Xeon D is targeted at the edge
Last week, Intel announced its new Intel Xeon D-2100 processor family, a follow-up to the company’s Intel Xeon Scalable Platform launch last year (read my initial coverage here). The new offering is geared towards bringing many of the capabilities of Xeon Scalable Platform to edge computing and other workloads that necessitate power and performance density. Xeon D is a new part in a family of processors, including the Intel Xeon Scalable Processor and the Atom C3000 processor, which vary by raw performance, density, power draw, and price. I see this as a promising addition to Intel’s portfolio, and one that fills an important niche. Here’s my take on the part.
Expanding on Xeon Scalable Platform
When the Xeon Scalable Platform was released mid-2017, I called it the company’s broadest datacenter and carrier play to date. Targeting workloads for enterprises, carriers, HPC, cloud, and AI, the launch marked a significant expansion of Intel’s datacenter efforts. The Xeon Scalable Processor promised on average a 1.65x performance increase over the previous generation of Xeon, and its comprehensive package of chip improvements, platform enhancements, accelerators, and software stacks targeted specific workloads, down to the silicon level. I wrote then that the Xeon Scalable Processors gave Intel growth potential in areas it has historically struggled in, while simultaneously allowing the company to defend its CPU turf. Flash forward to now, and it looks like the new processors are doing pretty well—Intel is seeing double-digit growth in its datacenter group, driven by Xeon Scalable chips, with 20% YoY growth in Q4 2017 versus Q4 2016. Sure some of this was pent-up cloud demand, but even a 10% bump would have been respectable.
Xeon D targets the “edge”
This brings us to Xeon D. The edge compute market is one of the fastest growing in the industry. With the proliferation of smartphones, smart cars, and drones, the approximately 10 billion current IoT end points are creating an incredible amount of data—and it is growing exponentially. This data needs to be processed either in the device, gateway, “distributed core,” or the traditional or cloud datacenter. When responsiveness and security are key, the data is best processed as close to the source as possible—hence, edge compute. Carriers, enterprises, and CSPs are completely overhauling their networks to address this data explosion, and with Xeon D, Intel is trying to get a piece of the action. With Xeon D, Intel is intending to serve the growing edge compute market but also replace older MIPS and PowerPC installs.
Compared to its predecessor, the D-1500, the D-2100 is claiming as much as 1.6x general compute performance, 29% better packet transfer performance for multi-access edge computing, 2.9x network performance, and 2.8x warm storage performance. In a nutshell, Xeon D-2100 is all about delivering the highest performance at the highest density and lowest power.
Intel announced a bunch of Xeon D-2100’s host of launch-day partners—Dell EMC, Ericsson, F5, NEC, NetApp, Palo Alto Net, and Supermicro. This says a lot about the industry’s confidence in the platform and includes who you might expect. To me, notable exclusions from the list were Hewlett Packard Enterprise and Huawei, both who play strong at the edge. One might also wonder how Lenovo fits into the picture, but it isn’t really playing aggressively on the edge yet.
Xeon-D launch partners
All in all, the Xeon D-2100 looks to be a great follow-up to the Xeon Scalable Platform launch, hitting many of the right notes to be attractive to those playing in the growing edge compute market and other power and performance-dense workload environments. Intel continues to broaden and diversify its reach and edge compute is a growth market for Intel—I look forward to seeing how the D-2100 lands with CSP, carriers, and enterprises in the coming months.