Intel’s Xeon family of processors dominates the November 2016 Top500 list. Intel holds over 92% market share in the most recent Top500 list. Extending on this lead Intel is launching a new Xeon E5-2699A V4. The company is also showing off Skylake-EP at the show.
The Intel Xeon E5-2699A V4 is a higher-clocked (200MHz) version of the Intel Xeon E5-2699 V4 we saw launched in March 2016. That clock speed increase, in turn, gives an increase in HPC performance.
We will have more from the SC16 show floor but Intel released more information on Skylake-EP. The company is targeting mid-2017 availability. They are also touting that the new architecture will have AVX-512 instructions included. Finally, the company is confirming the support for Intel Omni-Path interconnects.
Given this, we can understand why Omni-Path made a huge move on the Top500 list in November 2016. For those looking toward 2017 deployments, Skylake-EP and Knights Landing/ Knights Mill Xeon Phi generation parts have integrated interconnects makes the interconnect a low-cost leader.
One new bit of information that Intel confirmed at the SC16 press briefing was that Skylake-EP will utilize the 14nm plus process introduced first with Kaby Lake. Intel claimed improvements over the original 14nm process so it is great to see this technology utilized on Skylake-EP.
Here is an excerpt from the Intel SC16 press release on the new processors:
Intel Xeon platforms are already powering virtually all HPC and machine learning workloads and to push that further, Intel is introducing its fastest two-socket processor, the Intel Xeon processor E5-2699A v4.
- The Intel® Xeon® processor E5-2699A v4 delivers industry-leading performance for demanding, data-intensive applications ideal for customers who need high performance from the world’s most-chosen data center processor family.
- At Supercomputing 2016, Intel will publically demonstrate an HPC cluster featuring an early version of the next-generation Intel Xeon processor utilizing integrated Intel OPA and Intel® Advanced Vector Extensions-512 (Intel® AVX-512), complemented by Intel Xeon Phi processors.
- Intel AVX-512 processor instructions, launched June 2016 with the Intel Xeon Phi processor, accelerates compression and encryption algorithms. It provides hardware-enhanced data protection and fast time-to-insight, and can accelerate floating-point operations used in a wide range of HPC applications.