Today Intel has launched their lower-power series of microprocessors based on the Skylake server microarchitecture. The Xeon D-2100 series is very similar to the Skylake-SP (Xeon Scalables) processors but are specifically designed for dense edge networking, IoT and cloud servers.
|Intel Xeon D-2100 Series|
|Model||Cores||Base||Turbo (All)||Turbo (Max)||TDP||Price|
|Edge Server and Cloud SKUs|
|D-2191||18||1.6 GHz||2.2 GHz||3.0 GHz||86 W||$2,407.00|
|D-2161I||12||2.2 GHz||2.8 GHz||3.0 GHz||90 W||$962.00|
|D-2141I||8||2.2 GHz||2.7 GHz||3.0 GHz||65 W||$555.00|
|Network Edge and Storage SKUs|
|D-2183IT||16||2.2 GHz||2.8 GHz||3.0 GHz||100 W||$1,764.00|
|D-2173IT||14||1.7 GHz||2.3 GHz||3.0 GHz||70 W||$1,229.00|
|D-2163IT||12||2.1 GHz||2.6 GHz||3.0 GHz||75 W||$930.00|
|D-2143IT||8||2.2 GHz||2.7 GHz||3.0 GHz||65 W||$566.00|
|D-2142IT||8||1.9 GHz||2.5 GHz||3.0 GHz||65 W||$438.00|
|D-2123IT||4||2.2 GHz||2.7 GHz||3.0 GHz||60 W||$213.00|
|Integrated Intel QuickAssist Technology SKUs|
|D-2187NT||16||2.0 GHz||2.4 GHz||3.0 GHz||110 W||$1,989.00|
|D-2177NT||14||1.9 GHz||2.3 GHz||3.0 GHz||105 W||$1,443.00|
|D-2166NT||12||2.0 GHz||2.3 GHz||3.0 GHz||85 W||$1,005.00|
|D-2146NT||8||2.3 GHz||2.5 GHz||3.0 GHz||80 W||$641.00|
|D-2145NT||8||1.9 GHz||2.5 GHz||3.0 GHz||65 W||$502.00|
No low-power SKUs?
One thing that sticks out is the fact that there are actually no parts under 60 W. While those parts are in fact the lowest-power Skylake-based server processors, prior Xeon D generation (based on Broadwell) featured parts with a TDP as low as 20 W with most parts being around 35-45 W. This is a rather large change. It’s possible we might see lower TDP SKUs later on, but for the time being there is a missing upgrade path for many of those lower TDP parts.
Like prior generation of Xeon D processors, those new processors are also a single-chip solution. That is, both the chipset and the processor die is incorporated into the same package, resulting in a fairly dense server solution.
In addition to the 32 PCIe lanes from the microprocessor itself, provided from the chipset is 20 configurable high-speed I/O (HSIO) lanes. Those lanes can be confirmed as either up to 20 PCIe lanes, up to 14 SATA 3.0 lanes, or up to 4 USB 3.0 ports.
For the most part, all of those Skylake DE parts have all the features enabled. The main differentiation is frequency and core count. Where the segmentation comes in is in the application. All the SKUs can have one of the three suffixes:
- I – Integrated Ethernet. Those SKUs have four 10 Gbp Ethernet ports (just requires an external PHY)
- T – High Temperature. Those SKUs have a package TCASE rating of 90 °C
- N – Integrated Ethernet & QuickAssist Technology. Those SKUs have both integrated Ethernet as well as Intel’s QuickAssist technology
All the SKUs support 512 GiB of four channels ECC DDR4 memory which is double the memory channels and more than double the bandwidth versus previous generations.
Being based on the Skylake server microarchitecture means those chips do use the new Mesh architecture. All parts also have AVX-512 support but only support a single FMA unit.
Performance Claims Fine Print
As usual, we do not bother with reporting the manufacturer performance claims. However, one thing did catch our attention.
Benchmark results were obtained prior to the implementation of recent software patches and firmware updates intended to address exploits referred to as “Spectre” and “Meltdown”. Implementation of those updates may make these results inapplicable to your device or system.
This statement has started appearing in all recent Intel presentations since their disclosures. Performance claims were done before applying the recent Spectre and Meltdown mitigations to which those parts are inherently vulnerable to.
- ISSCC 2018: Intel’s Skylake-SP Mesh and Floorplan
- Intel Cascade Lake Brings Hardware Mitigations, AI Acceleration, SCM Support
- Intel Discloses 9th Gen Core, Refreshes Core X, And Reintroduces STIM
- ISSCC 2018: Intel’s Self-Powered Intelligent IoT Edge Mote
- Intel to launch Skylake-D in Q1 2018, followed by Xeons with integrated FPGA
- Intel Core i9-9900KS Special Edition Full Specs and Availability Announced
- Intel Launches Mainstream Cascade Lake Xeon W Workstation Processors
- NEC Refreshes SX-Aurora Vector Engine, Outlines Roadmap
- Japanese AI Startup Preferred Networks Designed A Custom Half-petaFLOPS Training Chip
- Samsung M5 Core Details Show Up
- SC19: Aurora Supercomputer To Feature Intel First Exascale Xe GPGPU, 7nm Ponte Vecchio
- A Look at Cerebras Wafer-Scale Engine: Half Square Foot Silicon Chip
- Groq Tensor Streaming Processor Delivers 1 PetaOPS of Compute
- Arm Makes Headway In HPC, Cloud
- Intel Announces Keem Bay: 3rd Generation Movidius VPU
- A Look at Spring Crest: Intel Next-Generation DC Training Neural Processor
- Marvell Lays Out ARM Server Roadmap