yeah it sounds crazy but we are still not even close to graphics looking like reality. as example when we see great lighting in games we think it's awesome and i waste a lot of time just looking at some nice spots but it's not really like in reality. i wonder if we would even look at that in games when it would be like reality?

i once sat in my car and had to wait for someone and because i was bored i looked at reflections on driving cars, signs, windows and whatever and it's crazy how much recognize then. but i never really recognized that before because i never looked for that. when a game really looks like reality, not sure if i would even look at lighting?

it's obviously different when you take photos, then you look at reflections/shadows and so on...

Could that power even be possible with chips made out of silicon? I don't think they can shrink the process nearly enough. It would need another material, which could mean who knows how many additional years of development before the breakthrough is done.

What's that other material they are trying to experiment on right now and which would allow transistors as thin as a couple of atoms? Graphite?

Nintendo already reached it with "project reality": 

The PS3's CELL can already do this, developers just haven't unlocked it's full potential ;)

I don't think when Game Devs try to reach reality quality graphics its about realism. I think its more of a quality level. Where just everything is perfect. Lighting no aliasing no blurriness no flaws and infinite detail as far our eye can see it. Mathematical models good enough to fool our eye in any case. What they then do with that power is up to devs.

Also AI research has bigger hurdles to overcome not just computing performance. Even 5 Petaflop machines will have unimpressive AI if people have no clue how to program amazing AI.

I would love to see games being programmed with 5 Petaflops in mind. 

But its a really long way still. 40 Tflops we get with 10nm approximatly. Maybe we can get to 100 tflop single GPUs using conventional means.

After that a long while nothing and than maybe graphene. It will be hard (impossible) to get to 5 Petaflop without major breakthroughs and an Industry wide revolution inbetween.

Before graphene we might get stacked chips, they already do it with DRAM/NAND which is usually a step ahead of the more complex CPU fabbing. :)

Besides, I don't think we *need* 5,000 TFLOPS to get near realism, the idea is to try to make things look as real as efficiently as possible and to use a variety of cheap tricks to fool the eye.
I mean think about it, things like culling of objects that you, the player would never see, improved compression algorithms, impostering etc'.

Yet, I feel strangely attracted to that 5,000 TFLOP number...

So basically, only EA, Activision, Sony, and MS can fund these in the future :P

Its about 1100 Titans.

Folding@home had 5 Petaflops in the year 2009 with tens of thousands PS3s and GPU/CPUs.

i couldn't care less for "reality graphics" in a game.

yeah, cause Sony is totally in a good financial state....