Built-In Intel Omni-Path Architecture
Intel’s Omni-Path Architecture is an end-to-end high-bandwidth, low-latency fabric that optimizes performance to handle HPC workloads and can scale to thousands of nodes and beyond. Intel Xeon Scalable Processor-based servers are the first general purpose platforms to offer integrated Intel Omni-Path Fabric, bypassing the need to utilize valuable PCIe lanes. With the ability to operate at 100Gb/s and direct interface into the CPU, Omni-Path is ideal to deploy wherever high-speed fabrics are needed.
Optional integrated network and built-in accelerators
The Xeon Scalable Processor provides up to four high-speed 10 GbE Intel Ethernet ports and integrated QuickAssist Technology (QAT) for high data throughput and low-latency workloads. Intel’s QuickAssist Technology is designed to provide both security and compression acceleration capabilities, while improving performance and efficiency on Intel architecture platforms. QAT dramatically increases the speed of compression and AES encryption by offloading these computationally intensive functions from the processor to a purpose build accelerator.
Increased performance with NVMe
The new Intel Xeon Scalable processors focus on NVMe by removing the legacy bottlenecks inherent to SATA and SAS interfaces allowing the SSDs in each system to reach their full potential. Servers built on Intel Xeon Scalable processors provide the SSDs the headroom they need via a direct PCIe interface to the SSDs and a new block level protocol for non-volatile storage media. This is made possible by the increased number of PCIe lanes to accommodate additional NVMe drives allowing servers to run demanding workloads simultaneously with increased system utilization for greater responsiveness.
High-speed connectivity delivered
Intel’s C62x series chipset Platform Controller Hub (PCH), integrated into the Xeon processor, has 4 lanes of DMI 3.0, up to 10 USB 3.0 ports, 14 USB 2.0 ports, up to 14 SATA 3 ports, and 4 10Gb Ethernet ports. This allows the Xeon scalable processor to scale up its connectivity to unprecedented levels, bumping the core count up to 28 and threads up to 56. Since there are up to 4 10GB Ethernet ports native to platforms, Intel’s Xeon scalable processor family reserves PCI express lanes for NVMe. Open PCI lanes equal faster connectivity and faster Ethernet throughput.
Built-in Virtual RAID on CPU (VROC) enables faster RAID assemblies
VROC is a new RAID volume management software package released by Intel that is designed to take advantage of the Xeon Scalable Processor’s new connectivity features. With VROC, direct attached NVMe drives can be setup into a RAID stack run directly by the CPU. This innovation provides the most bandwidth, throughput, and compute power for storage, while eliminating the need for battery backup or super-cap memory backups for data write holes in the event of a power failure. The overall effect lowers costs and simplifies RAID volume management.
Radically increased responsiveness with Intel Optane memory
The biggest change to memory architecture in 25 years, Intel Optane boosts speed, endurance and memory density for demanding applications, without sacrificing storage capacity. Built on 3D XPoint technology and available only with NVMe or direct DDR4 interfaces, Intel Optane will enable up to four times the memory capacity available in a system relative to previous platforms reliant on DDR4 technology. Get ready for a new tier of non-volatile storage that blurs the line between memory and storage with Intel Optane and 3D XPoint.
Ready to build on Intel Xeon Scalable-based platforms?
UNICOM Engineering is ready to help you transition your next-gen solutions to Intel Xeon Scalable-based systems. Visit our Intel Xeon Scalable landing page or call a UNICOM Engineering expert at 781-332-1000 or toll free at 800-977-1010 to learn more.