MiSTer FPGA Forum

Watching the powerful systems that the DE-10 Nano is able to emulate, I keep wondering... What if it was not bound to emulating a pre-existing system, and instead a brand new design was started from scratch? How powerful can it get, with the CPU and the video? Would the CPU be RISC or CISC type? Would it be better to beef up the CPU, and make it as fast as possible, or to make it simple and support as many cores as possible?

And what would be best for video? The NeoGeo has probably the most advanced graphics hardware currently supported. If we're not bound by an existing hardware, what would be possible? Like, 3D capabilities? Shaders? Could such a system overpower, say, the Nintendo Switch?

As an example, this guy has explored these fields a little bit, but not much: http://www.chrisfenton.com/the-zedripper-part-1/ - He created a new, not previously existing computer of his own design, but he used a pre-existing Z80 core as the basis. He also used a more powerful FPGA than the one in the Cyclone V.

I don't see what the point of new system base on fpga. FPGA are great for preservation but any newer system, even open source, can do much more with standard arm / x86 and some graphic core.

Sorry to disagree, but FPGA's main area of application is prototyping and highly flexible "circuits". So new "systems" is the main purpose and not preservation of old systems.

Ya but all the point of reproduction of old system is the fact that you have already a full catalog of games. If you want to develop something new you already have everything you need in ARM / x86 world, a solid base, api, graphic engine etc. Here you need to develop something new in the hardware and software side, and the tools to make it work. Maybe it could be interesting for a team that want to develop a new retro game and they need specific thing that require a chimera of old system. But i am no dev or hardware engineer so...

Okay agreed, but in what way does your answer has something to do with the original question?

Take for example this here:
https://github.com/sajattack/Hack_MiSTer

This is an fpga implementation of the "Nand to Tetris" computer, which is a new computer. https://www.nand2tetris.org/

And the question was what kind of new systems a DE10-Nano could probably implement.

I think you are a little to much fixed on the "Emulation of old systems" view. This is simply not was the DE10-Nano was originally build for.

It looks like the DE-10-Nano can emulate the new Propeller 2 processor. (https://forums.parallax.com/discussion/ ... evelopment) Not sure how many cogs and smart pins you can have. I have done some projects with the Propeller 1 years ago but the new Propeller is miles ahead of that. I don't think they released the code for a DE-10-Nano build yet. Soon you will be able to buy production chips so there won't be any need for emulation.

Certainly the DE10-Nano is capable of many things - that is the reason for its existence.
But keep in mind that the MiSTer platform is targeted primarily at gaming and retrocomputing - especially preservation of arcade boards, consoles and computers as closely as possible (within the limits of the hardware).

The fundamental problem of any new platform is that you will have no users and no software. It's all well and good but who would bother implementing something for it? Contrast to retro computer and consoles, which come with a huge library of games and a group of enthusiasts of original hardware.

But you can build something self contained, like this "demoscene"-style production based on a custom system made in an FPGA:
https://www.youtube.com/watch?v=h42neZVvoMY

Newsdee wrote: ↑Sun Jul 05, 2020 12:26 pm But you can build something self contained, like this "demoscene"-style production based on a custom system made in an FPGA:
https://www.youtube.com/watch?v=h42neZVvoMY

Thanks, that's pretty awesome! It does indeed explore the limits of what can be accomplished. "CPU: 16-BIT, 63 MHZ", "10 DSP-LIKE SHADER CORES". It doesn't seem capable to come close to a Switch, which was my initial conjecture, but it does seem to be able to go beyond what an NDS can do.

pgimeno wrote: ↑Sun Jul 05, 2020 1:30 pmIt doesn't seem capable to come close to a Switch, which was my initial conjecture, but it does seem to be able to go beyond what an NDS can do.

The Switch and other modern consoles have a top-of-the-line GPU with programmable shaders, gigs of RAM and gigahertz frequencies... you just can't replicate that with a single FPGA chip.

Actually if you think about modern computing and 3D graphics, then even a Raspberry Pi 4 would be more capable than the DE-10.

Where FPGAs excel at is real-time signal processing, hence the low latency / low lag that can be achieved for old consoles. It is also good for upscaling, audio processing, etc.

Edit: I just remembered this project, they discuss how a traditional CPU wasn't good enough, so went for am FPGA approach:
https://hackaday.com/2019/07/24/xfm-a-3 ... n-an-fpga/

The MiSTer framework can also be used to implement things that are not old games consoles nor computers.

For example Conway's Game Of Life :
https://github.com/hrvach/Life_MiSTer

Two interesting development ideas for which Terasic DE10-Nano seems ideal could be:

- Winamp's Milkdrop-like video generator taking audio as input, with analog VGA and HDMI output.
- VGM player for multiple synthesizers using a top board with an OLED screen and buttons as the interface.

I really like the music visualizer idea. Though that would mean I'd probably have to add the analog audio input and modify my case for it.