ARM's director of embedded solutions, Ian Ferguson, has spoken out about his company's work to get a chip more commonly associated with smartphones out of people's pockets and in to the data centre.
It's something ARM has been working on in the past: the company's chip designs offer a high performance per watt, making them extremely well suited to highly parallel tasks where a large number of processing cores is more important than the actual horsepower in each core.
ARM isn't the only company looking to break into this market, however: as Intel convinces its customers to try a many-core Atom system while it works on its 50-core MIC architecture, newcomers Tilera and Adapteva both offer a many-core low-power architecture of their own.
"The first consistent piece of feedback we hear from multiple points in the ecosystem is that one size no longer fits all," Ferguson writes in a post to ARM's Community Blog. "Fundamentally, server workloads do not all look the same. The mix of CPU performance, memory and IO bandwidth requirements vary by application.
"For many of these workloads, we believe the critical decision criteria is moving away from pure performance and instead viewing performance (of that particular workload) [per] watt [per] dollar," Ferguson writes - unsurprisingly, really, given his position in the world's largest designer of low-power processors.
"ARM has been looking at this area for a while," Ferguson points out, referring to an experiment in 2009 where a webserver was constructed from of-the-shelf hardware and software modules as a proof fo concept device. "Of course, much more work would need to be performed, from both hardware and software perspectives to construct a value proposition that was commercially viable."
Ferguson goes on to explain the work his team has been carrying out to further the adoption of ARM in data centre applications. Third-party companies are also heavily involved, he points out, menioning licensee Calxeda's work alongside Oracle's demonstration of a C2 Just-In-Time Java compiler for ARM - providing 20-40 per cent improved performance over the existing C1 JIT compiler at the cost of latency.
"This is important for applications like Hadoop, an application many people are seeing as a good fit for ARM Powered servers," Ferguson crows, before admitting that Oracle has been reticent to suggest a launch date for its C2 compiler.
"Pulling an ecosystem together does not happen overnight – it will take some time," Ferguson warns. "This initiative will provide the market place with a range of options and allow the market to select the solution that best aligns to their particular requirement. So it will be more than one size and more than one winner. We look forward to the road ahead."
ARM's move into the data centre will have chip rival Intel worried, even as it attempts its own assault on ARM's traditional market of embedded and portable computing.