[Ti] Re: multi-core chips / more speed

[Carbotron]

this is not a flame, just real. please if you really want to flame, do 
it to JUST me, not the list.

I really have to disagree with the "all hype" AMD has really brought 
affordable power to the desktop with the hammer(a64) and before that 
the athlonxp. currently, if anyone should be accused of hype it's apple 
with the under achieving G5s. basically taking the performance per 
clock away so clock speeds can go up, the marketing guys must be happy. 
Intel has just stopped doing this, with their pentium M and they were 
really the hype-pushers. the biggest problem with the PC world is that 
they've somehow gotten stuck with the lousy X86 architecture. blame it 
on microsoft, intel, IBM, or whoever you want, it still is the weak 
link in pc land. amazing what can be done despite a lousy architecture! 
although apple uses the power architecture proudly and makes a big fuss 
of it, IBM has been doing incredible things with it, on all scales. 
anyway, in userland, where real world speed is necessary, just my 
personal experience:

i do most of my real work (maya, CAD, etc) on my pcs, which are 
actually just AthlonXps. stacked up against top of the line G5s vs a 
nearly two year old $700(then) slightly overclocked 2.5ghz single 
athlonxp machine, tested four months ago, the athlon is still 
consistently on par even with a quarter the ram(1gb vs 4) and only one 
processor vs 2. nothing else i use is available for OSX, but they are 
seriously processor intensive, and maya raytracing and shaders being 
quite processor intensive itself has proven to be a good barometer of 
real world computing power for me. a 1ghz g4 is ~3.5 times slower. my 
800 tibook is about 5 times slower, so i don't even bother. results are 
similar on scenes that uses 80mb or ram or 900(pretty hard to use this 
much ram in my scenes). my father built a current budget 64bit hammer 
system for the express purpose of FEA. his machine is a good 35% faster 
than my main box, and here's the killer, it's that much faster running 
linux, using all the windows apps(my maya) in a VMware winXP! that 
machine was only about 800 bucks last month, and i don't think he'll 
ever use enough openGL to max out the 6600GT. I dont' think i really 
need to compare a laptop with this kind of power vs any powerbook. and 
they're out there. even a decent pentium M laptop can outmuscle the 
fastest powerbook. but osx is so much nicer and easier to work with 
than anything readily available for the x86 architecture that the raw 
available computing difference isn't apparent until you really start to 
crunch it, and by that time, you're already in love with your mac. i 
know i am. they're just "nicer" machines, and osx seals the deal.
i use my mac all the time. i wish i could use osx on my pcs. all 
non-power work(like spreadsheets, business docs etc.) i do is done on 
the much more reliable and enjoyable tibook.
so if you can stand it when you want real computing power, a current 
dual hammer(proccessor or soon, core) setup can rock hard the g5s, and 
for much much less power(watts) and dollars. x8664 isn't behind the 
curve, their curve is in a different graph, in terms of power, ahead. 
makes sense really, the hammer is just an x86ified evolution of one of 
the best processors of all time, and still one of the fastest : the 
alpha. yes, so good, they end up making more every time they announce 
canceling production. DEC really made the best processor way back in 
88. nobody really wants to admit it, but DEC was always two steps ahead 
of everyone. it's a shame the last and best, alpha EV8 never saw 
production, and x86 goes on. yuk. but i guess it works. btw, i'm typing 
this from my mac with the big HKS on it, and my main power pc has two 
old "power computing" stickers on it, right by the mopar, vortech, 
perma-cool and EBC ones. so confusing this allegiance thing...

	-Jason Economou
	Carbotron Composites
	kradcarbon.com