Mister FPGA: The Future of Game and Computer Emulation | Hacker News Reader

Mister FPGA: The Future of Game and Computer Emulation

racketboy.com

109 pointsecliptik64 comments

I would say miSTer's weakest point is that it isn't designed for the open FPGA synth, place and route stack.

It depends on proprietary tools tied to Intel's FPGAs.

This is somewhat understandable, as the open stack wasn't ready when miST was conceived.

Part of the appeal of mister is it runs on cheap "dev" FPGA boards that have their price reduced to appeal to students and educators. Those are typically boards from Intel and Xilinx. It's been awhile since I looked at open FPGA synth tools, but I remember that Intel support was non-existent and Xilinx support was rudimentary. You couldn't get the same performance and packing as you can with the proprietary tools.

The Xilinx -7 series of FPGA is fairly well supported by SymbiFlow these days. https://symbiflow-examples.readthedocs.io/en/latest/building...

Unless it changed recently, I think it doesn't have full support for BRAM, and no support for BSP and hard blocks, right ? I wouldn't be surprised if some were used in the Mister project. (Doesn't change the fact that Stmbiflow is an impressive feat)

I don't know the state of Xilinx family 7 open source support, other than there has been work on it for years.

A competitive chip that is well supported (including BRAM etc) is the Lattice ECP5, as found in e.g. the ULX3S dev board.

>price reduced, cheap, etc

Note that the chip used in that board is listed in retailers such as digikey. It's not so expensive as to prevent third party boards.

It would even be well feasible to use a larger FPGA or a higher speed grade. FPGA chips used in devboards are typically the lowest grade.

I'd say its weak point is the appalling user interface. Really hard to understand all the menus and modes and how to navigate. (Yes as a programmer I ought to pitch in to fix it)

Are you talking about the dev tools or the mister itself? I have a 7,4 and 3 year old. All three prefer the mister menus over retroarch and they can put their favorite games on and do save states etc themselves no problem.

The MiSTer itself. It's still a great system.

Surprising. What about the menuing itself do you dislike. My kids can’t get around retroarch no matter the theme. Mister they just go to the core. Select it. It drops in in the cord directory of games. They select it. And they play. Retroarch is Roms folder then they select the core and if they don’t select the right one their save states aren’t there and their configs aren’t there. Good luck doing 5x scaling with retroarch too.

That's kind of my experience with the MiSTer UI - confusing, deep menus, unclear what button gets to what menu or how to get back, etc. I now wonder if there may be several MiSTer UIs and I'm just using the wrong one?

My only experience with Mister is playing SNES games with the device my brother set up on a CRT tv. Playing an SNES game is just a matter of choosing the SNES core, picking the game rom from the default directory, and learning the hot key combination on the controller to go to the save menu to save your in game save file to the sd card.

For this use case it's easier than Retroarch- you don't need to fiddle with resolutions since everything outputs natively to the CRT and you don't need to fiddle with cores since only 1 core exists per system.

Maybe there's more complex features that are harder to use, but as a replacement for a super nintendo I find it to be dead simple to use once setup.

The Mister Multisystem appears to be a great improvement over the de10-nano. LGR did a video on it:

https://www.youtube.com/watch?v=qx45r-BRHxY

https://rmcretro.store/multisystem-board-only/

This plugs into the FPGA dev board, doesn't it?

Yes. It still requires the nano.

Thanks for the reminder. This is very high on my want-list. Do you know what the "deluxe edition" mentioned at the store page is?

The Deluxe includes the DE-10-Nano FPGA board. The other editions require you to buy the DE-10-Nano board yourself.

The Mister ecosystem is based on open hardware / software, besides the FPGA which is a necessary evil (and can be replaced in the future).

The "mister multisystem" (which is a proprietary design) comes across like a slap in the face. Don't support blatant opportunistic profiteering.

No one is taking the ecosystem from you. This board is just one expansion option.

There's a fair few commercial sellers for MiSTer related stuff. Yes some just build existing open hardware designs, but not everything is open hardware.

E.g. the MiSTercade https://misteraddons.com/products/mistercade or the MiSTer mini ITX Ironclad Plus https://www.d3fmod.com/mini-itx-ironclad-plus/

None of these are trying to lock people into their solutions, they're just providing other options.

The mister multisystem is not an improvement over the de10-nano. It's just an alternative in a different form factor to the open sdram module and the analog io board. You still need a de10-nano to plug into it.

It's not the first time I see this confusion but I'm a bit surprised by it because they also make it very clear that "A DE10-Nano is required to operate"

Oh my, you are correct. Looking at the board and nomenclature, "multisystem", this should have been obvious to me. Thank you for the clarification.

When it comes to the detailed timing and edge cases of hardware, especially peripheral hardware, software-based hardware emulators try to cover the most important cases. As far as I understand, an FPGA-based emulator would try to emulate the structure of the hardware, not its behaviour, thus resulting in more accurate emulation. Where do they plan to get information / documentation about the structure of the hardware? AFAIK most of the behavior is reverse-engineered by trial and error.

At least in the case of the Neo Geo core, the chips were decapped and fully reverse engineered. I'm unsure about other cores.

https://www.retrorgb.com/furrtek-demonstrates-decapping-proc...

If you're decapping old chips, do you even need to reverse engineer them, or can you just blindly convert a high-res photo to something an fpga can run?

If you look up Robert Peip, he's detailing a huge amount of what he's doing as he builds his PSX core. It's a combination of work done on emulators, getting expert feedback on the MiSTer, previous work on supported chips and docs from where ever he can find them. It's pretty incredible, I've supported him on patreon for about a year and have been exposed to so much interesting work.

The best thing about this project is the seemless CRT TV support. There is no simple way to use a PC emulator and support CRT output.

Specifically 15 kHz. If you're okay with rough approximations then 480p VGA will be okay. If you really want the original art then you need to bend over backwards to get the original, strange aspect ratios and non-square pixels. It is indeed nice to have a project that outputs authentic analog video signals.

This is a 2019 article and quite a bit out of date. There's a dedicated forum for the project: https://misterfpga.org/

Many more cores have been supported or further polished, well into the 32 bit era - Playstation and Sega Saturn cores are in the works.

The general purpose computer cores are what interest me the most about the MiSTer project. I'm keeping an eye on the 486 PC and Macintosh Plus cores. Emulators are mostly fine for what I do but I would just love to have dedicated hardware instead. It's a shame the DE-10 Nano is out of stock everywhere...

The ao486 core is much improved since this article was published in 2019 - it's had caches and a VGA chipset added, which make most early 90's games playable to some extent.

Might want to check out the Vortex86. Other than sound, it might support what you want.

Why do you want to emulate a 486 with an FPGA?

What happens when the DE10 Nanos run out?

Presumably it gets ported to another architecture.

Or, simpler, somebody comes up with a OSHW board that's actually fit for purpose, rather than leverage a generic devboard.

https://rmcretro.store/multisystem-board-only/ Seems like a good start

That still needs a dev board mounted on it.

There are lots of boards being designed, some of which have been open, but they tend to be expensive because of the small runs.

The open cores tends to end up ported to multiple platforms over time.

Is that simpler? Dev boards are a dime a dozen, and far from everyone who wants to use FPGAs to mimic game consoles wants (or can) solder them themselves... I feel like if this platform runs out, people will mostly switch to another generic one with not too much effort.

By now, I would dare say miSTer has enough traction that a custom board would make sense.

As a reminder, few people use bare DE10-Nano. It is generally used together with custom boards for RAM and I/Os, designed for miSTer use.

There's nothing magical about FPGAs or SMD as to prevent a main board designed for the purpose.

Is there any upside to replacing it with an ASIC specifically for miSTer?

I can't think of any, relative to a FPGA. MiSTer cores are regularly upgraded; an ASIC would defeat the point.

What's being proposed is a board that's designed for miSTer. Such a board would include an appropriate FPGA, alongside memory, a better cooling solution and ports useful for miSTer use.

These are currently accomplished via mods/addons to the generic devboard, which is far from ideal.

More downsides I'd say, the point of the FPGA is that you can always replace and update the cores according to what you need any given moment, just "download your game console" essentially. Physically swapping hard ASICs isn't as convenient. :)

Electrical engineers are cheap. Porting to a slightly different layout can be easily crowdfunded.

I've been hoping it somehow ends up available for pano logic client devices. No idea how big a project that be but it would be pretty cool since there are a reasonable number of them out there, they are kinda cheap, and are already in a nice case.

Hopefully terasic will keep making them for a good while and when they do retire them have a replacement.

It's not as simple as just develop a new board as the DE10 Nano offers a great price / performance ratio. There are more capable FPGA boards out there but they're also more expensive. Any purpose developed board would also likely cost more.

Amazon.com: DE10-Nano Kit

  Currently unavailable.
  We don't know when or if this item will be back in stock.

Also, on Mouser:

  Stock: 0
  On Order: 2,877 Expected 1/28/2022
  Factory Lead-Time: 48 Weeks
  Long lead time reported on this product.

ooh glad I squirrelled 2 of them away pre-COVID

I believe they already have, at least interntionally.

Can it do 3dfx?

No. And it can barely do anything without crashing. I hate mine.

Reply on news.ycombinator.com