NEW YORK -- AMD is rolling out its server CPU platform based on ARM’s 64-bit technology -- the first of such collaboration between the two firms. The Opteron A1100 series chip, codenamed Seattle, is fabricated using 28-nm process technology, and the chip maker plans to sample it this quarter.
In addition, AMD is planning to contribute to the Open Compute Project with a new micro-server design that utilizes the Opteron A-series, along with other architecture specifications for motherboards that Facebook helped developed called "Group Hug," an agnostic server board design that can support traditional x86 processors, as well as ARM chips. The company says it is looking to collaborate with others in the industry to create a software ecosystem around 64-bit for ARM-based designs in order to target essential workloads in web-tier and storage datacenters. The Opteron A-series will be accompanied by a standard UEFI boot and Linux environment based on the Fedora Project, which is distributed by Linux and sponsored by Red Hat.
AMD detailed the two projects on February 12.
Suresh Gupalakrishnan, corporate vice-president and GM for AMD’s server business unit, told EE Times that the company is going to offer a development kit with tools and software so that users can port their software to 64-bit ARM technology, along with a server board for the Open Compute Project using this particular processor.
This move is being spurred by two distinct but converging trends. The first is the rush to mobile, especially smartphones in the mid-range of $50 to $100. The second is the need to store all the data people are downloading and uploading to their smartphones in ever complex and sprawling datacenters. Gupalakrishnan believes that a number of these tasks are best suited for ARM CPUs, including storage, big data analysis and messaging, or “collaboration” as he put it.
“As ARM comes into the server market, its success will depend on the software ecosystem surrounding it,” Gupalakrishnan said. “It has to be compatible with existing datacenter infrastructure. It needs stable and predictable functionality. We are bringing our expertise in the x86 server market into the ARM market so that end customers can use ARM and X86 in the same datacenter.”
Gupalakrishnan lauded ARM’s servers for being especially compact in their implementation.
“It allows us to pack a lot of servers into a given space,” Gupalakrishnan said. In addition, ARM is the right choice at this moment for AMD because of the broader range of options it offers. “There are a lot of customers asking for a choice other than x86,” he said. “They are looking for a wider variety. Also, ARM allows you to customize processors at a faster rate than with x86.”
As for how all of this affects competition with Intel, Gupalakrishnan skirted the question, avoiding mentioning AMD’s nemesis by name, but addressing the substance of the rivalry nonetheless with a bold yet tentative prediction.
“We are going to continue making x86 processors,” he said. “We project that ARM will have 25 percent unit volume in the server market in 2019. So that means, even by then, 75 percent of the market will still be x86 processors.”
As for the overall direction of the server market, Gupalakrishnan predicted single-digit growth for the foreseeable market. Within those parameters, however, he expects double-digit growth for a select few segments, specifically storage, big data, and cloud servers.
— Zewde Yeraswork, Associate Editor, EE Times