Wednesday, April 2, 2008

thoughts on intel and its foray into consumer electronics

for most of the last 30 years or so, there's been a dichotomy between the world of computing and the world of consumer electronics (ce). computing has been dominated by the wintel dynasty (or monopoly depending upon your perspective) in a constant race for more performance. on the ce side there really isn't any dominant platform from a silicon or software perspective. there hasn't been a need for it because our toys weren't expected to do the same things as our tools (i.e. PCs).

that's been changing in recent years as we've "gone mobile" 24x7 with our PDAs, mp3 players, PNDs, games and who knows what else. as has been shown by the success of the iphone (some may want to argue with that characterization), we increasingly want our toys to perform like our tools. historically, the way we've accomplished that is to take software from the computing world and adapt it to run on the processors that have serviced the ce world. that may sound rather straight-forward but not all of these processors are created equally. in many cases, the resulting "solution" may well come with a penalty in performance and/or functionality.

now, intc believes that it has come up with an alternative for the ce world that will allow OEMs to take advantage of the treasure trove of x86-based software. intel's new atom processors are a new family of microprocessors targeted at the one market the company has never successfully penetrated. there was a very good reason for intel's lack of success in ce. intel's parts are generally far too expensive for most applications and, more often than not, overkill from a performance perspective. there aren't a lot of cell phones that need a multi-core processor with 400-plus million transistors on it. even if there were, the 90 second battery life would probably be a limiting factor...

what intel hopes it has done is created a new family of microprocessors, designed from the ground up that are simple, powerful, very stingy with electricity, fully x86 compliant and very, very small (think cheap). rather than simply attempt to pack more transistors into a given area using leading edge process technology (45nm), they took the same leading edge process and applied it to targeted price/performance targets within the ce markets.

the company introduced the atom line at its developer forum in shanghai and expects the parts to be available later this year. at this point i doubt that we'll see much of an impact from the atom line in 2008 but i would venture a guess that we'll see some products using them at the ces in january of 2009. this is going to be a very interesting test for intel because, while our demand for new tools has matured, our desire for new toys is almost insatiable.

No comments: