MiSTer FPGA Forum

Hi,

I am very new to MiSTer fpga and just starting to learn about it. I love arcade games especially capcom CPS1, CPS2 and neogeo MVS

I want to know if these arcade boards are perfectly replicated using MiSTer fpga

Thanks in advance

The MiSTer wiki has all systems listed that have had public releases but:

Neo Geo is working AES/MVS
CPS1 has partial releases, you can support Jotego on Patreon if you like his work.
CPS2 no, have not seen any cores yet, maybe there are people working on it, maybe not.

Neo-Geo is OK now, some tweaks were being worked on last time I checked, but mostly, it's accurate enough to ditch your original hardware and software.
CPS1 is being worked on and is already quite mature to be playable and sell your original hardware.
CPS1.5 is being worked on and CPS2 will be done once 1.5 is done.

So, basically, if you really play your games one at a time in order to 1cc each, then you can invest in the MiSTer and Jotego's Patreon and start with Neo-Geo library.

CPS 1.5 and CPS2 are likely to be available in the next months as Jotego is working on them actively. Be aware, thought, that CPS2 will require 128MB of RAM for a number of games -no more details are known about which exactly-.

The arcade boards aren't technically "perfectly replicated", but the logic of their digital circuitry has very accurately been reimplemented in FPGA. Probably a semantics thing, but it's a common misconception.

aberu wrote: ↑Thu Oct 15, 2020 8:09 pmProbably a semantics thing, but it's a common misconception.

Would it be more accurate to say it's as good as it gets unless somebody clones the board chip-by-chip?

Newsdee wrote: ↑Fri Oct 16, 2020 3:51 am

aberu wrote: ↑Thu Oct 15, 2020 8:09 pmProbably a semantics thing, but it's a common misconception.

Would it be more accurate to say it's as good as it gets unless somebody clones the board chip-by-chip?

The interesting thing is if the logic is identical, the timings are the same (where it matters), etc... it shouldn't matter and it wouldn't be worth the effort to 1:1 represent all of the circuits accurately. Nowadays, for instance, modern CPUs are designed in software and the hardware architecture is automatically generated for the most part. The PSX was the first gaming console to have an automated design for its architecture, for instance. To play a PSX game on FPGA 100% accurately and to represent the logic of the system would not require perfectly mirroring all of the components architecture entirely in FPGA.

Thanks for the replies guys, its very enjoyable and informative.

I have also been also looking at another route to play neogeo games which is much more expensive. It is to use a dedicated flashcart like the NeoSD AES flash cartridge to run on an actual neogeo AES console.

Do the games using this method run closer to arcade perfection than to MISTer, or is the latter closer to arcade?

The Neo•Geo core is very close to perfect accuracy, possibly only surpassed by some of the console cores (e.g. SNES). Probably one of the best value-for-money cores on the MiSTer, alongside PC Engine.

There is nothing to lose by just diving in with MiSTer, there is value as is and way more as time goes by. I really regret getting a MiSTer said no one ever. All ready you have the classic arcade games, plus Neo Geo plus CPS 1 plus all the game consoles plus all the handheld consoles plus all the computer cores plus all the extra cores. really name one place where you can get the same value per dollar. Mind you to get the same quality you would have to have hdmi / rgb mods for all those consoles / systems and other mods. It is a no brainier. If you had this just for Neo Geo it would still be worth it even ignoring everything else it can do. Mister Addons is working on a Jamma arcade addon that would make any existing arcade machines you have on hand a simple plug and play addon. So the one question I am asking is are you game?

aberu wrote: ↑Fri Oct 16, 2020 4:07 pm The interesting thing is if the logic is identical, the timings are the same (where it matters), etc... it shouldn't matter and it wouldn't be worth the effort to 1:1 represent all of the circuits accurately. Nowadays, for instance, modern CPUs are designed in software and the hardware architecture is automatically generated for the most part. The PSX was the first gaming console to have an automated design for its architecture, for instance. To play a PSX game on FPGA 100% accurately and to represent the logic of the system would not require perfectly mirroring all of the components architecture entirely in FPGA.

That may be so, but this approach kinda cancels the whole "hardware preservation" angle, which is a big pro-MiSTer argument.

@rikdenio: if you want 100% perfection then nothing will beat original hardware. Thing is, it's extremely unlikely you will be able to tell the difference compared to MiSTer implementation. The original HW also has the "realness" factor though and many people like to use it because of nostalgia, and to keep the old tech alive. So if you reallylove NeoGeo games and have the spare funds then go for it.

Problem is, you will still need some solution to play the other arcade games. I personally use MiSTer for some of them (0 lag) but also have a GroovyMAME PC + CRT setup, which has also a very low latency. If you want all-in-one setup for arcade and aren't that interested in computers/consoles cires then this is probably the best way.

akeley wrote: ↑Mon Oct 19, 2020 7:58 am

aberu wrote: ↑Fri Oct 16, 2020 4:07 pm The interesting thing is if the logic is identical, the timings are the same (where it matters), etc... it shouldn't matter and it wouldn't be worth the effort to 1:1 represent all of the circuits accurately. Nowadays, for instance, modern CPUs are designed in software and the hardware architecture is automatically generated for the most part. The PSX was the first gaming console to have an automated design for its architecture, for instance. To play a PSX game on FPGA 100% accurately and to represent the logic of the system would not require perfectly mirroring all of the components architecture entirely in FPGA.

That may be so, but this approach kinda cancels the whole "hardware preservation" angle, which is a big pro-MiSTer argument.

Hardware variation from the manufacturer, especially in later systems with automated hardware design and smaller architectures (like the PSX core coming up), means this is probably pointless to pursue. I think that for projects where it's practical or desirable, a 1:1 representation is fine, but there are many projects where this is both too difficult (meaning 5 years of 5-10 people dipping psx chips from multiple revisions into acid baths and using high powered microscopes etc...) and unnecessary. So I guess can be convinced to sort of share your concern about a potential difference in your use of hardware preservation and how it's used here.

I do not see the average MiSTer user scrutinizing the digital circuitry themselves and caring, so I disagree with it being a big pro-MiSTer argument. With Analogue products being closed-source, there is no confirmation as to those being 1:1 hardware preservation either. The average user wants a highly if not 100% accurate gameplay experience that is just as low lag as it is on original hardware. That's the selling point of FPGA reimplementations to the average person. The market for people that want the hardware to be perfectly preserved has got to be far lower.

aberu wrote: ↑Mon Oct 19, 2020 10:21 pm I do not see the average MiSTer user scrutinizing the digital circuitry themselves and caring, so I disagree with it being a big pro-MiSTer argument. With Analogue products being closed-source, there is no confirmation as to those being 1:1 hardware preservation either. The average user wants a highly if not 100% accurate gameplay experience that is just as low lag as it is on original hardware. That's the selling point of FPGA reimplementations to the average person.

That's not true in my experience, and I'm a pretty "average user". I only got MiSTer a few months ago, but was following it (and FPGA scene in general) for several years, in retro media and forums. Two major arguments in its favour have always been the same: low latency and reimplementing the original HW . The latter is connected to preservation and being a big factor in the whole "feel" people like to talk about. I've seen it brought up countless times in the "vs emulation" arguments, and being mentioned as important by people inquiring about FPGAs.

Example: http://www.racketboy.com/retro/mister-f ... eservation

It is also worth stressing the FPGA technology is also one of our best ways to possibly preserve the “embodiment” of a console or personal computer for decades to come. SmokeMonster explains this so well:

“I should preface the reason why so many of us are interested in FPGA-based hardware emulation by denoting exactly what FPGA code is. Verilog and VHDL are hardware description languages that were originally designed to document chips both for manufacture and research. So, instead of random blueprints for electronic circuits, manufacturers standardized them in Verilog & VHDL. They could describe a single chip or an entire system. For example, if you have a Neo Geo custom IC documented perfectly in Verilog, you can send that off to factory and produce the actual chip. Verilog and VHDL are the languages used to write FPGA cores, and offer a theoretical level of perfect accuracy to original hardware. Of course, that always comes down to the quality of an implementation and of other technical restrictions, and few cores ever achieve true “cycle accuracy”. But that is the ultimate goal of any core, even though it could take a decade or more of open-source contributions and research to get there. Open-source FPGA cores are permanent preservation of hardware, so the MiSTer project is long-term. Although it’s currently based on the Terasic DE10-Nano, it will be relatively simple to port its cores to future devices. When all of the original hardware and custom chips have died, FPGA recreations can be used to emulate or even produce physical replacements“

Reimplementing the original hardware ≠ replication of the original hardware.

SmokeMonster, as far as I'm aware, has not developed any of the cores extensively. I have not either btw. My understanding comes from talking to a few of the developers of the MiSTer cores online here and there. I could be wrong, sure.

But the only usages of "replication" in that article come from SmokeMonster and Colin (who I believe is Porkchop from MisterAddons). Smokemonster's comment about Burgertime and Ghosts 'n Goblins running noticeably faster in MAME for instance is troubling since you can achieve cycle accurate emulation in software regardless, this isn't an argument against software emulation itself. You cannot however achieve reliable and 100% accurate experience, down to the latency, without original hardware and/or FPGA. Software emulation has to do tricks to try and hack the latency difference (run-ahead) at significant cost with many bugs introduced. You also cannot reimplement the parallel logic in digital circuitry, so to speak, in software emulation the way you can with FPGA.

I don't know if SM would phrase things quite the same way today, over 1 and a half years later.

It's also apparent to me that Sorg uses the term replicate here and there. I don't think he may have meant it identically to the way you mean it here, given his long history on github issues of turning down certain ideas. For instance, the ao486 core won't have a complete ODE in FPGA, it's doing a workaround and doesn't currently read CD's in the same way an IBM-compatible 486'ish PC would. Is this an inaccurate travesty? Probably not. People have asked for full MSU-1 support for the SNES core because it would be technically complete, it's been shut down since it's probably nto going to fit or not practical and not worth their time, etc...

The current region code detection in the Genesis core, which is something I understand a tiny bit about, is not literally having the software check the registers of the specific chip for the current region the way hardware would blindly, because... the Genesis core is not literally a region-coded chip. It adapts. the Genesis core checks for the header of the ROM first, then sets up the register so when the ROM does a region-code check it sees the desired register. A decision was made to have auto-region detection built in, and this is not representative of the original hardware at all, just the original logic after someone hypothetically plugged in a cartridge into the right or wrong system and pressed power.

This really isn't about semantics or how many times the word "replication" was used. The context is pretty clear, from the headline itself to certain quotes, such as "Open-source FPGA cores are permanent preservation of hardware" or "FPGA recreations can be used to emulate or even produce physical replacements". That's the larger narrative I was talking about earlier, and which is very popular outside specialized forums such as this one.

What you say about software emulation using "tricks" is irrelevant in the context you mentioned before: average user does not care how something is achieved as long as it is working. I'm not sure what "reliability" is, but accuracy is actually better in emulation for many cores (mainly microcomputers) and probably on par for consoles. The latency is still the major selling point for MiSTer, no doubt, though it's also not as huge a factor as some would have it. You can achieve good results in Retroarch or other emus such as WinUAE already, even without runahead.

Anyway, I'm not going to argue about it, I'm just pointing out this angle to some lurkers, so they can make their own mind up. Myself, I will still promote MiSTer to people, but definitely without that "preservation" angle.

akeley wrote: ↑Tue Oct 20, 2020 2:31 pm This really isn't about semantics or how many times the word "replication" was used. The context is pretty clear, from the headline itself to certain quotes, such as "Open-source FPGA cores are permanent preservation of hardware" or "FPGA recreations can be used to emulate or even produce physical replacements". That's the larger narrative I was talking about earlier, and which is very popular outside specialized forums such as this one.

What you say about software emulation using "tricks" is irrelevant in the context you mentioned before: average user does not care how something is achieved as long as it is working. I'm not sure what "reliability" is, but accuracy is actually better in emulation for many cores (mainly microcomputers) and probably on par for consoles. The latency is still the major selling point for MiSTer, no doubt, though it's also not as huge a factor as some would have it. You can achieve good results in Retroarch or other emus such as WinUAE already, even without runahead.

Anyway, I'm not going to argue about it, I'm just pointing out this angle to some lurkers, so they can make their own mind up. Myself, I will still promote MiSTer to people, but definitely without that "preservation" angle.

You? Argue?

Your wrong on many of your points there but I for once can't be bothered to point it all out. There are only so many hours in the day!

I will say a lot of the emulators can't claim they are cycle accurate and sorry, but lag is a HUGE issue on PCs which largely doesn't exist on MiSTer.

WinUAE is probably more advanced than the minimig core, but still suffers the same issues mentioned above.

No emulator can claim it will natively support OG display devices either (and if they do, they're lying). Groovymame is the closest it comes (even CRT output through retroarch isn't perfect and I can vouch since I worked on it).

Your trivialisation of the project is insulting to be honest. It IS a preservation project, just as emulators are. A lot of time, blood, sweat and tears have gone into both.

retrorepair wrote: ↑Tue Oct 20, 2020 3:45 pm Your wrong on many of your points there but I for once can't be bothered to point it all out.

Perhaps because it's easier to launch into a load of hysterically emotional guff instead? Nothing new here indeed, I had dubious pleasure of sampling this style of yours before.

It's not for the first time you choose to ignore everything that was written before and launch into a pointless rant instead. I actually have a spare minute, so let's break this down, shall we?

-I did not say all emulators are cycle accurate, but that they are at the moment more accurate in general, which involves many other factors than cycle accuracy.
-I've already stated that low latency is a huge selling point for MiSTer. That does not change the fact that emulation is not that far off. Such experienced coder as yourself, who has after all worked on Retroarch, should be aware of this simple fact.
-not sure what your comment about CRTs has to do with anything, apart from scraping the barrel, but since we're at it, yes crtemudriver is pretty close (some would say right there), and you can get composite out of RPi natively, RGB with an add on (just like in MiSTer, eh?). There's also Wii, no too bad a solution either.
-if you had any reading comprehension at all, and actually bothered to read this thread you'd understand that my comment about not promoting MiSTer as preservation project stems from the overarching narrative in the wider world, which implies that this is the only true preservation method vs emulation, which isn't. Of course, they both are, but none is significantly better than the other in this regard.

To cap it off, your inane accusation of me "trivializing" this project is symptomatic of cultist attitude which unfortunately some of people involved with it flaunt quite often. It is, frankly, pathetic. MiSTer is an awesome project indeed, but trying to be objective in regard to its strengths and weaknesses can often set off some zealous reactions. It's rather tiresome, so like I said this is my last "explanation". From now on I will simply ignore it.

My PC running bsnes will consume about 100W of power on average. I could probably put together a similar PC today for about $500 (excluding case).

For $150 (DE10 + RAM) I can get a machine that has the same kind of accuracy as higan but uses only ~5W of power (1 amp 5V). All other things being equal in term of accuracy, I can run 20 MiSTers for the same electricity cost as my PC.

FPGAs open source is preservation in the sense that 1) the output of several cores has been objectively measured to match reference material (e.g. using Artemio's MDFourier), 2) the open source nature makes cores portable to new platforms as tech advances, and 3) cores can be cut down in pieces to create replacement parts of the original hardware. In theory you could make a motherboard for a SNES or NES with a tiny FPGA (removing cart mapper logic) and put it in an original shell to accept carts, and nobody would know the difference.

In terms of input lag, it's no secret that MiSTer uses USB as its controller solution and achieves low latency by forcing USB polling at 1ms. That is perfectly possible on other platforms, but not all of those platforms give you the option to override the polling in the same way (especially not the random emulation boxes that are now popping out from various companies)

Don't get me wrong. I think the NES / SNES classic are perfectly valid solutions to play the included games on HMDI if one doesn't get too hung up on the compromises they have. But if somebody starts talking about hacking it to add games, or whether it's a good replica of the original experience, then they can't hold a candle to an FPGA device (MiSTer or otherwise). A DIY Linux box with CRT output could get close to a MiSTer, but it does have much higher setup effort and higher hardware cost, at least for the average person.

akeley wrote: ↑Tue Oct 20, 2020 4:40 pm From now on I will simply ignore it.

That's the first thing you've said that I can identify with.

Just keep talking nonsense, your the one who ends up looking like an idiot.

I would say that unless you are doing the "hardware" coding in the Mister....then you are just like the rest of us....beggars. No need to argue or insult each other. We have people re-creating the experience of original hardware....and due to the nature of FPGA is less laggy than software emulation. And the Thread is titled "Mister as an Arcade System".....from my very limited knowledge.....it would appear that Mister would make a nice Arcade system.....not perfect....but most arcades I remember playing had a lot of issues of their own....no reason to clone the errors. So I believe the answer to the first question is No...not perfectly replicated from a hardware perspective....but perfectly replicated from a fun "lag less" experience that simulates almost perfectly the fun that an arcade represents.

Maybe instead of arguing....you guys can figure out how to get more people to buy them and sponsor the people that program them....because that is a good use of time. And frankly, I love both hardware and software emulation....and they both have their place in gaming use.

But your argument is definitely a good read for sure.