Check Out These Hot Products Using New Skylake Purley Intel Xeon Scalable Processor
Intel's New Skylake Purley Intel Xeon Scalable Processor
Intel rolled out its new Skylake family of server processors aimed specifically at data center applications with the new mesh-based Purley architecture to reduce latency at high core counts.
The new family, the Intel Xeon Scalable Platform, is also at the center of a wide range of new data center servers offering new levels of performance, density, and scalability. Vendors with new Skylake servers ranged from the top-three branded companies to specialty custom system builders.
But the Skylake processors are not just about servers. Several other IT vendors in the storage and cloud market segments jumped on the Intel Xeon Scalable Platform bandwagon this week as well.
Intel expects its new processors will help customers and partners find new ways to transform data centers to meet the requirements of new cloud, networking, and artificial intelligence applications. Here's a look at what Intel's vendor partners are doing.
Targeting Modern Data Centers
The new Intel Xeon Scalable Platform processors feature a new mesh architecture that offers improved connectivity between processor cores than the ring architecture that has been a feature of Intel's data center processors since 2009, Intel told CRN.
At the high end of the family, the new processors offer 28 cores per socket, and support up to eight sockets with up to three Intel UPI (UltraPath Interconnect) uplinks. They also support up to 1.5 TB of 2,666MHz DDR4 memory.
Options that can be integrated into the processors include the Intel Omni-Path Architecture, a high-speed proprietary fabric for high-performance and low-latency applications; Intel Ethernet connectivity; and the Intel QuickAssist AVX-512. Also available are Intel Optane SSDs based on the 3D XPoint memory technology jointly developed by Intel and Micron Technology.
Intel said it has already shipped 500,000 production processors to over 30 customers including OEMs, ODMs and cloud service providers.
Amax Matrix On-Premise GPU Cloud
Matrix is Fremont, Calif.-based system builder Amax's platform for artificial intelligence development and deployment. Powered by Bitfusion Flex, Matrix includes key components of artificial intelligence environments including deep learning frameworks and GPU virtualization technology. Amax said Skylake-based Matrix can pull the latest containers into clusters with a single command without disrupting running workloads, and are fully integrated with Jupyter Notebook to create Interactive workspaces. It also includes GPU over Fabrics technology for sharing and scaling of large numbers of GPUs across systems, and dynamic allocation of GPUs across multiple jobs and users.
Cisco Unified Computing System M5
Cisco Systems introduced its Cisco UCS M5, a new generation of servers based on the Intel Xeon Scalable Platform. The Cisco UCS M5 offers a performance boost of up to 86 percent over previous generations, the company said.
The M5 generation of servers include the half-width B200 M5 blade server, the B480 M5 full-width blade server, the C240 M5 two-socket rack server, the C240 M5 storage-optimized and I/O-optimized rack server, and the C480 modular rack server.
Cisco also unveiled enhancements to its software, including Cisco UCS Director 6.5, which extends automation capabilities beyond infrastructure with new automation, including for FlexPod converged infrastructure and Cisco HyperFlex hyper-converged infrastructure, and Workload Optimization Manager, which uses intent-based analytics to match workload demand to infrastructure supply.
Dell EMC PowerEdge 14th Generation Servers
Dell EMC is preparing to ship the PowerEdge 14th generation server family it first previewed at Dell EMC World 2017. Dell EMC said the new servers, based on the Intel Xeon Scalable processors, are optimized for NVMe-based storage and targeted at such applications as software-defined storage and hyper-converged infrastructure thanks to significant latency decreases in VMware VSAN clusters, high-speed seamless live virtual machine migration, improved storage performance and integrated security.
Models include the PowerEdge R640 for dense scale-out computing and storage in a 1U, two-socket chassis; the PowerEdge R740 with balanced storage and application acceleration in a 2U, two-socket chassis; the PowerEdge R740XD storage-optimized system in a 2U, two-socket chassis; the PowerEdge R940 mission-critical server in a 3U, four-socket chassis; the PowerEdge M640 and FC640 modular servers; and the PowerEdge C6420 performance server in a 2U, eight-socket chassis.
DigitalOcean High CPU Droplets For Cloud-Based Developers
DigitalOcean, a New York-based developer of cloud for developers, released High CPU Droplets featuring the new Intel Xeon Scalable processors for compute-intensive applications requiring the highest levels of CPU performance and reliability. The High CPU Droplets provide maximum access to the new CPU resources to support applications such as ad exchanges, game servers, highly active front-end application servers, and queue/batch processing, the company said.
DigitalOcean is launching five new High CPU Droplet plans that start at $40 per month, or $0.060 per hour. They scale from two to 32 virtual CPUs with 3 to 48 GB of RAM and 20 GB of local SSD storage. The company also provides standard Droplets for deploying typical applications, as well as High Memory Droplets for running large-scale databases or distributed in-memory caches.
Gigabyte High Performance Computing Servers
Taiwan-based Gigabyte, with U.S. operations in City of Industry, Calif., unveiled a new generation of servers based on Intel's Skylake Purley architecture targeting applications from the enterprise to the data center to high- performance computing.
Initially, Gigabyte will offer four new 1U and four new 2U- servers as well as two new motherboard SKUs that support the Intel Xeon Scalable processors with a range of options for storage and expansion slots. The company's value-add to the new Intel-based platform includes uniform backplanes that support exchangeable expanders offering SAS and/or U.2 connectivity, along with high thermal ratings and 80+ Platinum or better power supplies.
Hewlett Packard Enterprise ProLiant Gen10, Synergy Gen10
HPE, which in June previewed its HPE ProLiant Gen10 Servers and HPE Synergy Gen10 compute modules, will base them on the new Intel Xeon Scalable processors. The company claims them to be the first industry-standard servers with built-in silicon-based security, which gives customers the ability to address firmware attacks. This includes designing security directly into the iLO chip to create an immutable fingerprint in the silicon.
HPE is also providing new management software including HPE OneView 3.1 to automate complex tasks and simplify life-cycle operations, HPE Intelligent System Tuning to better tune the server to match workload profiles, HPE Scalable Persistent Memory for improved performance, and HPE Flexible Capacity and HPE Capacity Care Service to let companies to pay based on server utilization.
Huawei FusionServer V5
China-based Huawei unveiled its FusionServer V5 portfolio based on Intel Xeon Scalable processors. The FusionServer V5 family includes modular servers such as traditional two-socket to eight-socket rack servers, the X series high-density servers, and the E series blade servers. Also included is the company's G series of heterogeneous computing platforms.
Huawei FusionServer V5 servers are engineered to be automatically adaptive, and were upgraded with five major smart features in computing, storage, network, management and energy efficiency to meet what the company termed the flexible demands of various scenarios while delivering industry-leading computing and operations and maintenance efficiency.
IBM Brings Intel Xeon Scalable Processors To Cloud
IBM said it expects to be the first major cloud provider to launch bare-metal servers powered by new Intel Xeon Scalable processors globally. The company introduced new IBM Cloud bare metal servers powered by the Intel Xeon Silver 4110 processor and Intel Xeon Gold 5120 and 6140 processors targeted at delivering high performance and fast insight from big data workloads. In addition to increased performance, the new bare metal deployment options provide a dedicated, security-rich environment customizable for sensitive big data workloads, IBM said.
IBM Cloud bare-metal servers powered by Intel Xeon Scalable processors are slated to be available in IBM Cloud data centers in the U.S., U.K., Germany and Australia in the third quarter.
Kaminario All-Flash Arrays
Kaminario, a Needham, Mass-based developer of all-flash storage arrays, is building its next generation of K2 all-flash storage array controllers to incorporate a module powered by the Intel C6268 chipset with Intel Quick Assist Technology, or QAT. This configuration lets the Kaminario K2 offload inline compression from the new Intel Xeon Scalable processors to a dedicated compression engine.
Kaminario said the configuration increases the data reduction ration by 30 percent with no impact on latency, enhances the level of ensured effective capacity as part of the company's ForeSight business assurance program, and lets the server recapture CPU cycles that can be used to handle more data and customer-facing services.
Lenovo ThinkSystem
Lenovo's new ThinkSystem portfolio, based on Intel Xeon Scalable processors, includes 14 configurable servers across rack and tower, mission-critical, dense, and blade platforms. The company is aiming them at such challenging workloads as real-time analytics, cloud, high-performance computing and software-defined architectures.
Lenovo expects its ThinkSystem portfolio of servers, storage and networking offerings to provide streamlined IT infrastructure and lead to increased service levels of data center operations that tie directly to business growth.
Maxta Supports Hyper-Converged Infrastructure On The New Processors
Santa Clara, Calif.-based hyper-converged infrastructure software developer Maxta unveiled support for any server designed for Intel Xeon Scalable processors as a way to bring the performance of the new processors to HCI to customers who cannot wait for hardware-based offerings.
The company claims a Maxta HCI cluster configured with new Intel Xeon Platinum 8168 processors, Intel data center SSDs, and NVMe connectivity was benchmarked to offer a performance gain of 120 percent in IOs per second with under half the latency when compared with the previous generation of Intel technology.
Nor-Tech Voyageur
Nor-Tech, a Burnsville, Minn.-based custom system builder with a focus on high-performance computing, is using the new Skylake processors in a wide range of HPC offerings based on its new Voyageur servers and Voyageur clusters.
According to Nor-Tech, the new Voyageur Skylake Clusters feature six 2.5-inch hot-swap drive bays per node, up to 2 TB of DDR4-2,666 memory in 16 DIMM slots, two PCIe slots and an optional NVMe slot, either Intel’s Omni-Path fabric or Mellanox FDR InfiniBand, and 2,200-watt Titanium-level redundant power supplies.
Quanta Second-generation Servers
Quanta Cloud Technology (QCT), a San Jose, Calif.-based part of Taiwan-based Quanta Computer focused on data center and cloud hardware and software, unveiled a new line of servers based on the Intel Xeon Scalable processors.
The company introduced four new servers including the QuantaGrid D52B-1U, a 1U, two-socket model with up to five PCIe expansion slots for high-performance storage; the QuantaGrid D52BQ-2U, a two-socket rackmount server for software-defined workloads; the QuantaPlex T42S-2U four-node server, which each computing node supporting up to 16 memory modules and four storage drives; and the QuantaPlex T42SP-2U four-node server with NVMe support. QCT also plans to introduce storage servers, four-socket servers, high-density multi-node servers, and artificial intelligence acceleration servers.
Supermicro X11 Server and Storage Offerings
Supermicro, San Jose, Calif., is basing its new X11 server and storage gear on the new Intel Xeon Scalable processors backed by support for NVMe storage and 100G/25G Ethernet.
The company's new portfolio includes its highest-performance 1U and 2U UltraSuperServers; its BigTwin high-performance, high-density 2U, four-node servers supporting 24 DIMMs and six hot-swap NVMe drives per node; its 4U FatTwin with a variety of configurations optimized for cloud, high-performance computin, and enterprise applications; its SuperBlade two-socket and four-socket blade servers with NVMe for enterprise, cloud and HPC applications; new SuperServers with GPUs for artificial intelligence and deep learning; SuperStorage arrays including 45 or 60 bays in a 4U chassis; and other new storage and server offerings.
Tyan High Performance Computing Platforms
Tyan, the Fremont, Calif.-based subsidiary of Taiwan's Mitac, introduced a series of new motherboards servers and storage appliances based on the Intel Xeon Scalable processors. The devices are targeted at HPC, machine learning and technical computing workloads.
Included in the lineup are single-socket and dual-socket motherboards, single-socket and dual-socket servers with up to eight GPUs, a two-socket server node targeting hyper-converged infrastructure and HPC workloads, other servers for data center deployments and IoT workloads, and a number of storage appliances.
Vexata VX-100 Memory-class Array
Vexata, a San Jose, Calif.-based developer of enterprise storage systems, unveiled new versions of its Vexata VX-100 arrays that will integrate the Intel Xeon Scalable processors to improve throughput, IOPS and latency performance compared with its arrays based the previous generation of Intel Xeon processors.
The new arrays take advantage of the new processors with up to 28 cores per chip, increased number of memory channels, memory running at up to 2,666 MHz, and Intel Optane SSDs to offer up to 7 million IOPs and 80 GBps of bandwidth at 40 microseconds latency, the company said.