facebook rss twitter

AMD and ARM to collaborate

Computer chip manufacturer AMD, founded in 1969, has been producing cutting-edge microprocessors for more than 40 years. AMD is the second-largest worldwide supplier of chips based on the x86 architecture and also one of the largest suppliers of graphics processing units marketed under the consumer Radeon and professional FirePro cards.

Though AMD has a rich heritage in producing CPUs based on the x86 architecture, including inventing the AMD64 (x86-64) 64-bit instruction set that is widely used by modern processors for efficient processing, the company is partnering with ARM Technologies to release a number of system-on-chip (SoC) processors ideally suited to entry-level server and workstation markets.

Why choose ARM?

The collaboration with ARM is the first time that AMD will not use the existing x86 architecture in executing server strategy. ARM chips are based on a different architecture called RISC (Reduced Instruction Set Computer) that, when compared to the incompatible x86, simplifies chip design and is ideally positioned to offer excellent performance-per-watt characteristics. It is no surprise that the vast majority of chips that power smartphones and tablets are also based on ARM technology.

The focus on energy efficiency has become more important in recent years, especially as the need to provide electricity is the largest cost of running servers and datacentres, which are specialised buildings holding hundreds of servers. Offering customers x86-like performance at a lower overall energy cost is a key selling point that AMD hopes it will leverage with its upcoming ARM-based chips.

The Seattle chips

Examining the details, AMD has announced that it will release the "Seattle" System-On-Chip (SoC) in the second half of 2014. Seattle is to use the forthcoming ARM Cortex-A57 64-bit architecture, with either eight or 16 cores integrated into an SoC. AMD expects that these SoCs will run at speeds in excess of 2GHz.

An SoC is more than just a processor of course, and AMD says Seattle SoCs will support up to 128GB of RAM, 10 Gigabit Ethernet, and integrate a new AMD technology called "Freedom Fabric," which enables a large number of servers or networking equipment to be easily connected into a many-machine cluster.

AMD is bullish on Seattle's pure performance, too, with the company saying that the ARM-based Seattle SoCs will offer up to 4x the performance of its recently-announced X1150 and X2150 "Kyoto" x86 server processors. These also-low-power chips are based on the Jaguar cores that will be used next-generation consoles such as the PlayStation 4 and Xbox One, as well as on AMD's current energy-efficient mobile chips.

The ARM gamble

A pioneer in the development of x86 chips, AMD sees significant, worthwhile merit in exploring new computing architectures for the lucrative server and workstation markets. Teaming up with ARM is a clear sign that the incumbent x86 architecture, championed by Intel and currently dominating the entry-level and mainstream server space, may not necessarily be the ideal fit for the kinds of energy-efficient chips actually required by the enterprise market.

Collaborating with ARM on a hardware level is only the first step, AMD understands, as enterprise software is currently optimised for the x86 architecture. AMD, ARM and their partners needs to encourage the larger enterprise-specific software companies to invest in optimising for the newer, simpler instruction set run on ARM chips. Only time will tell if the AMD/ARM cooperation, pregnant with untapped promise, delivers as expected.

Click here to read more about AMD APU