"If you compare us with Intel we have 2000 people compared to Intel's 100,000. We have a small fraction of the revenue that they do," Iain Drew, ARM's EVP of strategy said at CES last week, repeating an oft heard pitch from the chip architect.
MOUNTAIN VIEW, Calif.-- On the one hand, ARM loves comparing itself to Intel, if only to whimper that the firm is so insignificantly tiny in terms of pure employee numbers.
On the other hand, ARM points out it has a completely different business model to Intel and ships significantly more products.
So, which is it? A tiny, no-budget company from Cambridge, or the architect IP brain behind an octopus-like network of partners which has its tentacles in all aspects of the embedded space?
A tiny virus can take down an elephant, but the point is well taken. ARM's true strength, however, is their ability to grow and use their partner relationships. When you add that to equation, ARM dwarfs everyone.
It is indeed the power of its network that can rival with Intel, not ARM on its own. That said, many people think ARM is much bigger (revenue and profit-wise) than it is in reality. Taken on its own, it is indeed a very small company compared with the likes of Intel, Google and Microsoft.
"So, which is it?" It's a case of comparing apples with oranges. ARM holds IP. Intel builds all manner of finished products.
But here's a thought. Remember back when there was this fierce battle between RISC and CISC? At that time, the battle between the two raged for products that looked essentially like desktop PCs, as well as for embedded applications. What happened then was that the performance of CISC chips became comparable with that of RISC, and the CISC won the war.
I think, this is very much the same sort of war. ARM is RISC, but its simplicity used to reduce power demands rather than to achieve greater performance. The move now goes to CISC, i.e. x86, to even that score.
This seems to be what Intel is claiming they have done. Last time they succeeded with performance. They may well make it with power requirements this time around. I wouldn't count them out.
Intel's problem is addiction to the higher margins of its processors. Its the same problem that has driven Kodak into bankruptcy - they had the effective monopoly of film processing with 70% margins:
Can we please put this "CISC won the war with RISC" myth to bed. Intel won by adopting a RISC core and translating x86 instructions into RISC "micro-ops".
So in fact RISC won the war by adding on the fly instruction translation.
Will Intel try to compete with ARM at all levels?
Even down to the ultra low power Cortex M0 level? Bring back the 386SX! :-)
Seriously though, I think the one big piece of disruptive technology could be the battery. Batteries just have to get better for the automotive market. Even 2X the storage of today's batteries could make a lot of ARM's power efficiency irrelevant, or at least buy extra peformance that Intel is better positioned to exploit.
Intel can obviously build a competent RISC core. I cant see that saddling it with the x86 instruction set brings anything of worth in the mobile market. With Android and Metro being the likely environments. It would be just as easy to develop new Android applications on a new Intel RISC platform. Unless you want to run Excel and Word on your phone. Perhaps a PC-Phone that docks or bluetooths to keyboard screen combo, or to a virtual window of the nearest PC?
The future belongs to multiprocessors, Tilera is making 64 core chips now. In that game, whoever can build cores with the least transistors wins by packing more of them on a die. That sounds more like ARM than Intel.
That future won't materialize until the software industry makes a breakthrough in multicore programming. Right now, there is no easy way to efficiently scale a piece of software for multiple cores. It's too early to say anyone has an edge.
Long-time editor covering embedded systems at EE Times and Embedded.com influenced a generation of industry readers. Anyone who has touched the pages or viewed the pixels of EE Times or Embedded.com over the years has read and benefited from Bernie’s insights on embedded systems and software.
While the role of DNA as a biological memory is well established exploring its potential as a data memory is relatively new. DNA data memory has not quite yet reached the stage where a blob of DNA can have some wires attached to it to write and read its data content, good progress has been made.
Carmakers are at a crossroads. They can either follow the footsteps of the mobile industry by allowing two platforms -- Apple or Android -- or they can turn to open-source platforms like Automotive Grade Linux.