 |  |  |  |  |  |  |  |  |  |  |  |  |  |
|
 What's happening to ARM: A RISC for all vendors? W
ith many big semiconductor players eyeing ARM 32-bit RISC microprocessor licenses these days, you may be wondering what that means in the big scheme of things. Two or three years ago, there were reports that the original licensees were worried about diluting the pool with too many competitors. How come it doesn't seem to be an issue any more?
There's no doubt that Robin Saxby, pr esident and chief executive officer of Advanced RISC Machines (ARM), has cleverly guided his company through the rapids--going from "a handful" of licensees to a dozen was tricky. But at 18 or 20? That shows, he says, that the ARM architecture is becoming a global standard. One way Saxby has maneuvered ARM is by raising awareness of the trading of intellectual property (IP), in the form of system-level macros. It has gained credibility through his company's own work and the creation of the Virtual Socket Interface Alliance. Semiconductor companies are realizing that it is not just what cores they have that make sales, it is what they do with them; how easily, completely and quickly they can get their customer to market. Major semiconductor companies are coming on board and established licensees cease to complain because--What can they do? Not having the ARM license could lock them out of some valuable business. With so much competition it might not be as much business as they once hoped, but it's s till valuable. Semiconductor companies are now feeling compelled to get an ARM license to fulfill customer requirements; an offering that will compete against other cores they have licensed or developed internally. In this context, NEC springs to mind, and it seems to be a similar case with Philips, Lucent, Motorola and, possibly, SGS-Thomson, if these companies all take licenses. But there's the rub. As the "Thumb" core, for example, becomes a global standard for mobile-phone handsets, it is ceasing to be a differentiator for those semiconductor makers who have access to it. It may be necessary to have the Thumb on their books to be in the bidding, but they may win the business--and be able to charge a premium price-because of some other technology or service. In the dim and distant early 1980s, when the first version of the ARM RISC was designed to power an Acorn personal computer, a 32-bit RISC was a pinnacle of electronic-engineering achievement. Now it is being reduced to a familiar pre requisite. In the early 1980s, every major semiconductor maker had paid the appropriate licenses and had a family of 74-series logic parts in its data books--sound familiar?. So, in these days of the IP revolution, expect semiconductor vendors to take licenses on multiple RISC cores in many different flavors linked to many different OSes and RTOSes and optimized for different applications. The good news for ARM is that the ARM architecture is not just one core, but a family of cores: ARM7, Thumb, ARM8, StrongARM. The race moves on, and what remains for ARM to do is to show significant gains in key performance parameters in future cores--ARM9, ARM10--and perform the trick all over again. And, of course, there are other avenues for ARM to pursue, such as licensing of software and hardware-software combinations.
|
| ||||||||||||||||
