Intel recently gave the details about its latest and fastest Xeon chipset, the Xeon E5-2699A v4, and the upcoming Skylake-based chipsets at the SC16. The newest processor from the Santa Clara-based tech firm delivers a punch and is aimed at helping with high-performance computing (HPC) tasks.
Xeon E5-2699A v4 is based on Intel's fastest server chip, the 22-core Xeon E5-2699 v4. The former is the tuned up and tweaked version of the latter so it can deliver more CPU Performance. The new chipset also belongs to the Broadwell line of processors, which have already done an outstanding job when it comes to HPC. Intel is determined to improve this performance and their presence in the HPC market by developing more efficient variants of the Broadwell-based chipsets.
Given that Xeon E5-2699A v4 is based on its predecessor, Xeon E5-2699 v4, one can expect the same specifications, except for the 200 MHz boost. From Xeon E5-2699 v4's 2.2 GHz, the new variant now ends up with 2.40 GHz and features 55 MB of L3 cache. As noted by NextPlatform, the E5-2699A v4 is recorded to have a 4.8 percent better performance compared to its predecessor, based on the LINPACK parallel Fortran benchmark results. Although the new processor has an impressive improvement in the specs, it comes with a major downside - the Xeon E5-2699A v4 costs $4938.
The launch of Xeon E5-2699A v4 is proof that Intel is working harder to improve CPU performance. New chips usually have around 10 to 20 percent CPU performance boost. During a recent phone call, Intel executives discussed putting more speed into server chips to cater to applications requiring high-computing performance, PC World reports.
Meanwhile, Intel also gave some details about its upcoming Skylake-based processors. The next-generation chipsets will become available sometime in mid-2017. According to WCCFTech, the new line of processors should make its debut at Computex 2017. The Skylake-based chips are expected to have optimizations focused on HPC, including 512-bit Advanced Vector Instructions for more efficient encryption algorithms and floating point calculations. For high-speed networks, the Omni-Path architecture is also included.
