News has just rolled in that Intel has unveiled a major refresh of its famed Xeon D System-on-a-Chip processors. In case the name doesn’t ring a bell, Xeon D is a line up of high-end processors built for high-density servers. The newest addition to the lineup is the Xeon D-2100.
The Xeon D is the go-to chip for many data centers that want to bring their power as close to end-user devices and sensors as possible in order to keep TCO and application latency under check.
The original Xeon D processors started out with just eight cores before being upgraded to 16 cores in the follow-up Xeon D 1500 series. And now, the latest product line refresh has pushed things up a notch by equipping the brand new Xeon D-2100 series with 18 cores and 36 threads.
Here are some of the key specs from the new lineup:
- 14nm process technology
- Up to 18 cores, 36 threads
- Up to 512 GB DD4-2666 ECC
- Up to four channels memory support
- Intel Mesh Architecture
- Rebalanced Intel smart cache hierarchy
- Intel Hyper-Threading Technology (Intel HT Technology)
- Intel Virtualization Technology (Intel VT Technology)
- Intel Ethernet, up to four 10 GbE adapters
- Up to 3.0 GHz with single-core Intel Turbo Boost Technology 2.0
- Up to 32 PCI Express 3.0 lanes
- Up to 20 lanes of configurable Flexible High Speed I/O
- Intel AVX-512 acceleration with up to 1 FMA
- Intel QuickAssist Technology with up to 100 Gpbs of crypto, encrypt, and decrypt accelerated processing offload
Needless to say, it’s a big leap forward for the Xeon family in terms of TDPs and overall capabilities. Among other things, these upgraded specs have brought Xeon D processors closer to the Xeon E5, Xeon E3 v5, and the new Xeon Scalable families.
If you are operating a data center with limited space and power, the Xeon D 2100 series will most likely grow on you. Granted, these new chips fall short of the per-core clock speeds of the “proper” Xeon processors, but they bring many more cores than the E3 family, whereas the price and the TDP are much less compared to the E5 lineup.
How does it affect the performance of your server infrastructure?
Well, among other benefits, the lower power cores with less clock speed are surprisingly efficient with burstable tasks like serving up web pages where the system can generate many threads simultaneously and maintain them over a prolonged period without requiring a high amount of processing power. As soon as the server gets requests for new pages, the cores can automatically turbo boost to competently meet that demand.