LONDON — ARM Holdings plc (Cambridge, England) has fired off the latest salvo in a battle against Intel Corp. for domination of the world's microprocessor slots and sockets.

The launch of a dual-core Cortex A9 processor hard-macro called Osprey, ready for manufacture by TSMC in a 40-nm general-purpose process, with plenty of comparisons to the current generation of Intel Atom processor, signposts ARM's aspirations — to take Intel on in its traditional performance domain.

Osprey is also clearly aimed at the netbook space, but the argument that by designing in TSMC's 40G process ARM is not going to tread on the toes of some of its semiconductor partners, who are getting ready launch their own Cortex-A9 chips in 40-nm low-leakage processes, is not clear cut.

According to Eric Schorn, vice president of marketing for ARM's processor division, the semiconductor partners who have already licensed Cortex-A9 in soft form are aiming at "wireless."

Those licensees include Texas Instruments, STMicroelectronics, Samsung, NEC, Renesas, Toshiba and NXP. But any notebook, netbook, smartbook, or indeed just about any computing device these days needs to be connected to the internet. And mobility demands no wires, which means wireless and low power consumption in the processor and every other aspect of the design. The distinction would appear to be subtle if not moot.

Now ARM is bringing forward Osprey so companies will be able to go to TSMC and quickly develop system chips aimed at higher performance applications. Perhaps the key can be found in Schorn's statement "We are into unlocking some new markets; netbooks, smartbooks, MIDs, consumer electronics in TV and entertainment devices, and enterprise networking, such as things like printers."

One interpretation could be that the existing soft, configurable and architectural ARM core licensees are not unlocking those markets for ARM, or at least not yet. Meanwhile the Cortex-A9 hard macro already has it first licensee "extending the Cortex-A9 into new markets with significant performance uplift."

In other words, if the fabbed, soft core licensees are not going to make a dent in Intel's netbook processor progress then perhaps some entrepreneurial chip companies and system houses taking a hard-core will? ARM is backing both horses.

And if we take ARM's numbers and benchmark assertions at face value and conclude it has an Atom-killer we must also remember that ARM is comparing an EDA simulation of a 40-nm processor core against a 45-nm/40-nm Intel processor that is in the field.

ARM's Osprey has not gone through TSMC as a system test chip yet. So by the time those Osprey processors are also in the field Intel will have, no doubt, fired its riposte.

Intel will likely have done what it always has done. Turn the handle on its fearsome manufacturing machine to drive the high-priced state-of-the-art down to 32-nm thereby allowing it to drop the price at the 40-nm node to undercut any Osprey user that dares to try and run against them.

As with all good sagas of battles between giants, the match is not won quickly. ARM's announcement has merely set the scene for battle on a broader front.

In enterprise and other tethered applications ARM's Osprey strategy may work. Linux is gaining ground in embedded applications. For the Mobile Internet Device ARM's success is inextricable from that of Windows, whose support it does not yet have, Linux and Android.

Related links and articles:

Opinion: In 'phoney war' Intel attacks ARM at home

ARM processor runs applications in some Intel-based PCs, claims executive

ARM processor-count can exceed that of Intel in modern PCs