Xeon Phi uses the same 22 nm, 3-D tri-gate
transistors
Intel has official launched
the new Coprocessor called Intel Xeon Phi based on the “Many Integrated Core
Architecture" (MIC) structure. Previously, Xeon Phi was developed under
the Knight Corner code name. Xeon Phi uses the same 22 nm, 3-D tri-gate
transistors as Ivy Bridge, but one tiny difference is that it comes as PCIe
card with more than 50 cores and at least 8GB of GDDR5 memory. Xeon Phi is used
to support the CPU Xeon E5-2600/4600, assisting computers to highly-parallel
processing. Xeon Phi is suitable for the station systems, the data centers or
even the super-computers. The product is expected to be released at the end of
this year.
|
Coprocessor is a processor that a computer uses to
support the CPU. The tasks that Coprocessor can carry out are calculating
floating points, graphical processes, signals, strings, coding… Coprocessor
plays an important role in reducing tasks for CPU, which help to increase the
system’s processing speed. Coprocessor allows the customers to customize
their own computers. Therefore, the payment depends on their needs and
demands, users are not supposed to pay the untouched performance.
|
In addition, Xeon Phi has the 512b wide SIMD support and
allows for multiple elements to carry out in one single instruction from RAM.
Intel revealed that every Xeon Phi chip could speed up at 1 teraflops. This
chip works well with x86 current programming models.
Every Xeon Phi
Card can be considered “a computer node”
Every Xeon Phi Card can be considered “a computer node”
running on the isolated Linux, independent of the host computer OS, performing
a highly paralleled processing, maximizing solutions for the High Performance
Computer. Intel said 44 companies had already signed up to use Xeon Phi,
including Dell, HP, IBM, and NEC… It promised to reach the Exaflops speed by
2018.
The Xeon Phi introduction will help
Intel to highly compete with NVIDIA. This producer originally produces PCIe
cards supported by GPGPU Tesla which show off the highly strong calculating ability
and they are used to support the High Performance Computer.
