Sashkin - Fotolia
Cisco, Dell EMC and Super Micro Computer Inc. pumped out server upgrades this week, piggybacking on Intel's launch of new processors, storage class memory cards and networking products.
Intel positioned its Tuesday product blitz as a "data-centric" approach to help data centers of varied sizes implement AI at scale. The x86 processor giant took the wraps off its second-generation Xeon Scalable processors, based on the Cascade Lake chipset.
More than 50 standard Intel Xeon Scalable SKUs were introduced, led by the Platinum 9200 flagship, which scales to 56 CPU cores, or 112 threads in a two-socket server. The latest Intel processors embed Intel Deep Learning (DL) Boost software on the chip to accelerate machine learning.
DL Boost is built for Intel Optane dual inline memory modules (DIMMs), which integrate 3D XPoint hybrid dynamic RAM and NAND flash on the same device.
"The addition of machine learning [capabilities] into the chip ... is good for specific inference workloads," such as recommendation engines, said Patrick Moorhead, principal analyst at Moor Insights and Strategy.
The Intel Xeon Scalable processors are available with 32 cores, 48 cores or 56 cores. The high-density Platinum 9282 supports 3 TB of DDR4 memory storage across two sockets.
Along with the new Xeon Scalable processors, Intel introduced the Xeon D-1600 high-density system on a chip, its first internally developed field programmable gate array (FPGA), and expanded its line of Optane DC Persistent Memory DIMMs.
Intended use cases for Xeon D-1600 devices include control planes, security and midrange storage in edge environments with limited energy and data center space.
Intel Agilex FPGAs are designed with 10-nanometer process technology and a multi-die interconnect bridge. The design allows other FPGAs, CPUs or accelerators to be integrated in Agilex devices, said Dan McNamara, general manager of Intel's programmable solutions group.
"That allows us to bring heterogeneous functionality onto one device," McNamara said.
Navin Shenoy, an Intel executive vice president of mobile and client platforms, said Xeon Scalable hardware optimizations are designed to defuse side-channel security vulnerabilities, such as Meltdown and Spectre variants, which could potentially expose data.
Dell EMC said it will start selling 14th Generation PowerEdge storage servers with Intel Xeon Scalable and Optane support later this month. Dell EMC enhanced its OpenManage software platform for managing PowerEdge clusters.
New OpenManage features include prescriptive security, enhanced encryption keys and augmented reality mode for threat detection. The OpenManage tool set allows each server's BIOS to be tuned for different workloads.
Customers have been slow to go full-bore on AI adoption, largely to avoid the cost of adding compute servers, said Ravi Pendekanti, a Dell EMC senior vice president of product management for servers and infrastructure systems.
"We think the new Intel Xeon Scalable processors and Optane persistent memory will help more of our customers make that transition," Pendekanti said.
Cisco will also support Xeon Scalable processors on Unified Computing System (UCS) servers, which provide compute for Dell EMC VxBlock and NetApp FlexPod converged systems. Intel's new chipset in UCS helped improve clock speed and memory performance, Cisco's senior product marketing manager, Bill Shields, wrote in a blog post.
White box server-maker Super Micro Computer said it rolled out more than 100 configurations for its X11 line of servers and storage systems. Hewlett Packard Enterprise postponed announcing similar planned upgrades to its ProLiant servers.