[Ti] A real world comparison

[Chris Olson]

On Jun 13, 2005, at 3:26 AM, <illovox at comcast.net> wrote:

> Though Shawn's existential proclamations regarding the certainty of 
> knowledge and confidence and the futility of extrapolation are 
> philisophically sound, I am fairly content with the assertion that 
> such an approach is not particulary applicable to Intel's general 
> puruit of knowledge, economy and leverage.

Intel's processor roadmap reads quite plainly for at least the next two 
years.  And it's quite easy to extrapolate even after that because 
Intel follows AMD's technology, and has for the last 6 years short of a 
few flings with disaster such as IA64.

> Maybe Steve figures not many really use laptops for vector 
> applications?

Not necessarily vectorized apps, but 64-bit.  You can buy a Wintel 
portable with a 64-bit processor (AMD64).  That will not be available 
from Intel.  Even on the desktop 64-bit computing is an add-on, not a 
design, in the x86 world.  Once you boot an x86 box with 64-bit 
pointers turned on you're locked in - you need a full 64-bit OS, device 
drivers, apps, etc..  Gone is the seamless transition to 64-bit like we 
had on PowerPC where we could support both 32-bit and 64-bit computing 
with one release of the OS.  And you could run a 32-bit core (which is 
faster in most instances anyway than pushing 64-bit address space 
around for apps that don't need it) and only turn on 64-bit for apps 
that need to access huge amounts of memory.

My prediction - IBM doesn't get their panties in a bundle over stuff 
like this - they get even.  In 2007, just as Apple's transition is 
about complete, they'll release a new desktop PowerPC processor with 
virtual cores (aka the Power6) that will flat blow everything in the 
ancient world of x86 right out of the water.  And they'll market it 
through Lenovo with a 64-bit linux operating system.

It's been said there's 10 types of people in the world.  Those that 
understand binary, and those that don't.
--
Chris