Qualcomm Takes on Intel With Server SoC

Qualcomm has introduced its long-awaited server SoC, an ARM-based processor with which it hopes to challenge Intel's x86 server silicon hegemony. Xilinx and Mellanox are providing complementary silicon.

Brian Santo, Senior editor, Test & Measurement / Components, Light Reading

October 9, 2015

3 Min Read
Qualcomm Takes on Intel With Server SoC

Qualcomm, allied with fellow chip vendors Xilinx and Mellanox, has introduced its first IC for the server market, a 24-core system on a chip.

The SoC is based on the ARMv8-A instruction set, and constitutes the physical heart of what Qualcomm Inc. (Nasdaq: QCOM) is calling its Server Development Platform (SDP).

Qualcomm has entered long-term relationships with Mellanox Technologies Ltd. (Nasdaq: MLNX) and Xilinx Inc. (Nasdaq: XLNX) to provide silicon that complements its server SoC. Mellanox is providing Ethernet and InfiniBand interconnect optimized for scalable server and storage infrastructures. Xilinx, meanwhile, is going to provide FPGAs that will enable rapid development of a range of products, ranging from card level to highly integrated solutions. Target applications include compute acceleration, big data analytics, machine learning, storage and CloudRAN, the companies said.

Qualcomm signaled its intent to enter the server market last year. Facebook was identified as a customer interested in seeing what Qualcomm could come up with.

Entering an entirely unfamiliar market carries risks. But Qualcomm is dominant in the market for ARM chips used in cellphones, and it is riding a trend toward the use of ARM-based processors in the server market. Furthermore, it is among the small handful of silicon vendors with the heft to challenge Intel Corp. (Nasdaq: INTC), which utterly dominates the multi-billion-dollar server chip market with processors based on the x86 architecture, which has figuratively defined PCs and servers literally for decades.

Want to know more about the New IP? Check out our dedicated New IP content channel here on Light Reading.

Data centers are designed to minimize energy consumption, but that's because they already suck tremendous amounts of juice (and generate ungodly amounts of heat). Anything to further reduce consumption is of interest, explaining why the industry has been exploring the use of ARM-based processors in data center servers.

Cellphones are notoriously power-constrained; efficient power management is essential, and thus power management has become a defining characteristic for ARM-based processors. More and more companies are coming to believe that applying that capability to server systems might be a useful thing to do.

Marvell Technology Group Ltd. (Nasdaq: MRVL) was among the first silicon vendors to introduce ARM-based server chips in 2012. In just the past year, Lenovo Group Ltd. (Hong Kong: 992), HP Inc. (NYSE: HPQ) and Dell Technologies (Nasdaq: DELL) all said they are exploring building servers around ARM-based processors.

Qualcomm is targeting its SoCs at hyperscale data centers. (Because who in the server business isn't targeting hyperscale data centers?) Data center applications include infrastructure as-a service (IaaS), platform as-a service (PaaS), big data and machine learning.

Part of the announcement of the chip was a demonstration of the SDP running the entire software stack: Linux kernel version 4.2 along with KVM virtualization, OpenStack DevStack for OpenStack cloud orchestration, guest virtual machines running a standard Linux distribution along with Apache web server and WordPress.

The 24-core SoC was built using FinFet technology. It integrates all of the standard server-class features, including PCIe and storage, the company said. It was designed by the Qualcomm Technologies unit.

Qualcomm said this processor is only the first in a line of production single-chip server SoCs that will include a fully custom core built on FinFet technology.

Mellanox said its 10, 25, 40, 50, and 100Gbit/s Ethernet and InfiniBand products, when paired with Qualcomm Technologies' server CPU, will enable data centers to scale out with efficient, intelligent building blocks, to meet the needs of enterprise, cloud, Big Data, Web 2.0, and storage workloads.

Xilinx said its FPGAs in combination with the Qualcomm SoC should enable customers to gain performance, performance/watt and lower total cost of ownership.

— Brian Santo, Senior Editor, Components, T&M, Light Reading

About the Author(s)

Brian Santo

Senior editor, Test & Measurement / Components, Light Reading

Santo joined Light Reading on September 14, 2015, with a mission to turn the test & measurement and components sectors upside down and then see what falls out, photograph the debris and then write about it in a manner befitting his vast experience. That experience includes more than nine years at video and broadband industry publication CED, where he was editor-in-chief until May 2015. He previously worked as an analyst at SNL Kagan, as Technology Editor of Cable World and held various editorial roles at Electronic Engineering Times, IEEE Spectrum and Electronic News. Santo has also made and sold bedroom furniture, which is not directly relevant to his role at Light Reading but which has already earned him the nickname 'Cribmaster.'

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like