Thursday, October 23, 2008

I pause from the "usual" processor technology talk, to address another topic that's been a favorite interest of mine since grade school: space travel. Buzz Aldrin thinks that we should send the first explorers to Mars on a one-way trip to the red planet:

I can't completely agree with him. Much like the early plot of Kim Stanley Robinson's Red Mars, the first explorers should return to Earth in order to be able to relate the experience to the world. Otherwise, I think the exploits of the Mars explorers will be too remote, too unreal, to the people who must fund supply missions. Only when we have the technology to make the "settlers" self sufficient, should we consider the one-way trip option.

What we should be investing in is better robotic AI ("Open the pod bay doors, Hal") and the underlying technology needed for colonization. Both these endeavors could have additional benefits here on Earth.

I hope I live long enough to see the pictures of Man on Mars. Just going there and not colonizing is a waste of money. Look at what happened when we went to the Moon - close to 40 years later and we haven't been back.

Sunday, October 12, 2008

Sorry for the lack of posts - worklife and funlife has been busy. Frankly, not really much new to report on the processor front. The Larrabee watch continues - still no sign of real silicon.

The AMD "asset smart" split was announced and it didn't help the stock much. A lot has been written about it, although the present economic crisis has overshadowed it. I see it as mixed news. The infusion of more capital should help AMD compete with Intel - more money for the fabs will build the capacity AMD can't afford to fund itself. But you still have a foundry completely dependent on AMD for its revenue and AMD is almost completely dependent on the foundry company for its silicon. It will take at least 2 years for the foundry company to build up a reasonable portfolio of processes that will attract other customers (it needs bulk silicon). Most importantly, the foundry company will have to learn a new mindset as a foundry, not just a captive fab. IBM made that transition, but it is still not a flexible and cost effective as TSMC.