In enterprise circles, nearly all eyes are on the cloud to see whether it will actually live up to its promises of a more flexible and cost-effective IT environment.
But for those whose lives revolve around infrastructure, there is a different revolution taking place. ARM processors are poised to move into virtually every area of the enterprise, most notably servers, which have been dominated by the x86 platform for some time.
The movement got a kick-start last month with word that Microsoft is about to launch an ARM-friendly version of Windows-albeit in a move to capture more of the smartphone/mobile market rather than any fear that it was about to lose its grip on servers. Still, the same basic question remains: Can the technology deliver?
It can if it's coupled with that other processor that's been working its way into enterprise data settings: the GPU, according to Nvidia. The company has just taken the covers off Project Denver, a move to integrate an ARM processor, most likely the Cortex-A9, with a graphics co-processor. The intent is to outclass both the x64 and x86 architecture in terms of processing ability and efficiency.
As Nvidia Chief Scientist Bill Dally explained it to TG Daily:
A high-performance CPU with a standard instruction set will run the serial parts of applications and provide compatibility while a highly-parallel, highly-efficient GPU will run the parallel portions of programs. The result is that future systems - from the thinnest laptops to the biggest data centers, and everything in between - will deliver an outstanding combination of performance and power efficiency. Their processors will provide the best of both worlds, while enabling increased battery life for mobile solutions.
But servers aren't the only boxes likely to be caught up in the ARMs race. Networking is also a wide-open field. Cavium Networks is also looking to the Cortex A family for a new generation of SoC solutions intended to power home, enterprise and carrier broadband devices. Matching the lower-powered Cortex with Cavium's ECONA platform is expected to drive processing into the GHz range while pushing power consumption under the 1 watt threshold.
The buzz surrounding enterprise use of ARM technology is hard to ignore, and on paper the technology appears ready for heavy data environments. But there is one nagging problem: the 32-bit RISC architecture. A 40-bit memory system is in the works, but we're not likely to see true parity with x86 machines until a 64-bit version is available.
For the moment, ARMs can make up for that deficit through sheer numbers. Multiple ARMs can still operate under the power envelope of a single Core2 or Opteron, even with the addition of a GPU to handle the parallel processing requirements. But as you continue to add complexity to the basic ARM architecture, is there a danger of jeopardizing the very advantages that ARMs provide in the first place?