• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Nintendo 64 MiSTer FPGA officially out today; 99% of a full N64 in FPGA

VGEsoterica

Member
Ive posted some milestones here and there when the N64 FPGA core hit certain goals; first game booting / 100% of North American library booting, overclocked core...etc etc etc...but that was always on test cores uploaded that 9/10 MiSTer owners never touched or wanted to test. Everyone wanted to wait until it hit update_all

well that day is today. Outside of about five games that have some issues a full fat N64 is in FPGA with crystal clear HDMI out or RGB analog out all for the princely sum of "zero...just gotta own a MiSTer"

I am sure some people on GAF will be stoked to hear that and maybe a few others might want to get into MiSTer now

 

RagnarokIV

Battlebus imprisoning me \m/ >.< \m/
Awesome. Will update all when I start yet another Legacy Of Kain marathon on the weekend.

Looking forward to replaying Glover, Pokémon Stadium and a few others on this.
 

Esppiral

Member
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
Tb7yIoT.jpg
 

dave_d

Member
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
In a word, timing. So Windows and Linux are not real time OS'es and they make no guarantee on anything timing related. They don't make any guarantees when your program will get a timeslice on the CPU, how long it will take to load it, or how long it will get to run on the CPU. What this means is that when the emulator runs it can run at a different speed than real hardware would. The FPGA tries to guarantee not only the same result (data) but also running in the same amount of time. Admittedly the emulator authors have gotten really good at getting them to run and very close to the same speed so you'll probably never notice this.
 
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
Well for Analogue products the ability to run carts in real time. Latency is also a big one. While there are ways to lower latency via software emulation, you do need a relatively powerful PC to handle that and even then the consistency of latency won't be as good as a dedicated FPGA device.
 
So what happens when everything before 32-bit is emulated perfectly with the MiSTer? What would be left to work on, or would we move to different hardware?

Will there ever be a direct successor that allows for a better UI?
 

spons

Member
So what happens when everything before 32-bit is emulated perfectly with the MiSTer? What would be left to work on, or would we move to different hardware?

Will there ever be a direct successor that allows for a better UI?
I don't think there is an accessible successor in terms of hardware. The Cyclone V was released in 2011, and there's something from 2017 that's not really an improvement in terms of LE. Somewhat odd if you ask me, but there probably isn't mass-market appeal for anything drastically more powerful.
 

VGEsoterica

Member
So what happens when everything before 32-bit is emulated perfectly with the MiSTer? What would be left to work on, or would we move to different hardware?

Will there ever be a direct successor that allows for a better UI?
Dreamcast / NAOMI are in the realm of doable on more advanced yet still affordable hardware FPGA wise

PS2 gen and on forget about it. Not only would the hardware cost a fortune but this isn’t a “one man develops a core in 1-2 years”…you are talking a team of people working full time do it in 1-2 years. The level of hardware complexity from N64 and even DC to PS2 / GameCube / Xbox is staggering
 

Pegasus Actual

Gold Member
Yo we got like a buyer's guide thread? I'm a little confused... I think DE-10 Nano is something I need... then a case... and then you need add-on boards for audio/video and input? And SD card for storage? That about the long and short of it?
 

Pagusas

Elden Member
Dreamcast / NAOMI are in the realm of doable on more advanced yet still affordable hardware FPGA wise

PS2 gen and on forget about it. Not only would the hardware cost a fortune but this isn’t a “one man develops a core in 1-2 years”…you are talking a team of people working full time do it in 1-2 years. The level of hardware complexity from N64 and even DC to PS2 / GameCube / Xbox is staggering

Considering how close in power the DC and PS2 were that’s surprising, is it because of the PS2’s more custom hardware?
 

Puscifer

Member
Analogue's 3D console won't be long now..
Tbh considering the power it takes to get N64 working accurately the ond releasing this year might end up being a pocket for consoles but who knows if they suddenly have pocket cores out of nowhere again.

What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
By and large, accuracy. Mother 3 is shockingly busted in ways if emulated, the music combos are badly timed but with the pocket it worked just fine. I sold mine and borderline want another one because I thought my steam deck would be a great replacement but... It's just not? Bust a groove is also infamously broken, duckstation is the closets to getting it, and PCSX2 is also really good with rhythm games (I tested with Brittneys dance beat, feels 1:1)

It also is the closets thing you'll likely ever get to preserving your physical collection of retro games. Those consoles and handhelds are getting older and more expensive. For instance, a pocket is the same price or cheaper than a used GBA and GBC and the additional cores.


Sega Saturn also is having this resurgence, but again, low console sales + scarcity + near impossibility to emulate properly = even higher prices. It could benefit from an fpga as well

I guess all in all it's a collectors things I guess
 
Last edited:

nkarafo

Member
Literally perfect, 100% accurate emulation for the purists
In a word, timing.
By and large, accuracy.

Not this again... VGEsoterica VGEsoterica my bro, could you please put a disclaimer every time you post a thread, that FPGAs are not inherently more accurate than software emulators? You obviously love FPGAs and like to inform people about them, you could help stop this misinformation that always spreads in your threads, thanks.

Edit: You could use this link like a signature: https://archive.ph/4Umsk


What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
See the link i posted above but here's a summary:

-They are more compact than a PC powerful enough for the same accuracy in emulation.
-They are more efficient, power wise. A regular CPU needs to draw more power to deliver the same amount of emulation accuracy.
-They don't have an OS, which helps with input lag. Though software emulators have different ways to reduce latency.
-It's easier to connect an old CRT TV with them.

When it comes to accuracy they are no different than a software emulator other than they are more parallel, which is more expensive on a regular CPU performance wise.

Other than that, accuracy only depends on the author, their knowledge and skills, just like every emulator. There are software emulators more accurate then FPGAs and vice versa. The chip has nothing to do with it.
 
Last edited:

dave_d

Member
Not this again... VGEsoterica VGEsoterica my bro, could you please put a disclaimer every time you post a thread, than FPGAs are not inherently more accurate then software emulators? You obviously love FPGAs, you could help stop this misinformation spreading, thanks.
Umm sorry but timing is the one real difference and latency is a specific example of a timing difference. BTW you bring up the thing from Byuu and they mention the whole turing complete point. If you do a CS program they'll go way in depth on that and it will be pointed out that anything that one turing complete device can compute any other one can compute with one huge caveat. As long as you don't care how long it takes to run, which can be faster or slower. (Literally you could write an emulator for say the SNES in excel because excel is turing complete. It would just run with really weird timing.) BTW people should look up how little you need to have a turing complete device. (It's not very much.)
 

nkarafo

Member
Umm sorry but timing is the one real difference and latency is a specific example of a timing difference.
You can have the same perfect speed timings on a software emulator. Input lag is a different issue and isn't a problem caused by the emulator itself but the OS. This is also explained in the article i linked.
 
I love your channel friend, I recently got an Analogue Pocket and it's been a blast. Maybe one day I'll get a MISTER once they become a bit more user friendly.
 

dave_d

Member
You can have the same perfect speed timings on a software emulator. Input lag is a different issue and isn't a problem caused by the emulator itself but the OS. This is also explained in the article i linked.
Again, you're talking about OS'es like linux and Windows that don't guarantee timing. (That's literally one of the things why they're not a real time OS. One of the standard definitions of a real time OS is if an answer is late it's wrong. On an OS like Windows and linux they're not real time so they can absolutely be late. For example you have a ton of other programs running sucking down time slices so your emulator doesn't get its time slice when it needs it.)
 

nkarafo

Member
Again, you're talking about OS'es like linux and Windows that don't guarantee timing. (That's literally one of the things why they're not a real time OS. One of the standard definitions of a real time OS is if an answer is late it's wrong. On an OS like Windows and linux they're not real time so they can absolutely be late. For example you have a ton of other programs running sucking down time slices so your emulator doesn't get its time slice when it needs it.)
Are you talking about timings that relate to the emulation itself, like game speed and frame rate? That's not an issue for emulators, these timings can and have been perfect in software emulators.

Are you talking about input lag? Different issue. But again, this isn't an issue of the emulator. Like the article says "there is nothing preventing an emulator from running on bare metal, without an operating system in the way". It's just that nobody has bothered doing that yet. The difference is too small for this to be worth it i suppose.

There are emulators that are still more accurate than FPGAs btw. Ares is more accurate than Mister N64, BSNES is more accurate than any FPGA SNES core. Beetle Saturn, i'm pretty sure it's more accurate. I'm not sure about most other systems because there isn't enough information about this. I do know that some standalone arcade board FPGAs are more accurate than MAME.

It depends on the author and their work, not on the type of chip used.
 
Last edited:

dave_d

Member
Are you talking about timings that relate to the emulation itself, like game speed and frame rate? That's not an issue for emulators, these timings can and have been perfect in software emulators. There are emulators that are still more accurate than FPGAs btw. Ares is more accurate than Mister N64, BSNES is more accurate than any FPGA SNES core.

Are you talking about input lag? Different issue. But again, this isn't an issue of the emulator. Like the article says "there is nothing preventing an emulator from running on bare metal, without an operating system in the way". It's just that nobody has bothered doing that yet. The difference is too small for this to be worth it i suppose.
I think we're actually talking about the same thing now that I think of it. The emulator itself can be perfectly accurate. The system that runs it can cause it to run faster or slower and that depends. IE every frame of video that the emulator generates could absolutely match one generated by hardware perfectly for example. However the amount of time it takes between frames might not match. Admittedly I know I'd have no chance at noticing a difference as small as that. Not that "Hardware" is inherently more accurate. For example different revisions of the Genesis hardware sound different even though it's all hardware.
 

nkarafo

Member
I think we're actually talking about the same thing now that I think of it. The emulator itself can be perfectly accurate. The system that runs it can cause it to run faster or slower and that depends. IE every frame of video that the emulator generates could absolutely match one generated by hardware perfectly for example. However the amount of time it takes between frames might not match. Admittedly I know I'd have no chance at noticing a difference as small as that.
Ok i think i know what you are talking about. But again, it's not an issue with the emulator. It relates to the OS/Video driver and API. It has to do with syncing the frames with whatever screen you are using and this can also be an issue with FPGAs btw. Doesn't have to do with emulation timings and accuracy though. It's also something that can be fixed depending on the API you are using, the panel and the emulator.

Some emulators will try to change the timings so you can have perfect sync with the screen. That's to avoid screen tearing or stutters and have a smooth scrolling. Yes, that's not an accurate behavior but you don't have to use it. And if you have a VRR display like myself, you can have absolutely perfect 1:1 displayed frames even with games that have odd refresh rates.


Not that "Hardware" is inherently more accurate. For example different revisions of the Genesis hardware sound different even though it's all hardware.
Since you mentioned that, there are emulators that emulate the different audio chips used in Mega Drive 1 and 2. I think there were 3 different revisions of the audio in general, all emulated. By default you are usually getting the first version, which is also the best for most.
 
Last edited:

Ozzie666

Member
To the people who buy into it? Pure Snob value of not being down among the plebs with their software only emulation.

In a blind test, put a properly configured emulator next to a FPGA core running the same game and I'm willing to bet no one could tell the difference.

Your probably right, I'm most likely one of those Analogoe / 8 bit do snobs. FPGA does seem better overall in some circumstances. But for me, having a replacement system that supports cartridges, controllers and accessories is really nice to have. Maybe it's that nostalgic feeling, but playing older consoles on something close to the original, just feels better and more enjoyable. Input lag better? maybe.
 

VGEsoterica

Member
To the people who buy into it? Pure Snob value of not being down among the plebs with their software only emulation.

In a blind test, put a properly configured emulator next to a FPGA core running the same game and I'm willing to bet no one could tell the difference.
Lol as someone who does tons of software emu stuff tutorial wise, has been accused of being an FPGA snob AND a hardware purist…

There are some emulators where you’d be hard pressed to tell the difference between it and FPGA and others where it’s pretty obvious right off the start. Just depends on which one.

But FPGA cores are not some smoke and mirror “look how much I paid for this” addition to retro gaming. Nor are the end all, be all of it either. It’s a different way to accomplish the same goal and when done properly can deliver better results. Are those results worth the board cost? Everyone gets to decide

That’s why I do stuff like “MiSTer FPGA N64 core” videos right next to “N64 on Steam Deck”…because all I really care about is people having fun playing games. Not as much so how they play them
 

nkarafo

Member
But FPGA cores are not some smoke and mirror “look how much I paid for this” addition to retro gaming. Nor are the end all, be all of it either. It’s a different way to accomplish the same goal and when done properly can deliver better results.
Could you elaborate? What do you mean by "better results"? Your response is somewhat vague.


That’s why I do stuff like “MiSTer FPGA N64 core” videos right next to “N64 on Steam Deck”…because all I really care about is people having fun playing games. Not as much so how they play them
I watch your channel ever since it was only about obscure hardware showcases. And right now it's probably the biggest (and best) channel on the FPGA topic. But i feel like you don't clear up things enough for your audience. Sometimes you try to be informative and educational and even attempt comparisons between FPGA and emulators (which i personally find a bit biased) and on the other hand you don't seem to care about the whole FPGA vs software emulation accuracy misinformation in your Youtube comments or your threads here.

You always seem to respond with "all I really care about is people having fun playing games" and i know you are honest. But your channel has became the go-to for people to get informed about FPGAs and even software emulators since there aren't many (or any) good Youtube channels on the topic. There are probably many people who get all their info from your videos. For better or worse, this means you have some responsibility.

Case in point, the title of this thread. What does "99%" means? What is that number reflect? It's not the games that simply boot, the Mister can boot 100% of the games. Is it the games that run perfectly? No that's not either because not even Ares (which is more complete than the Mister core) can claim such high percentage right now. But when you say it like that "99% of a real N64" in an FPGA topic, many people who are already misinformed or biased (and they are many) will automatically assume this percentage is games that have no difference VS a real N64, accuracy wise.

Truth is, timings on the N64 are hard. A lot of games have timing/speed issues in both the FPGA and the best emulators. Some subtle, some very obvious. You'd be surprised by how many they are. This is why i'm really curious about the Analogue N64 when it gets released. It has to be perfect in order to be sold. But something tells me i will see many of the same issues there too while the seller will boast about how perfect it is compared to emualtors, just like how they did with their previous products.
 
Last edited:

Dacvak

No one shall be brought before our LORD David Bowie without the true and secret knowledge of the Photoshop. For in that time, so shall He appear.
What are the five or so games that don’t work?
 

Ulysses 31

Gold Member
What are the five or so games that don’t work?
Conker's Bad Fur Day
Jet Force Gemini
Resident Evil 2
Pro Mahjon Kiwame 64
Battle for Naboo
Gauntlet Legends
Polaris SnoCross
Wonder Project J2(turb core doesn't crash)
Iggy's Reckin' Balls
 

nkarafo

Member
Knife Edge is also unplayable because it runs too fast. Which is an issue with all emulators as well. A lot of other games are playable but run at the wrong speed.
 

64bitmodels

Reverse groomer.
To the people who buy into it? Pure Snob value of not being down among the plebs with their software only emulation.

In a blind test, put a properly configured emulator next to a FPGA core running the same game and I'm willing to bet no one could tell the difference.
Fair but also, N64 is the one Nintendo console where Emulation for it is a pretty dogshit scene due to the complexity of the system, years of plug-ins and many graphical errors. Not to mention all the decompilation efforts are making the system more irrelevant by the day
 

VGEsoterica

Member
Could you elaborate? What do you mean by "better results"? Your response is somewhat vague.



I watch your channel ever since it was only about obscure hardware showcases. And right now it's probably the biggest (and best) channel on the FPGA topic. But i feel like you don't clear up things enough for your audience. Sometimes you try to be informative and educational and even attempt comparisons between FPGA and emulators (which i personally find a bit biased) and on the other hand you don't seem to care about the whole FPGA vs software emulation accuracy misinformation in your Youtube comments or your threads here.

You always seem to respond with "all I really care about is people having fun playing games" and i know you are honest. But your channel has became the go-to for people to get informed about FPGAs and even software emulators since there aren't many (or any) good Youtube channels on the topic. There are probably many people who get all their info from your videos. For better or worse, this means you have some responsibility.

Case in point, the title of this thread. What does "99%" means? What is that number reflect? It's not the games that simply boot, the Mister can boot 100% of the games. Is it the games that run perfectly? No that's not either because not even Ares (which is more complete than the Mister core) can claim such high percentage right now. But when you say it like that "99% of a real N64" in an FPGA topic, many people who are already misinformed or biased (and they are many) will automatically assume this percentage is games that have no difference VS a real N64, accuracy wise.

Truth is, timings on the N64 are hard. A lot of games have timing/speed issues in both the FPGA and the best emulators. Some subtle, some very obvious. You'd be surprised by how many they are. This is why i'm really curious about the Analogue N64 when it gets released. It has to be perfect in order to be sold. But something tells me i will see many of the same issues there too while the seller will boast about how perfect it is compared to emualtors, just like how they did with their previous products.
The problem is the more granular I get, the more people tune out. Although I did a full episode explaining the inherent technological difference between how an FPGA handles code vs how software emulation does.

Better results on latency and in some cores accuracy. Take Irem M92 for example. MAME has corrupted graphics on the title screen of Ninja Baseball Batman. That was not an issue in the core. And I showed that side by side. That’s just an example off the top of my head that I recall being one of the more visually apparent ones
 

nkarafo

Member
Better results on latency and in some cores accuracy. Take Irem M92 for example. MAME has corrupted graphics on the title screen of Ninja Baseball Batman. That was not an issue in the core. And I showed that side by side. That’s just an example off the top of my head that I recall being one of the more visually apparent ones
Yes but again, these games are more accurate in the FPGA core not because it's an FPGA but because the author happened to emulate these games better. They had better skills or understanding or they just did better code. I mean, do you really believe these games can't be fixed in MAME only because it's a software emulator?

There are other games and systems that are more accurate on software emulators as well, that doesn't mean anything. I'm not gonna tell everyone software emulators are more accurate because of that.

Neither software emulators or FPGAs are inherently more accurate, that's what i'm trying to tell you. They are just the platforms/tools. It's all about the people who write the code.


Fair but also, N64 is the one Nintendo console where Emulation for it is a pretty dogshit scene due to the complexity of the system, years of plug-ins and many graphical errors. Not to mention all the decompilation efforts are making the system more irrelevant by the day
This isn't an issue with N64 emulators anymore. Ares doesn't even have plugins. And the only accurate video plugin that matters for Mupen/PJ64 is Parallel RDP. Simple 64 is also a Mupen fork that uses Parallel and doesn't let you change plugins either.

The whole plugin system is a relic from the past. Current N64 emulators such as Simple 64 and Ares have advanced a lot the last few years, it's not like it was during the plugin era. And the Mister N64 core has not yet reached the compatibility of Simple64 or the accuracy of Ares.

Naturally, the average Youtube retro games collector will use some ancient version of PJ64 with GlideN64. And that's if we are lucky because i see many are still using the previous Glide64 plugin. I see youtubers doing streams with Zsnes and ePSXe. Why? Because these are probably the first results they get on Google. I was watching a Game Chasers stream the other day with Zsnes and the game had all kinds of bugs and of course everyone on the chat were calling emulators out. Keep in mind Zsnes was considered an obsolete emulator in 2004. No wonder emulation gets a bad name in the retro collecting circles.

And i'm not even going to mention all the shitty Android and pi emulators that are configured to run on the worst performing systems.
 
Last edited:
What is the least you'd need to spend for accurate 32/64-bit era emulation via software vs. FPGA? Smallest form factor possible as well.
 
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
From my experience, less latency and more accuracy. I could be wrong but that's what I perceive when playing on it. Not refering to the FPGA itself in regards to the accuracy, just the accuracy of the cores written.
 
Last edited:

Buggy Loop

Member
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?

Each logic elements of an FPGA is programmable to match the data, adresses and instructions as exactly the old system was made like. It's basically to make a copy of the old CPU and other chipsets into an FPGA, there's no higher level interpretation from a different language to send then to a modern processor to create what it should be like, which is software emulation. Some cores are exact recreations, some not, but still very close to the real thing. Some peoples have even delid old chipsets to make better VHDL codes to match the circuit.

As someone who learned VHDL for FPGA during electrical engineering, I kind of respect madly FPGAs. You basically rewire a blank state processor to match to perfection another chipset.

For software emulators, processors from PC are generally just steamrolling the problem away, even though the clock cycles of all console components don't make sense for the x86 CPU, it iterates so fast that it kind of keeps up. FPGA since it can do all chipsets in a perfect circuit, each clocks given by the FPGA can run the whole system at 100% accuracy. If old system had a chipset that would maybe skip a clock or more for whatever reason, you program that too. So you need way less speed than a processor that is running software emulation. On FPGA everything is parallel, if the user wants, it can do the maximum systems the FPGA can contain and run it all in a single clock instantaneously, no queue, no thousands of stalls to wait on the old slugish ass processors instructions to be done in the sake of accuracy.

I suggest looking also at my answers at the bottom.

Not this again... VGEsoterica VGEsoterica my bro, could you please put a disclaimer every time you post a thread, that FPGAs are not inherently more accurate than software emulators? You obviously love FPGAs and like to inform people about them, you could help stop this misinformation that always spreads in your threads, thanks.

I suggest this read


PCs have brute forced their way into it but for ridiculous amounts of processing power to match puny decades old chips, and some still have problems.

This guy's whole video is a pretty good explanation, I suggest watching it, but the part about software accuracy more importantly.



What typically happens with all software emulators, even the accurate ones, is that it depends on a huge effort from the team to hack their way around the problems with tweaks on timing values, patch out certain needs and areas of code. It's a lot of effort. They've done it for most of the games of course, over decades of effort and countless man hours and a monumental gap in power between the original hardware and the modern PC.

While when the FPGA core is up and running, that's it. That problably covers 99~100% of all game library to run as it was intended.

There's so much focus on popular consoles like NES/SNES/SMS/Genesis, etc that it looks nearly perfect and I would say indiscernible from FPGA

Then software emulation completely fails an amiga 500 while mister FPGA will cruise on that. But not popular so not talked about.

You can have the same perfect speed timings on a software emulator. Input lag is a different issue and isn't a problem caused by the emulator itself but the OS. This is also explained in the article i linked.

Not all software emulators have the same accuracy. It's a tradeoff of performances. There's specialty emulators that aim for high accuracy such as Higan

As of a few years ago, it was the only emulator not bugging out on Yoshi's island. ZNES is notorious for inacuracy in general. SNES9x did a better job, but still off. Again, modern CPUs steamrolled it eventually, Higan was the go-to for the best experience, but the SNES emulators on say, a 3DS, would bug out on this game.

I don't have any horses in this race, I don't own any FPGA emulator nor any software handheld emulators. But having worked on VHDL FPGAs I understand the gist of the why they do it. Is it required to play retro games? No. Is it expensive? Yes. After that it's a choice.

Software emulators are dependent on OS changes, API changes, hardware changes. They patch something, break something. It's the typical cycle that devs are left with a clinging community to always support them even though the emulator is old as shit.

I think that's something FPGA is making it a lot more simple. If the core is good, that's pretty much it. There's no OS/Driver/API/hardware configs to debug. All mister FPGA cores could be ported to a bigger FPGA (Mars FPGA if it releases?) and remain the exact same performances, as long as the other FPGA is an upgrade in every senses. That's another thing for simplicity is how easy they are to have accurate results without tweaking. While software emulators are configured for general compatibility but then something is a bit off and you have to tweak it in the settings. Some want to pay extra money on that and I get it.

These guys who are good at VHDL probably don't even understand that if they focused their attention to aerospace industries they would be paid big sums of money compared to their emulator hobby as it's very tough to find good VHDL programmers nowadays, but aerospace industries use that heavily for how solid and reliable it is.
 
Last edited:

dcx4610

Member
What exactly is the benefit of a dedicated hardware FPGA over regular classic software emulators?
I like to call it "hardware simulation" vs. "software emulation".

Instead of software pretending to emulate hardware, an FPGA chip can be programmed to run the exact hardware specs of the system in question. So as long as the core is programmed correctly, it's 99.99999% like running on real hardware. From the music, to the timing, the game plays as if it was the original hardware with the added benefits of being over HDMI or even adding more features.

Unless I see FPGA, I'm not interested in playing older games any other way. FPGA + Flash Cart = perfection.
 

nkarafo

Member
Buggy Loop Buggy Loop

A few comments. Most of the stuff you wrote are basically one of my points on my first post. That, with a CPU you need more power to achieve the same level of accuracy. But you can achieve the same level of accuracy.

Not all software emulators have the same accuracy. It's a tradeoff of performances. There's specialty emulators that aim for high accuracy such as Higan
Yes, not all software emulators aim for cycle accuracy. Some do, some don't. I'm not sure what you are trying to say here though. It's not because you can't make a cycle accurate emulator if you want, it's because some want to make something different. Once again, it depends on the project/author/skills/knowledge. Not all FPGA cores are cycle accurate either so? Again, it depends on what the author wants to do or can do, i'm not sure why i have to repeat that so many times.

And it's not only Higan. There are plenty of other cycle accurate emulators for different systems.


As of a few years ago, it was the only emulator not bugging out on Yoshi's island. ZNES is notorious for inacuracy in general. SNES9x did a better job, but still off. Again, modern CPUs steamrolled it eventually, Higan was the go-to for the best experience, but the SNES emulators on say, a 3DS, would bug out on this game.
Sorry, but this is wrong. A few years ago? Zsnes has been obsolete since 2004. Back then SNES9X wasn't that great either but you could still play Yoshi's Island without issues on SnesGT on a Pentium 4.

A 3DS is not powerful enough for BSNES, sure. But that doesn't prove "FPGAs are more accurate" it proves that accurate emulators require more power. You don't expect the 3DS to offer accurate emulation for anything other than the NES or Game Boy maybe.


Software emulators are dependent on OS changes, API changes, hardware changes. They patch something, break something. It's the typical cycle that devs are left with a clinging community to always support them even though the emulator is old as shit.
SNES9X is not a cycle accurate emulator nor it tries to be. So i don't know why you are comparing that to FPGAs. Also, none of those things, the OS, drivers, etc, have anything to do with the accuracy of the emulator. They have to do with it's ability to run on the host machine.

Speaking about support and communities, did you know the Analogue NT people used BSNES as a reference to fix their own bugs?


There's so much focus on popular consoles like NES/SNES/SMS/Genesis, etc that it looks nearly perfect and I would say indiscernible from FPGA
These machines you mention are already covered by cycle accurate software emulators that are indistinguishable from real hardware. Some long before the FPGA cores showed up.

NES - Messen, PuNES
SNES - Higan/Bsnes
SMS - Ares
Genesis - BlastEM, Exodus


Then software emulation completely fails an amiga 500 while mister FPGA will cruise on that. But not popular so not talked about.
Sorry but the moment you use a Pi device in order to compare it to an FPGA... I can't take it seriously. The Pi is, literally, the worst platform to play emulators on accurately at full speed. It can't be done. Performance wise you are better of using a mid-range phone from 6 years ago. This is very well known so i'm not sure why you use it as an argument. You are only proving my point, that accurate emulation needs more CPU grunt compared to an FPGA. Not that it's impossible at all.


If the core is good, that's pretty much it.
And what being "good" depends on?

Yup. The author's work/code.

Just like software emulation then.


Instead of software pretending to emulate hardware, an FPGA chip can be programmed to run the exact hardware specs of the system in question.
FPGA is emulation too.



So as long as the core is programmed correctly, it's 99.99999% like running on real hardware. From the music, to the timing, the game plays as if it was the original hardware with the added benefits of being over HDMI or even adding more features.
Just like a software emulator then.
 
Last edited:
Since nobody answered my question yet (don't mean to repost; got lost in the mix) what is the ballpark for the minimum PC spec you would need to emulate anything MiSTer can run at full accuracy? Just curious.
 

nkarafo

Member
Since nobody answered my question yet (don't mean to repost; got lost in the mix) what is the ballpark for the minimum PC spec you would need to emulate anything MiSTer can run at full accuracy? Just curious.
The most demanding emulator that aims for accuracy right now (of the systems FPGA supports) is Ares for the N64. On my i5 4670 (a 10 year old CPU) i can play most games full speed at native resolution, except the most demanding ones, such as some Factor 5 games or Conker. Not sure what CPU i need to just meet the minimum required for those. Probably an 8th gen CPU? I can't be sure as i don't have access to better CPUs right now. You also need a GPU that can run Vulkan but even a lowly GTX 1060 should be more than enough even if you upscale.

Keep in mind the N64 FPGA is not complete and neither is as accurate as Ares currently.

For older systems i can run all other accurate emulators at full speed on my current i5 4670. I can even use CPU expensive features that reduce input lag at the same time. Not sure what the minimum requirements for each is though. But i think even a fast Core 2 Duo will handle anything up to Sega Genesis full speed. Not sure about Higan, i do remember it running full speed on the Core 2 Duo but that was an older version (which was already nearly 100% compatible/accurate).
 
Last edited:

NeoIkaruGAF

Gold Member
This discussion is going in circles. Every time VGEsoterica VGEsoterica releases a new Mister video, the whole charade starts a new loop.

I always think I want a Mister. Then I think about how much time I have for gaming these days, and how much of that is spent on playing retro games I’ve never played before, and how much of that is done on a CRT. And every time the math says, any emulator will serve me fine. I’d be better served with a good Android-based device right now, and even my 2014 tower PC can probably give me 99% of what I’d ever want from emulation.

If I had infinite free time I’d be all about playing a Mister connected to a CRT. But this very interesting piece of hardware really is for a specific audience, which I’m currently no part of.

Still, I love seeing the progress this project is doing.
 

nkarafo

Member
The argument that emulators can be just as accurate as FPGA is pretty asinine. Especially based off of a literal crazy post of a literal crazy person.
Speaking about asinine arguments, that very same crazy person you are accusing just so happens that he created an emulator that is just as accurate as any FPGA core ever was and more. One that was used as a reference to improve the Analogue NT among other things, imagine that 🤷‍♂️

Don't take my word for it. It's out there, it exists, anyone can test it if they want.
 
Last edited:
Top Bottom