Charles Wuischpard, the general manager of the HPC Platform Group at Intel, said the company hopes that including the supercomputer chip on desktops would allow more PC makers to follow suit.
Advertisement
Intel also informed us that preproduction second-generation Intel Xeon Phi processors (code-named Knights Landing) are running in several supercomputing-class systems including Los Alamos (Trinity system), NERSC (Cori system), and CEA in France.
It’s not just about the silicon, though: Knights Landing is the first in the Xeon Phi family to move away from the original 32-bit Pentium microarchitecture.
Furthermore, Intel’s previous generation of Xeon Phi chips, which were traditionally co-processors, are deployed in the world’s fastest supercomputers such as the supercomputer code-named Tianhe-2, which is installed in China.
Like its predecessors, Knights Landing will be a coprocessor alongside CPUs and assist in complex calculations. However, it is different in design from graphics processors like Nvidia’s Tesla, which are used for the same purposes.
The workstation will also include preloaded programming tools and other software programs. Knights Landing will deliver the same high performance as a few of the graphics chips used in a few of the worlds fastest super computers. Moreover, the chip has been compared to have the same performance range as of a few graphics chips used in the fastest supercomputers in the world today. Then, they’re bundled up-along with 16GB of on-package MCDRAM memory-into a PCI-E add-in card (like the one pictured above, which is an engineering sample of an older Knights Ferry model). The memory has five times more bandwidth than DDR4 memory, and Intel claims that the stacked memory is five times more power efficient than GDDR. OmniPath is a proprietary interconnect, and Intel has not detailed the technology.
The goal of new technologies like Xeon Phi is to squeeze more performance out of supercomputers without drawing more power.
Advertisement
Intel will equip business workstation desktops with its supercomputer 72-core processor that provides over 3 teraflops of peak performance.
