Laptop predictions for 2006

[onix]

I found this article really interesting:

http://www.notebookreview.com/default.asp?newsID=2719

[snife]

i vote for 'Buy when you need a new notebook'

[AlphaKilo470]

Well, I knew it was inevitable but if this guy is right, then that means in the upcoming year we'll see an end to the record long computer lives we've been seeing these past five years as technology will finally be moving at a rapid pace again.

However, unfortunatley, there's still going to be the inevitable setbacks as well. The dual core stragtegy is a pretty nice innovation and has much potential. On the other hand, once dual core has hit maturity, it will be likely to cause setback: companies will simply try to cram more cores onto a chip instead of actually improving the chip itself. The person with the most cores wins. It'll be just like the megahertz race in the 1990's when clock speed was more important than the chip itself. In the 1990's we had the megahertz rush, now we'll have the core rush.

Another standpoint, for the better and worse: technology really was moving quite slow in the early 2000's which is apparent from the fact that desktops from 5 and 6 years ago are still able to run a significant percentage of availible modern software, much more than than the age equivalent could have run back in 2000. A 2000 computer is much more usable an useful and modern today than a 1994 computer was back in 2000. With all the groundwork laid in 2005, we're sure to see some major big changes this year and next that will on the good side of things male the computer market a fast moving place once again but on the negative side will mean people who buy new computers now risk having to upgrade sooner after purchase than people who bought new computers a few years ago.

[K. Eng]

I think yes and no.

Some computational tasks simply do not lend themselves well to parallel threads. I'm almost 100% certain Intel realizes this. The upcoming Merom core has a 4-issue front end and much more execution hardware than P6/PM/Yonah. Don't count on single thread performance suffering just yet!

Multi-Core does work really well though for some types of computation where there are few dependancies. Look for chips with high speed general purpose cores as well as high speed specialized cores. Personally, I think the IBM/Sony Cell chip is the way things may go. Right now its not much of a threat because the individual cells themselves are of limited use and the general purpose core is really weak.

Pair a dual K8 or Merom with a bunch of improved cells and that'll get things rolling. Of course the die size would be horrendous...

companies will simply try to cram more cores onto a chip instead of actually improving the chip itself. The person with the most cores wins. It'll be just like the megahertz race in the 1990's when clock speed was more important than the chip itself. In the 1990's we had the megahertz rush, now we'll have the core rush.

Edit - I voted for "buy later." The warranty hasn't quite run on my current machine, and I want to wait a month or two to see how the new Core Duo machines turn out