They ALL matter. You just can't say that architecture A with X transistors, is better or worse than architecture B with Y transistors.
Understand?
Architecture B can require Y transistors in order to be competitive with Architecture A with X transistors.
Please, graspeth this concept.
Joe I understandeth thee. That's why I said 'even if they are different' that is even if they require different amounts of transistors and different speeds to be competitive, if the difference in amount of processing elements and speed diminishes or increases, it is a logical inference that the difference might tend to diminish or grow...
As I said if you compare a processor from today, even with different arch., to one of say 5yrs ago, Id' say (assuming they're designed for the same field/function/purpose) the one that's most recent will tend to be above in perf, and the one from the past will tend to be below perf wise.
One reason is because the modern one has a larger trans. budget, and is faster. The less time you give between processors, that is the more the gap in proc. elements and speed diminishes, the more the perf line will blur, and the larger the gap gets, the further perf will differ.
Examples: Take 3 proc.s:
1. proc. from 1yr ago..
2.proc... from 20yrs ago...
3.modern processor...
Assuming they're designed for the same purpose, the modern one will TEND to be closer in perf to the one from 1yr ago, the smallest gap, even if they differ in arch. Again I said 'TEND'...
Are they going to pony up and fight on the same scale, or are they going to play it safe and hope Sony stumbles? Questions, questions...
Well based on their previous actions, I'd say it's very likely they'll fight back... now there are two ways to fight back
1.) Waste a ton of transistors to beef up raw specs, if they try to do a ' we're 2-3X better', although anyone with any real notion of what's going on will realize they've sacrificed a lot to achieve, theoretical mumbo jumbo... and in the real world...
2.) Delay and launch superior $$$$ h/w a year later.
As i said before, 6-10mths down the road, CELL/R5X or whatever console technology, will be overtaken by the latest PC hardware.
It's possible, but the question is "by HOW MUCH?" a nigh 45nm arch, will require process equal or below 30nm to be significantly surpassed.
Again and then we've got to wait for software dev., if exponential leap is achieved, the gap between the low end and the high end will be RIDICULOUSLY large, thus it'll be yrs until this new "SUPAH" gpus are widespread enough to be viable for full support.
What about the dev. costs? Modeling of char.s, objects, dozens of times more complex, new advanced physics for said objects, etc. these all require MORE people, and people cost $$$. Most pc dev.s don't have the sale numbers necessary to sustain these significant cost increases...
And what about the bottlenecks, the fact that the cpu is likely to remain focused on gen. processing tasks, and not gphx, the ever growing size(in terms of mem. usage.) of windows, etc?
Face it, dude. Nothing in the PC world is going to come near Cell in the next five years or so if it turns out to be ANYTHING like the patent we've been discussing.
Well, I wouldn't necessarily say that... but if it's significantly surpassed very late in the game(a few yrs.), the time for software that will support said hardware to the max... won't likely arrive until the next console cycle... That is ALL that needs to be done, it can be equalled(somehow) a yr later, surpassed in 2yrs, and significantly(2-3X) outdone in 3yrs, but software will likely arrive TOO late for said 2-3Xhardware... by then we'll be starting the next console cycle...
Now We don't know what dev. might occur in the future yrs, these dev. can actually change everything, any model change, any new radical tech, say cheap diamond, dev. in the nano world, or rad. advances in lithography, etc... can alter everything, but there is a difference between ideas/discoveries and realworld applications, and that's that the later often arrives yrs later...
PS This is all assuming STI didn't seriously F@ck off when designing their arch...
As for my 2-3xhardware comment, it's that it does not matter if the gphx are equal or slightly above, in order to justify the price difference($300-500 gpu from a few yr laters), it has to show gphx that SIGNIFICANTLY outdo the ps3 ones, now 2-3x perf improvement might not even be enough to do this...
