Jesus. I'm sorry to say this, but talk about being delusional...!
Where do you propose to find games programmers that know how, or are even willing to program in LISP in this day and age? How would development be rapid and smooth as you claim when devs would have to throw away every single piece of code in their source library as well as all middleware, because none of it runs on Nintendo's wacky FPGA-powered LISP machine?
This sounds like complete and utter madness.
Lisp is one of the of the greatest programming languages ever thought up. There is no doubt about that.
It is hugely influential but none of the (partial) imitators have been able to match the power and productivity of real Lisp or one of the close variants.
You clearly need to read up on this.
And Lisp in games and 3D graphics? Naughty Dog does part of their game engine in Lisp (Scheme) precisely for power and productivity reasons. Some of the early 3D in movies and TV (whilst the two Lisp machine companies were alive) where done in Lisp.
The drawback of Lisp is that it runs rather slow compared to say, C on the current hyper souped-up but essentially neanderthal hardware of today (no offense to any neanderthals reading this
.
A real language specific machine made with todays fab process and bandwidth would blow any of todays CPU away on Lisp. And not by a little.
But the great thing about an FPGA language architecture is that it is almost completely flexible. You could take just about any great language and make it run many times faster when interpreting.
And late binding in general (done right of course) makes for much faster development, smaller code and easier maintainability and malleability, among other things.
What devs like is for hardware to be predictable and straight-forward, with good development software system. First having to design the hardware and program it into a set of FPGAs before being able to write a game for it doesn't sound like something that fosters rapid development. Who amongst today's top leading software developers would even be able to design a CPU from the ground up, or shit, a GPU for that matter. These things take years and hundreds of the smartest, brightest people you can find on this earth. If you want no games whatsoever to come out for Nintendo's next machine, this would clearly be the way to go...!
Predictable and straight-forward will only get you predictable and straight-forward results. IE incremental change.
What I'm talking about is a game changer.
Of course Nintendo would have to do all the hard work of defining the architecture(s) and making a whole sleek and ultra developer friendly package available from day one. If they had started in 2006 guns blazing, they could have been finished in 2011 or 12.
The difficulty of designing a CPU/GPU is overrated. Complexity-wise CPUs are much simpler than even moderately large programs. The ARM CPU was done by a small team, and the design team at ARM is still comparatively very small.
Much of the complexity in for example Intel CPUs is artificial/for other reasons than pure power and utility.
A clean slate design wouldn't be too complex for a company such as Nintendo to handle.
Also, FPGAs tend to not be cheap, and an array of them sounds like costs could easily gallop with no performance gains to show for it. FPGAs have terrible efficiency compared to an ASIC, they're much slower and use multiples more transistors for the amount of logic gates they hold.
FPGA's can be much better tailored and optimized to a specific purpose all while having very deep pipelines and dissipating very little power.
That's worth more than lot's of potential, but ultimately non-usable power from many more transistors on an ASIC.
Given an advance order of several million (or licensing) I think Xilinx would be able to respond with a very favorable price.
If you just want some low-powered shit that is good at playing adobe flash type games, there's already Ouya and whatnot. Why go to this extreme crazy length with FPGA arrays and wacky obsolete programming languages from 30, 40 years ago?
FPGAs are used in some supercomputers, exactly because of their great value for money. That is even with FPGAs that are not yet optimized for general purpose computing from the manufacturer.
C and C++ and all their variants are older than 30 years and are much more "obsolete" than, say Lisp for example.
Again you clearly need to read up on this.
The only thing moving in the computing field is Moores law. Just about everything else has essentially been standing still or regressed in various ways for the last 30 to 40 years.
Complexity has gone up, sure. But complexity is most of the time a sign of regression caused by an overhvelming task, lack of clearness and coherence. Not of greatness.
High game costs aren't going to be solved by changing programming languages. Not having to worry about the kinds of careful micro-optimizations that PS3 and XBox 360 needed should help a little, but I expect the bulk of game development costs will still be in generating the assets, especially graphics.
Asset generation is deeply intertwined with the how the engine works, if done right.
The solution to the asset problem is through programming, better modeling tools and methods and a less is more approach to the whole gameworld. Your game isn't going to be any better by taking place on a whole meticulously grafted tropical island with zero value to the gameplay and fun of the game.
Programming language specific processors never really took off not just because they were too domain specific but because in the long term they weren't actually faster than good implementations on general purpose CPUs. A good compiler or even JIT takes down a lot of the benefit of hardware acceleration of most languages, then you have to consider that general purpose CPUs have had many years of incremental improvement behind them, supported by huge development budgets. This comparison is for a real custom CPU too, not one implemented on an FPGA which would be multiple times less efficient in every metric (performance, power consumption, and cost).
That is pure BS. You are essentially arguing that specialized hardware doesn't pay off, which has been proven wrong too many times to be worth mentioning.
What went wrong with the Lisp machines was a mixture of arrogance, geek cat fighting and not having access to the newest fabbing like Intel and Sun.
But look at how coveted and loved the Lisp machines where and still are by the lucky few.
HLL language specific machines is an idea whose time has come (again).
Doing the GPU on an FPGA is even less practical..
There would be less to gain surely, but you'd be able to reclaim and reconfigure power much more easily than with a GPU. So overall it would be worth it.
A VR headset could be cool but I think launching a console around it right now would be suicidal. The industry needs way more time to figure out how to properly develop and utilize this, and the additional costs would be huge.
The tech to do it (good enough (tm) and cheaply) has been with us for some time (motion tracking and large cheap high res LCDs). It's the industry's complacency and mutual FUDD that has held it back. It came to the point where even a kid in a garage could make it, before things started rolling.
It would probably have been a bad idea to ship the console with it, only for cost reasons. But it could have been a relatively cheap and strongly endorsed, 1st party supported extra.
Think of how close the WiiU controller already is. The LCD would have to have double the resolution, and you would need some optics in front of the screen but otherwise it's not far.
And you wouldn't have to pay for miscellaneous, like the antiquated resistive touchscreen, the buttons, sticks, IR bar and camera.
