Is this first-degree wank?

[Dooey Jo]

I have someone claiming to be an electronics professor here, who says that he knows what the next generation of gaming consoles will be like. Personally, I find some of these things quite ridiculuos, but I'm a software person not a hardware person so I'll let someone else judge this also.


The 8th generation consoles of 2011 will all support 1080i, most likely on two channels.
Starts off reasonable enough. Two channels seem a bit unnecessary though.

They will have about 100 cores of about 20 GHz.
Uhm, a hundred fucking cores? Each of 20 GHz? Ignoring the costs for such a monster, I expect it to be hell to develop for.

The graphics will look at least as good as the effects in the movie Narnia.
Right. Not only is this system going to match ILM's and/or WETA's render farm, it's going to surpass it by doing it all in real time. Supposedly at the price of today's next-gen consoles...

Controllers will be wireless and look like Wii controllers. Advancements in battery technology will allow both motion sensitivity and multi-dimensional advanced feedback.
Advanced feedback? In a portable controller? What're they going to do? Put in really large or very fast gyros in the thing? Maybe this battery breakthrough also involves breakthroughs in reactionless propulsion physics...

The consoles will have 8-16 GB of memory.
Seems a bit excessive, not to mention fucking expensive. Even in five years.

The discs will store 250-1000 GB.
They're blu-ray discs, too. Is that even theoretically possible? Let alone economically feasible? Of course, there's no way game developers will be able to afford to fill up a 1 TB of high-resolution content in five years... And if they would, no-one would afford to buy the game...

The consoles will have sensors to detect where the player is so they can adjust the picture if you throw yourself to the side
I think seeing the TV would be a bigger problem if you throw yourself to the side. Besides you can already do this with the Wii-Wiimote.

The picture will be of hybrid 3D type. The monsters will appear to walk right out of the TV and into the living room.
That's what they told us in the sixties too. Seems like fake-3D really isn't that hot.

"I know this because of something called extrapolation. It's a very exact method to predict the future and my numbers and the facts will probably coincide very closely."
Sounds like a bullshit method to me.
However I may be biased, because I know that this is the guy who said that the Nintendo Wii would have some mysterious physics processing chip with more processing power alone than the PS3's Cell, as well as a similarly powerful texture chip. Overall, he said, that it would probably be more powerful than the 360, but less so than the PS3 (but not by much). Now I like the Wii (heh) and don't care that it's not that amazing, graphics-wise, but I think you can see why I think this guy is a total wanker...

Oh yeah, the 9th generation of consoles is impossible to predict because they will use quantum computing. That's ten years from now by his timescale. Am I out of my mind here, or should I contact his university to have this guy evaluated (if he really is a professor. For all I know he can be bullshitting about that too. He can't seem to spell either...)?

[Spanky The Dolphin]

I don't think he's so much a wanker as being completely from outer space...

[RedImperator]

100 cores? 1TB games? Photorealistic CGI graphics? What is this kid like 12 or something?

[General Zod]

If he's a 'professor' then you should be able to ask him to provide the following information: Where did you get your degree from? What school do you presently teach at? How long have you been in this field? What papers have you published in terms of research, if any? If he can't answer at least one of these, then I'd just as soon write him off as a complete whackaloon.

[Mr Bean]

To note there DOES exist storage media that can hold 500GB's of data on it in CD form. Last I checked it was still not outside the Lab yet.

Likewise 8-16GB of memory is @#$@$# silly. Memeory does not follow Moores Law. And he never @#$@$ explains how they are going to COOL 100x cores in one box... unless it's a ten foot tall console.

[RThurmont]

This might be what video game consoles are like in 2021, but in 2011 it strikes me as a tad absurd. Correct me if I'm wrong, but haven't consoles historically lagged well behind PCs in terms of performance and features? I've never been much of a console fan-I want every system I own to be able to run any application that it's possible to program.

[Spacebeard]

I don't know or care anything about game consoles, but it's worth pointing out that a machine with 100 cores each running at 20 GHz, assuming they were possible to build, would require some ~ 762 GB/s of instructions alone. The system's 16 GB of main memory might have a wee problem supplying that, so this hypothetical machine would be incurring millions of page faults a second just for instructions, without even considering data. Whoever came up with this obviously doesn't understand anything about computer architecture.

Likewise 8-16GB of memory is @#$@$# silly.

Well, high-end UNIX workstations today can have a 16 GB main memory, so I wouldn't rule out the possibility that a PC of 2011 could have that much RAM. I don't know about game consoles, though.

[Admiral Valdemar]

Samsung have made the first NAND based flash hard drive that has around 200 gigs or so. Given your hard drive would be pretty much like RAM with none of the million times slower access speed of a magnetic disk, it's very possible.

[lPeregrine]

He forgot one point on that list:

Game development time and cost will increase exponentially, as will game prices.

Photo-realistic effects like that are not cheap, fast, or easy even when they don't need to be interactive and real-time. Take a look at the average effects-heavy movie's production time and budget if you disagree.

Games are eventually going to have the ability to reach this kind of quality. It's only a matter of time once the rendering software can do something in still images/pre-rendered video. It's nothing more than adding more brute-force hardware power to get the calculations done faster. Whether we'll see that kind of power within 5 years is another issue entirely...

So it's really hard to predict what the future holds on that issue. We might get expensive amazing photo-real games, and see the market shift to rare major releases. Or we might just see game developers avoid the hardware's full potential, and work on improving other aspects of the game that haven't hit the point of diminishing returns.

[Arrow]

Because I feel like having some fun, here's my semi-informed/semi-out-of-my-ass assesment on this guy's statements.

The 8th generation consoles of 2011 will all support 1080i, most likely on two channels.

This is reasonable. SLI/CF can do this today.

They will have about 100 cores of about 20 GHz.

Complete and utter bullshit. I can see upwards of 16 cores by then, but even that's pushing it. And 20 GHz? Not happening. This dipshit hasn't noticed that processor speeds have not increased over the past few years? If we see 8 GHz in a server CPU by 2011, I'll be impressed.

The graphics will look at least as good as the effects in the movie Narnia.

Haven't seen Narnia, so I'm guessing he's talking about photorealistic quality, where you can't tell the digital additions from the true film. This one is iffy. Crysis is looking damn good, and if the real thing holds up, and the rest of the DX10 titles look that good, he may be close on this.

Controllers will be wireless and look like Wii controllers. Advancements in battery technology will allow both motion sensitivity and multi-dimensional advanced feedback.

Define multi-dimensional. I'm sure you can get something like this (say, two or three different kinds of feedback), but the battery life is going to suck hard.

The consoles will have 8-16 GB of memory.

8 GB is pushing it. 16GB is insanity. I don't think the desktop market is going to see 4 GB be realistic until a good bit after Visita's release and 64-bit computing is standard, and that's going to take a while yet. 2GB is probably much more reasonable, and 4GB is possible.

The discs will store 250-1000 GB.

I can believe this, but I doubt they'll be Blu-ray.

The consoles will have sensors to detect where the player is so they can adjust the picture if you throw yourself to the side

Hmmm, I don't think so. Maybe you could do it with a laser...

The picture will be of hybrid 3D type. The monsters will appear to walk right out of the TV and into the living room.

Bullshit. I'm not aware of any commerical technology, even in development, that can do this.

"I know this because of something called extrapolation. It's a very exact method to predict the future and my numbers and the facts will probably coincide very closely."

He's pulling it out of his ass, tossing in a few things that are close to reasonable to make it sound like he knows what he's talking about.

Oh yeah, the 9th generation of consoles is impossible to predict because they will use quantum computing. That's ten years from now by his timescale. Am I out of my mind here, or should I contact his university to have this guy evaluated (if he really is a professor. For all I know he can be bullshitting about that too. He can't seem to spell either...)?

They're still trying to figure out the best ways to make simple quantum logic gates! There's no way in hell Quantum computers will be ready for prime time in ten years. Maybe fifty, maybe.

As for this guy being a professor, I've met some strange ones when I was in college. A number of them had piss poor spelling. Some had very strange, irrational viewpoints. One even said Java was just as fast as C/C++ . But I'd check anyway, just for fun.

[rhoenix]

I won't reiterate what others here have said, as they've debunked this "professor's" claims thoroughly enough. To the OP, I'd advise not shaking his hand, and telling the gentleman who spouted these Amazing Glimpses into the Future to invest in Kleenex.

[Arrow]

They're blu-ray discs, too. Is that even theoretically possible? Let alone economically feasible? Of course, there's no way game developers will be able to afford to fill up a 1 TB of high-resolution content in five years... And if they would, no-one would afford to buy the game...

If a large dev puts raw, uncompressed source art on the disks, they could probably get close to 1TB. But I pity the GPU that has to render that.

[Drooling Iguana]

And all the games for it will be retreads and sequels to things we've been playing for the past decade!

[Pu-239]

As for this guy being a professor, I've met some strange ones when I was in college. A number of them had piss poor spelling. Some had very strange, irrational viewpoints. One even said Java was just as fast as C/C++ What the fuck? . But I'd check anyway, just for fun.

I thought the newer versions of Java really are as fast as C/C++, the problem is poor speed on startup (as the JIT compiler hasn't gone to work yet, which really does make it slow for desktop apps which get closed and opened, less so for server) and memory usage?

[Arrow]

Last I read, Java was about 1/3 the speed of native code. And unless I'm grossly mistaken and out of the loop (and granted, I don't follow the Java scene, since all my work is C/C++), it never will be as fast as C/C++, since it has to run through a VM and get translated into native code.

[Spacebeard]

Last I read, Java was about 1/3 the speed of native code. And unless I'm grossly mistaken and out of the loop (and granted, I don't follow the Java scene, since all my work is C/C++), it never will be as fast as C/C++, since it has to run through a VM and get translated into native code.

The JIT compiler translates bytecode into host machine code in advance; it's not going to perform like an interpreted language that translates instructions one-by-one as they are executed. Like Pu-239 said, desktop applications will "feel" slow because of the overhead imposed by initialling loading the JVM, but for long-running server or scientific applications the slow-down is minimal.

Sure, it's still fair to say that Java is slower at run-time than well-written C, but it's also not "strange and irrational" to say that Java can be, for all intents and purposes, just as fast as C for some applications.

[Arrow]

Since I've always dealt with Java in a desktop/UI environment, that's why I thought it was strange. Do you know the average amount of overhead the JIT incurs in a server/scientific application?

[Spacebeard]

Since I've always dealt with Java in a desktop/UI environment, that's why I thought it was strange. Do you know the average amount of overhead the JIT incurs in a server/scientific application?

Not offhand, but Google turns up this page with links to benchmarks of several long-running computations, some showing an average 20% slowdown for Java compared to C, and some showing Java faster than some C compilers.

[Netko]

Agreeing with the debunking in progress, I'm going to only nitpick on one thing:

The discs will store 250-1000 GB.

Its teoreticly possible that Bluray will hit the lower limit in that quote in 5 years. It has been designed so that multiple layers can later be added. The problem is manufacturing techonlogy and price. In the lab, TDK already made a 200GB (6 layer) prototype disk - however that is nowhere near mass production. And unless the discs are not much more expensive then their less layered brethren they won't be used. Which is unlikely since for mass production to take place they first have to find a stable process suitable for mass production and then modify/design new machinery. All of which means major $$$ spent which has to be recuped in sales.

By 2015 or 2020, sure, I'd even say likely, but by 2011? Maybe, but probably not - more likely there will be (if the discs even exist) rare limited editions with tons of extra crap that use those (6 layer) discs, while most stuff will use 2 or 4 layer discs.

Even in a most optimistic anylasis current tech like bluray will only hit near the lower end of the spectrum presented, which I doubt was the intention of that quote.

[Molyneux]

Fools!
I know precisely what the games in 2011 will be like; they will be Spore.
Do not resist the Sims...the time of Will Wright's SimEmpire is at hand!

[Dooey Jo]

The graphics will look at least as good as the effects in the movie Narnia.

Haven't seen Narnia, so I'm guessing he's talking about photorealistic quality, where you can't tell the digital additions from the true film. This one is iffy. Crysis is looking damn good, and if the real thing holds up, and the rest of the DX10 titles look that good, he may be close on this.

I'm actually very skeptical about that. I don't know if shader technology will ever be able to "cheat" enough to, in real time, simulate NURBS and true raytracing and radiosity effects to the level of quality seen in today's high-end CGI. Or even 1993's high-end CGI. Some of the CG dinosaurs in Jurassic Park doesn't look that good by today's standards, but they still look a lot better than any game I've ever seen (their biggest flaw is the animation, and games are even further behind in that department...).
And even if they would be able to render hair, fur, true motion and depth of field blur, muscle skin deformation, advanced cloth, and everything else, to the quality seen in Narnia (which I haven't seen either, BTW, but there are pictures), I can at most see it being done on a few models at best, certainly not on whole armies or anything like that.

Of course, there is another problem with completely photorealistic graphics. It is that people will most likely no longer accept that there can be ten people in the same village that looks exactly alike. So not only would the quality of artwork in games have to increase, but the amount and diversity would have to increase also, perhaps even more... All adding up to impossible production costs.

BTW, he now also said that production costs would be acceptable because the gaming market would be 2-3 times larger in five years. But there is simply no way that these changes are going to be only 2-3 times more costly in five years. FFS, a movie quality game for the price of a normal game production today? A large game of high quality costs single-digit millions of dollars to produce. A normal movie with high quality CGI? A hundred million, at least. I wonder what the movies will be like then

100 cores? 1TB games? Photorealistic CGI graphics? What is this kid like 12 or something?

27, actually...

[lPeregrine]

The graphics will look at least as good as the effects in the movie Narnia.

Haven't seen Narnia, so I'm guessing he's talking about photorealistic quality, where you can't tell the digital additions from the true film. This one is iffy. Crysis is looking damn good, and if the real thing holds up, and the rest of the DX10 titles look that good, he may be close on this.

I'm actually very skeptical about that. I don't know if shader technology will ever be able to "cheat" enough to, in real time, simulate NURBS and true raytracing and radiosity effects to the level of quality seen in today's high-end CGI. Or even 1993's high-end CGI. Some of the CG dinosaurs in Jurassic Park doesn't look that good by today's standards, but they still look a lot better than any game I've ever seen (their biggest flaw is the animation, and games are even further behind in that department...).
And even if they would be able to render hair, fur, true motion and depth of field blur, muscle skin deformation, advanced cloth, and everything else, to the quality seen in Narnia (which I haven't seen either, BTW, but there are pictures), I can at most see it being done on a few models at best, certainly not on whole armies or anything like that.

It's going to happen, given enough time. Well, at least the capability will be there eventually, whether game developers are willing to bring their budgets up to that level is another issue entirely.

None of those effects are anything amazingly complicated, they're just a matter of brute-force processing power. The hard part is developing the shaders/lighting/etc itself so you have working hair/radiosity/etc of any kind. Once that's done, all it'll take is hardware quality going up enough to render at 50 frames/second instead of .00001 frames/second.

If the industry puts some work into cheats and optimizing the methods for realtime games, we'll see that capability even faster. So "cheating" is still useful, but not necessary.

Of course, there is another problem with completely photorealistic graphics. It is that people will most likely no longer accept that there can be ten people in the same village that looks exactly alike. So not only would the quality of artwork in games have to increase, but the amount and diversity would have to increase also, perhaps even more... All adding up to impossible production costs.

This is what I said about massively rising production costs. Making photorealistic images/video is HARD, and incredibly time-demanding. A lot of the expense of those CG effects movies has nothing to do with the render farm costs, it's hiring the experts to do the huge amount of work making those shots.

[Dooey Jo]

It's going to happen, given enough time. Well, at least the capability will be there eventually, whether game developers are willing to bring their budgets up to that level is another issue entirely.

None of those effects are anything amazingly complicated, they're just a matter of brute-force processing power. The hard part is developing the shaders/lighting/etc itself so you have working hair/radiosity/etc of any kind. Once that's done, all it'll take is hardware quality going up enough to render at 50 frames/second instead of .00001 frames/second.

If the industry puts some work into cheats and optimizing the methods for realtime games, we'll see that capability even faster. So "cheating" is still useful, but not necessary.

Yes, what I was trying to get at was that we probably won't see graphics like those before the hardware is able to actually render like they do, which I don't think is going to happen in five years. I mean, a normal mapping shader is nice and all, but it's never going to be as good as the actual high-res model it was made from, nor will cube-mapped reflections be able to simulate all the effects of real raytracing. I still haven't seen anything close to decent usable real-time radiosity...

This is what I said about massively rising production costs. Making photorealistic images/video is HARD, and incredibly time-demanding. A lot of the expense of those CG effects movies has nothing to do with the render farm costs, it's hiring the experts to do the huge amount of work making those shots.

Absolutely. I think that if the game industry would completely move to highly detailed photorealistic art, probably around 80% of all the current artists would lose their jobs. Making a game model is quite a bit different from making Yoda...

[phongn]

"I know this because of something called extrapolation. It's a very exact method to predict the future and my numbers and the facts will probably coincide very closely."
Sounds like a bullshit method to me.

Extrapolation has its uses, but it is rather inaccurate.

Since I've always dealt with Java in a desktop/UI environment, that's why I thought it was strange. Do you know the average amount of overhead the JIT incurs in a server/scientific application?

A lot of enterprise software runs on Java and the performance hit is quite acceptable for most purposes (it isn't too bad, really, and the JIT compilers keep improving - even Microsoft is doing that with .NET) but hardcore scientific computing typically use C, C++ and especially FORTRAN. The frontend might be Java, though.

[Nephtys]

"I know this because of something called extrapolation. It's a very exact method to predict the future and my numbers and the facts will probably coincide very closely."

Because yes, by extrapolating from the 1960's, we should have a moonbase right now, and beginning construction of space cruisers to fight the Soviets with.