That is an interesting position. Why do you think this decade will be different from the last? As far as I know, despite the current crisis, Moore's law still can be applied.

 

Yeah, and taking into consideration what a piece of shit the FX series was at LAUNCH, it's impressive you're still dealing with it.

The basic growth of computers has drastically slowed, and there haven't been any recent ground shaking chips like there has been the last decade. My first computer was a 300mhz Pentium 2. It came with a 1gb hard drive, I forget how much RAM, and it was a total piece of junk. The computer I'm on now seems like it only has 12 times more mhz, but due to architecture changes, etc, it's worth a hundred of the P2 junker. Hell my videocard would destroy the P2.

Flash forward a decade, and the way folks thought about computers changed. They used to be for geeks and scientists, then with the invention of the internet, and then gaming, the folks who were making the chips, were forced to try to do better. We've now reached the point where your average grandma is using a computer many times more powerful then she'd ever need to look at pictures of your offspring, and check her ebay bids while sharing recipes for soup. If the PC holds up and it's more then enough, she won't upgrade, and the chip makers will suffer revenue hits. They won't keep spending billions on R&D just so we can get a better shader on our sniper in Call of Duty #whatever.

Check out how realistic many games are now, we're reaching a point where they won't be able to do much else, save for throwing even more things at you at once. And even now, unless you're using 16x FSAA, and 16xAnsio, set to high quality, your "average" high end rig only has a few things to even slightly challenge it, and hell even Crysis, for all it's harshness on a system, there's youtube videos of people running the benchmark utility and crushing it, with everything maxed at 80fps+ Throw in the fact that it's code is very, very wasteful and gobbles way more resources then it should actually need just in case it needs them, and you wind up with modern computers on the high end being barely slowed down by it if it the code was streamlined more.

The last few years, it's been all about graphical advancement and just basic bandwidth, with the quad cores mostly being PR spin, and the dual cores being more awesome because of the new ways they figured out to use the chip, and better RAM, instead of just the fact that they stuck an extra cpu on the chip.

If you read up on Nvidia for example, they talk about how they'll be focusing more on getting physics and various other things onboard their chips, since the average game doesn't stress out their video cards that much, or they wind up having a lot of room to spare graphically now they've gotten the chips so far ahead of the game.

Good quotes on what the expect in the future with Windows 7's and future OSes in regards to GPU computing from Nvidia's president

It's now how much more shit can we cram on a chip anymore, it's WHAT can we do with the chips we already have. GPU computer has the potential to speed up stuff our Core 2's will do 50-200 times, according to the President. What can we do with it? We don't know, we're researching it, and it'll be awhile before we figure out what the hell we can use that much power for. Eliminating loading times and installing games to the hard drive, etc.

20 years from now, sure, gaming will be probably be totally different. The last 10 years? It's been about increasing poly count to where humans look natural instead of like boxes with faces glued on, and making it so your realistic humans can move through a realistic environment. We've done that. Can it get even more realistic? Yes, absolutely. Will it require a dozen-hundreds of times more raw computer power, to get there? No. And R&D isn't going to be able to keep justifying spending the cash to make faster stuff each year, if what we got already isn't even being halfway challenged yet.