I'd like to throw one more $0.02 opinion into the pot.

First, I absolutely wish this project luck. There's plenty of room for projects like this in the world. If I were creating a game machine (actually, I am but it's project #459 out of 1000 projects), I would create it the way I wanted and hoped others would find it useful or entertaining. Even it it were only one other person. That's why they're hobby projects.

Now, if I were to create something for market share (no matter how small) and my intentions were to put them in the hands of hundreds or thousands, I would certainly do my research and see what succeeded and what failed.

Some of you may not know, but there is a very smart man named Andre' LaMothe. He's known for several things (especially books) but over 10 years ago he started creating some hardware game consoles for the purpose of teaching hardware to people. You could learn how to design your own console and learn how to program it. He had several variations. One of the first was the "XGameStation" based on a SX52 micro-controller running at 80MHz. I wanted it so bad that I bought it...paid over $300 for it. But it came with an awesome manual, example code, etc. It was a lot of fun.

Then I got bored and wished I hadn't paid so much for it (I still have it).

I always wondered how much money he made off those consoles. I know he worked INSANE hours on it and others. Not to mention huge amounts of money he had to put upfront (his words).

Point is...there's a world of difference between hobby and professional. I would probably start with hobby, keep it dirt simple....and drum up interest.

_________________
Cat; the other white meat.

Agreed on that. (If you want to keep a low price, you need to be very focussed on what parts you use and how big your board is - and you need the confidence and the market to buy and build in volume. Even for a modest-priced gadget, buying parts in quantities of 100s is quite a lot of money, and still not a big enough volume to get great prices. I see the Gigatron is now shipping kits in quantities of dozens, at a price of €160 or so. I don't know what their purchasing volume is.)

True.

Now, in Andre's defense, the XGameStation is a large board. And, this was nearly 15 years ago so I don't think the super-cheap Chinese fabs were as plentiful then. Plus, the book you got with it was top-notch on hardware design. Including how to create NTSC colorburst signals, etc. I believe it was worth the money...I just wished I didn't buy it since I rarely used it.

_________________
Cat; the other white meat.

I started a new thread in the Forth section --- this is getting way OT from game-machines.

I don't have any ideal game --- I'm expecting the users to be more creative than I am and think up something --- I'm hoping that users develop games with more strategic depth than the mainstream games which seem rather repetitive and non-thinking (those first-person shooter games are not only bad because they promote violence, but they would also bore me to death after about 2 minutes).

I can provide a development system and a code-library with pseudo-sprites that would be sufficient for writing a 1980s game such as Centipede or Ms Pacman --- possibly sufficient for a 1990s game such as Mario and Yoshi from the SNES.

I also don't have the time or inclination to write games myself. I can see myself writing something like Centipede which was one of my favorites from the old days.
Mario would require way more time than I have (actually, it would require a team effort, as nobody has that much time).
Mario would also require way more memory than the PIC24 has --- I don't know how much memory the SNES had, but it was likely upwards of a megabyte.

This whole game-machine idea might be a bad idea. It seems interesting to me to interface with the hardware directly --- may not be interesting to anybody else --- people could just use a desktop-computer, or at least a Raspberry Pi or some such thing.

This thread could really fizzle out as far as I'm concerned, as I've already said that a 16-bit processor such as the PIC24 would be needed for a game-machine. Most people would recommend the ARM Cortex because it can address megabytes.
I don't think the 65c816 is powerful enough to be interesting in the 21st century --- according to that video discussing disassembly of the Mario game on the SNES, they were using self-modifying code to eke out as much performance as possible for the game, so the 65c816 was barely adequate in the 1990s --- discussing a game-machine on the 6502 forum isn't interesting, except possibly for a retro-rebuild such as a Commodore-64 in an FPGA, which was discussed elsewhere on this forum.

Well then, what part of that is most interesting? Focus in on what makes it interesting to you, share what that focus is, and build the system around that.

In my opinion, if you just have a framebuffer there's not much interfacing with the hardware that needs to be done, compared to more featureful sprite/tile/layer systems. It would basically be a software defined graphics system.

_________________
WFDis Interactive 6502 Disassembler
AcheronVM: A Reconfigurable 16-bit Virtual CPU for the 6502 Microprocessor

Okay, we get it. You think the 65C816 is a piece of junk and the 65C02 is not far behind it.

A parting thought for you: the SNES's Ricoh 5A22 MPU (a 65C816 derivative) was clocked at a leisurely 3.58 MHz during non-bus cycles and was slowed down to 1.79 MHz while reading or writing the controller port serial-access registers. That's a far cry from the 20 MHz at which the 65C816 has been run in some applications. You think that might have had something to do with why self-modifying code got into the picture?

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Here's how I see it: pretty much any platform used for games ends up with people doing the most extraordinary gymnastics to get the most out of the platform. So the existence of those gymnastics isn't evidence of a platform being deficient in any way.

I think there's a big difference between a games machine designed for games and a games machine designed for education. In the head post I got a sense that the idea was education. That's what leads me to thinking of a simple machine, and possibly a simple virtual machine, because details of sprite engines or tiled graphics are one step removed from software controlling a screen. They are both examples of hardware acceleration, or augmentation, and both very handy ideas to have, but from an educational point of view they are details upon details and a distraction. Not that I'm an educator, and nor do I play one on TV.

When it comes to the games machine which isn't for education, a whole different set of tradeoffs and constraints become appropriate - some people will demand to see a through-hole CPU and others will be completely happy with an embedded emulation, and there are stopping-off points in between.

This, of course, is why it's handy that forums have threads, with titles - each conversation can have a topic, and the head post can set out what the idea is. After a broad discussion, it's quite OK to have a new thread which extracts one idea, summarises the context, and takes it forward.

Oops, something I hasten to clarify: when I said "at opposite extremes," I only meant within the context of various Forths.

Okay, good call -- a new thread is a good idea.


PS, BTW and FWIW ...

Self Modifying Code has both advantages and drawbacks, and one needs to weigh the matter objectively. I suspect SMC became stigmatized largely because it requires careful commenting in the source code, and that effort wasn't always forthcoming.

In most case you'll just modify operands -- not opcodes -- and thus the technique needn't be terribly complex. But SMC isn't reentrant and isn't ROM-able. And there can be cache issues, but those don't apply to 6500 family processors.

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

I don't see how you can make a blanket statement like that based on a few games in the 90's for one system. The SNES was never known to be super fast (anyone remember blast processing? lol). The '816 in the SNES ran at 3.58 MHz to keep in step with NTSC colorburst, etc. That's a far cry compared to what it *could* have ran. The '816 was also purposely slowed down for the Apple IIgs because Jobs didn't want it competing with the Mac.

The '816 has had an unfair (IMHO) rep in the computing world because it's only had two large scale products (that I know of). And both of those were either crippled on purpose or crippled due to budgets.

I think the '816 has more of a punch that most people think.

_________________
Cat; the other white meat.

I could be interested in a thread about products using the '816. I'd be sure to mention Acorn.

Nintendo's genius, to me, is that they are not a company focussed on technology, but a company focussed on products which entertain and divert. As the company has survived since 1889 (I had to look it up), their approach could be said, I think, to have been a good one.

Regarding hardware performance for games.

Historically, all hardware is "too slow" for the games folks want to make. They always want to do more and more and more, only to run in to limitations. Wether this was on the original Space War or the the recent AAA titles from the large game publishers running on paired GPUs.

Is there a middle ground of games that don't really tax their systems? Of course. But top shelf games, games that draw attention and acclaim. Games that make you go "wow", have always been trapped within their hardware. The designers always wanted to do more.

A 640x480 4 byte per pixel at 60 FPS is ~74MB/s. A single frame is ~1.2MB. For the record, my phone is 1125 x 2436. ~11MB frame, and over 650MB/s at 60FPS.

This is an atypical example because there are many parts in there, and it is a very niche thing with a narrow audience it appeals to. The enthusiasts are typically not the millennials who pride themselves to be "full stack developers". Still the batches are large enough that you could also buy a car for it. Think of it as an Apple-1 order of magnitude operation, except that it is world-wide instead of regional. The BOM is long. Some parts have >10 weeks lead time, so you must buy those in larger quantities than you're comfortable with if you don't want a waiting list. Some combination of faith and ignorance comes in handy. It only works out if you ignore the risk of ending up with thousands of otherwise useless chips. Luckily, the batches are small enough to start worrying about parts for a second series after one 8-bit guy episode on youtube.

The real costs are not parts but labor: design, logistics and support. The kit version only works because those hours are not reflected proportionally in the selling price. (We hear similar stories from other hobby kit makers.)

IMHO, for getting hands-on low-level hacking with educational value, the Arduino is truly brilliant. Not just the board, but more so the ecosystem. It doesn't do video games well, but the kind of video games it could do is not something teens can appreciate anyways. Even "Far Cry" is vintage for them, so why bother them with the nuances of different types of sprite or tiling systems. Controlling robots, motors, sensors is timeless however, and the platform naturally leads you to towards low-level hacking when doing that. At 34C3 I witnessed somebody teach rolling your own dispatch loop to a child... When used for gaming, an LCD or LED display will then do. There is no "this-is-not-as-good-as-Crysis" push back when the game is not on a monitor screen.

It is difficult to explain what technical stuff is interesting to me, and I doubt that anybody cares what is interesting to me.

I can explain, however, what is NOT interesting to me:
I am not interested in learning how other people's software works.

I'm very much in the minority on this. Most "programmers" are actually script-kiddies --- their primary skill is to quickly "get up to speed" on some software --- I have always been quite slow at learning any software system, and in fact would prefer to dodge the issue completely by writing my own software.

So, if I develop a game-machine, my goal would not be to provide a full-blown game-development system and create my own army of script-kiddies who brown-nose me, telling me that I'm a genius and a great leader, and in exchange I grant them expert status --- I would prefer to surround myself with real programmers who think for themselves --- of course, this is not a business plan that is destined to make much money (there are not many people on this planet who think for themselves, and most of them are bogged down in arguing with each other and/or defending themselves from trolls who call them stupid on the internet because they aren't following the herd).

Well, I'd say the same things as before. In order for discussion to be valuable, we need to understanding what you're pushing for, not what you're pushing against. That core vision would drive the design by illuminating well-defined boundaries and priorities. The discussion gets very non-cohesive otherwise.

_________________
WFDis Interactive 6502 Disassembler
AcheronVM: A Reconfigurable 16-bit Virtual CPU for the 6502 Microprocessor