will use a 48poorts switch chip, where the maximum for InfiniBand 36 is located. The use of optical cables to the processors and other components in supercomputers to connect, to speedups and improved scalability lead.
Comments
(44)