Advanced cores for MiSTer future FPGA hardware

[seastalker]

Many of us know there are FPGAs more advanced than the DE10 NANO, but prohibitively expensive... for now. The latter is most important to keep in mind with this post.

A hypothetical future I see once most of the wanted cores that fit on a DE10 Nano are done:

Developers will be faster than the hardware prices become reasonable: I can see the possibility that oft discussed projects like a N64 core or Dreamcast core will BEGIN life knowing it may/will not fit the DE10-Nano. Yet, if developed early, they may get streamlined to fit a FPGA board say two generations from now instead of a "Mister 4". If the goal is PRESERVATION, I can see cores getting developed even if not available to run on a Mister or the follow up to Mister.

If compared to film, this may be akin to a film scanned into 2k, BEFORE 2k or 4k BluRay devices are released.

[aberu]

N64 and DS have already had some preliminary exploration and work done on them by FPGAzumSpass. There's someone I've talked to who has worked on PS2 FPGA emulation at very preliminary steps, but he plans on just doing hybrid emulation with a PCIe FPGA card and a computer, mostly to offload the 128-bit calculations and other stuff that software emulation can't do accurately on even the fastest processors at normal speeds.

But Devs generally want to build things that people will be able and want to use, so the motivation probably doesn't hit you so hard if there's no platform and no hardware for people.

[Chilli_Vibes]

How is it "Preservation", when it comes to N64, Dreamcast, PS2, etc... ?
There are millions of them out there on eBay, second-hand shops, etc... It's not like it's hard to get one right now.
What I personally think is more important is the hundreds of machines released from the 70s, to the 80s, and 90s, that still have either no emulation, or no FPGA core on any FPGA device. These machines will eventually be lost to time, either sent to the scrap heap, lost in someone's attic, or sitting on a collector's glass shelf, in their silly glass cabinet, for only them to look at.

[Bas]

There used to be millions of Commodore Amigas out there for next to nothing. They're getting more expensive. Any device that's out of production will be a candidate for preservation efforts at some point.

[MostroW]

How is it "Preservation", when it comes to N64, Dreamcast, PS2, etc... ?
There are millions of them out there on eBay, second-hand shops, etc... It's not like it's hard to get one right now.
What I personally think is more important is the hundreds of machines released from the 70s, to the 80s, and 90s, that still have either no emulation, or no FPGA core on any FPGA device. These machines will eventually be lost to time, either sent to the scrap heap, lost in someone's attic, or sitting on a collector's glass shelf, in their silly glass cabinet, for only them to look at.

There will be a time when these stop functioning and will be less common to acquire.
Also when there's a trend going on where people are looking for this hardware, there's always enough parasites willing to speculate and capitalize on that trend.

Also this is not something we'll be seeing in the near future, the hardware required to get these projects going will be a pricetag above what most people are willing to pay for, unlike the fanatics or (extreme) hobbyists.

This is something for the future, just like when people said that Dreamcast and Playstation 2 emulation was unfeasable because pc's would never be fast enough to run these games on emulators.

[dmckean]

Price depends on what we're talking about. The board being targeted for the current N64 core in development is a $479 Multimedia dev board. Robert has said to replicate the 3D hardware of the Nintendo DS, even the largest $4000 development board isn't feasible.

[FPGA64]

Almost every old 80's 90's Computers have been recapped now. Sadly some never get done and when they stop functioning they are disposed of. The same will happen with the N64's

[Chilli_Vibes]

There used to be millions of Commodore Amigas out there for next to nothing. They're getting more expensive. Any device that's out of production will be a candidate for preservation efforts at some point.

Oh, I agree. BUT, to use a C64 today isn't that easy, especially if you don't have any modern way of loading stuff, or displaying it. I have two C64 machines, a C64, and a Breadbin, and while I also have a Commodore 1701 monitor, I do think these are a pain in the arse to setup and use. Whereas a Mister is just switch on and use.
The N64, Dreamcast, and newer systems, are still usable today, even on modern TVs. They don't have to deal with tape media, or floppy media. Switch on, play. They're also cheap, by retrro standards, and the only way of playing the real thing properly - emulation isn't quite there for N64, or Dreamcast.

There will be a time when these stop functioning and will be less common to acquire.

PS2 sold close to 158 million units, world-wide. There won't be a shortage for years. There will be, at some point, a failure of optical media - rot, mold, failing plastics, and optical drive laser failures - but these are still years away. And any newer than PS2, and we're going from preservation, and stepping into piracy problems. My own opinion is that we should concentrate on systems that definitely ARE going the way of the dodo, especially 70s, and 80s systems that hardly sold anywhere, and now fetch hundreds/thousands on eBay. If we don't get these systems documented, and put into emulation, and FPGA cores, then they could be lost forever. N64 isn't important, now. As is Dreamcast, they are for future Mister hardware that capable of doing them.

[aberu]

How is it "Preservation", when it comes to N64, Dreamcast, PS2, etc... ?
There are millions of them out there on eBay, second-hand shops, etc... It's not like it's hard to get one right now.
What I personally think is more important is the hundreds of machines released from the 70s, to the 80s, and 90s, that still have either no emulation, or no FPGA core on any FPGA device. These machines will eventually be lost to time, either sent to the scrap heap, lost in someone's attic, or sitting on a collector's glass shelf, in their silly glass cabinet, for only them to look at.

They won't be functioning and readily available in perpetuity. Your objection seems kind of like a fallacy of relative privation to me. "Why preserve the N64 when the RDI Halcyon isn't preserved yet?" has an obvious answer, because developers typically like to work on things that people will actually use and benefit from.

[Chilli_Vibes]

Well, an FPGA core of these systems is going to happen eventually, that I can agree on.
There is no objection, this is a discussion forum, is it not ? I just think there are far more important systems to preserve than the N64. And besides, the N64 onwards is mute, anyway, because these systems are not going to fit on the DE-10.

[Malor]

Well, an FPGA core of these systems is going to happen eventually, that I can agree on.
There is no objection, this is a discussion forum, is it not ? I just think there are far more important systems to preserve than the N64. And besides, the N64 onwards is mute, anyway, because these systems are not going to fit on the DE-10.

Expecting hobbyists to ever duplicate the Cell processor in FPGA format is, I think, wishful thinking written large.

In real life, the first CPUs were things that individual people could do, but then later generations were funded by the huge sales of the earlier chips. Companies could devote large teams to chip projects. The Cell processor project took more than four years of effort, cost four hundred million dollars, and the chip ended up with 250 million transistors.

Hobbyists are not going to be recreating those in their basement. FPGA re-implementations will take oodles of money and time. It seems unlikely to me that FPGA emulation of a PS2 will ever happen. It's hellishly difficult to do even in software.

[pgimeno]

As for the Dreamcast, I imagine that 200 MHz is out of reach for the MiSTer, but is it just a matter of MHz or is there something else?

[MostroW]

Expecting hobbyists to ever duplicate the Cell processor in FPGA format is, I think, wishful thinking written large.

In real life, the first CPUs were things that individual people could do, but then later generations were funded by the huge sales of the earlier chips. Companies could devote large teams to chip projects. The Cell processor project took more than four years of effort, cost four hundred million dollars, and the chip ended up with 250 million transistors.

Hobbyists are not going to be recreating those in their basement. FPGA re-implementations will take oodles of money and time. It seems unlikely to me that FPGA emulation of a PS2 will ever happen. It's hellishly difficult to do even in software.

The "Cell processor" was PS3, the PS2 had the "Emotion Engine" and the first models also had a full PS1 onboard.
I Wouldn't rule out that these are impossible tasks, highly daunting but with the correct teams i bet they could be done, just like software emulation.
The biggest hurdle i'm guessing is finding documentation / schematics on how these things are built / function?

[Chilli_Vibes]

Expecting hobbyists to ever duplicate the Cell processor in FPGA format is, I think, wishful thinking written large.

Why are you quoting me in your reply ?
I am not the one advocating for a PS2 core. Secondly, PS2 CPU was MIPS III R5900, so not quite sure what you're on about with cell - cell was PS3.

[H6rdc0re]

Problem with more advanced consoles is their dramatic increase in complexity. Just compare transistor counts of PS1 vs PS2 vs PS3. The PS1 is about 2 million transistors while the PS2 is 53.5 million and the PS3 is about 550 million. The jump from something like a N64 to a Dreamcast is huge. Going from a PS1 to a N64 is already pretty big.

I think we should just be glad with what we got already and everything else like Saturn, Jaguar, 3DO and some extra arcade cores are just a bonus. For more complex cores on better FPGA boards I think cores will either take a very long time to develop or developers need some kind of AI assistance.

[aberu]

I think someone meant to quote me. Yes, PS2 has been somewhat worked on in the preliminary phases, and then I think this person dropped off on working on it because of the cost-prohibitiveness.

https://www.youtube.com/watch?v=6QbRuhe7WMQ

[seastalker]

I am hoping this stays a friendly, light-hearted discussion. It was intended as highly speculative and or wishful thinking of a hypothetical future of FPGA progression. I have no intention of creating undo stress or pressure on developers either.

I am sure the majority of us here are incredibly thankful to every one of the core creators, and many of us are Patreon subscribers to show our support.
Likely, most of us know that N64 and Dreamcast (or beyond) won't fit on the DE-10, hence this discussion of a future even decades from now. Maybe future developers will be the grandchildren of current core developers, or new individuals that are drawn to this unique niche and time in history.

There were discussions during 1980s computer days that "no one would ever need THAT much RAM" or CPU power, etc. I never dreamed that one day we'd be printing plastic, let alone in 3d... anything can happen. I don''t want to join any debate on which cores SHOULD take priority over another. I prefer the idea that individuals will be drawn to individual passion projects in their own time. The fact that someone even tried to start a PS2 core for a platform that doesn't exist yet shows a lot of promise towards these ideas.

[Duffygag]

What's the point of this thread ?

[MostroW]

What's the point of this thread ?

Speculation with the occasional high horse sentiment and bashing of course dear sir!

[held]

This is just speculation but I think some of this stuff can be automated using AI in the future.
If IA is ment to replace humans in other processes why not here?

[aberu]

I personally think AI is vastly over-hyped given my experience with it in Industry 4.0 applications. It has potential, but it's a lot further away than those trying to sell you the product are willing to admit.

[redsteakraw]

preservation wise we have ODEs, and softmods, N64 and DS is basically done waiting a more powerful FPGA that is reasonably priced. Problem is that things won't settle for a couple years and that doesn't even factor in potential economic collapse and supply line collapses so best enjoy the PSX and Saturn cores we have as the world collapses around us. Cheers!

[Malor]

I am hoping this stays a friendly, light-hearted discussion. It was intended as highly speculative and or wishful thinking of a hypothetical future of FPGA progression. I have no intention of creating undo stress or pressure on developers either.

I am sure the majority of us here are incredibly thankful to every one of the core creators, and many of us are Patreon subscribers to show our support.
Likely, most of us know that N64 and Dreamcast (or beyond) won't fit on the DE-10, hence this discussion of a future even decades from now. Maybe future developers will be the grandchildren of current core developers, or new individuals that are drawn to this unique niche and time in history.

There were discussions during 1980s computer days that "no one would ever need THAT much RAM" or CPU power, etc. I never dreamed that one day we'd be printing plastic, let alone in 3d... anything can happen. I don''t want to join any debate on which cores SHOULD take priority over another. I prefer the idea that individuals will be drawn to individual passion projects in their own time. The fact that someone even tried to start a PS2 core for a platform that doesn't exist yet shows a lot of promise towards these ideas.

All I'm trying to say is that the low-hanging fruit has mostly been picked. Going much further forward in time will require enormous effort. Chip complexity just exploded. You may well see the PS2 started over and over, but actually finishing it will probably take a substantial team and a lot of time, probably multiple years, since it's just hobbyists.

In terms of creation date of the original systems, it's probably not going to advance much past where it is right now, no matter how big the FPGA boards get. Rather, you'll see more breadth, probably arcade games that used custom chips small enough to be amenable to re-creation, but which nobody's gotten to yet.

edit: in other words, if I haven't been clear enough, the limit on FPGA emulation is probably going to be people, not the hardware.

[Grumpy-Old-Gamer]

Sorg stated on the FB group. You won't see systems from year 2000+ on any FPGA

fROxgx6.png (512.04 KiB) Viewed 7561 times

[Mr. Encyclopedia]

I agree that FPGA isn't a good solution for consoles 6th gen and beyond. At that point, hardware and software gets sophisticated enough that the benefits of FPGA are outweighed by the limitations, as Sorg stated above. It's rarely cycle accuracy that's the goal for software emulation of these systems, rather removing inaccuracies in the emulator or overcoming hardware limitations of the machine running the emulator. A future MiSTer that can play these later consoles will probably be some kind of open hardware SBC with a powerful mobile chipset that can be targetted by emulator writers for newer consoles paired with an FPGA for older consoles.

[pgimeno]

Sorg stated on the FB group. You won't see systems from year 2000+ on any FPGA

The good news is that many of these systems are dedicated to emulating older systems that are already supported by MiSTer

[seastalker]

The last several responses are exactly the kind of discussion I was hoping for. Thank you! For me personally, I'm less likely to play past the mid-90s games as the disc based games can take up months to complete. Sometimes I make exceptions (Resident Evil 4, Mario Kart), but I only dabble in Dreamcast and OG Xbox due to the Sega connection. I'd be fine with emulation of the disc-based systems after that but thinking beyond my use case, hopefully that hardware gets preserved in the best way by some future generation of enthusiasts.

I agree that our current Mister platform could hit a stride in more rare and obscure arcade machines once the main systems like 3DO and CD-i come to fruition. Sorg's ideas are also interesting to me that even IF a future 'stronger' FPGA was miraculously affordable and the stars aligned, it may not even be the best tool for the job for preserving such 2000 and beyond systems.

[Mr. Encyclopedia]

It's certainly possible that future FPGAs may be fast enough to match the processors of these newer old systems, but it's also true that speed has never been a goal for FPGA architecture. They've always been meant to be customizable ASICs that can perform one job at exactly the speed required, which is rarely more than a few hundred megahertz.

Because of MiSTer and Analogue and Polymega people treat FPGAs as this holy grail of retro gaming but as they exist today they really, truly don't make sense for the faster CPU/GPU architectures with orders of magnitude more transistors that computers and consoles took up after the turn of the century. A software emulator tailored to run on specific hardware would be more than capable of preserving these consoles at a MiSTer-like fidelity. We're even seeing how this is true now with platforms like RetroPie able to squeeze the maximum amount of performance out of well-understood hardware. A better chipset, similar to today's high-end phones, and well-written emulators could easily bring us to the mid 2010s if our goal is to have a single open-hardware solution for preserving legacy software.

[Newsdee]

There's a threshold where it doesn't make any difference whether software or hardware emulation is used. That's where I expect to see the natural limit of FPGAs - not because they can't do it, but because there won't be further developer and user interest.

Not sure where the threshold is exactly but I'd guess the sixth generation of TV consoles (PS2 onwards) could fall in this category already.
At the very least I wouldn't be surprised if a hybrid approach is taken for them (as was mentioned earlier in this thread).

As 3D games became the norm, most code stopped being optimized for a specific platform. Some game designers used interpreters that could be rewritten for another platform to get good performance (e.g. GOOL used for Crash Bandicoot or GOAL for Jax & Dexter). GPUs relied on libraries for programmers to use them (e.g. GlIDE for 3DFX cards). On top of that, from the sixth gen consoles had an OS, and they started to use common standards converging to computer formats e.g. USB controllers. All that means that the natural FPGA advantage of being "bare metal" becomes more and more marginal.

It's not a bad thing. Architecture of consoles has converged into X86_64 meaning they are closer and closer to a powerful gaming PC.
You would want a powerful GPU in order to upscale a lot of those later 3D games into 4K and beyond, add Ray Tracing, etc.

Mind you, there is still plenty to cover in TV consoles up to the 5th gen (and portables up to the 6th gen), so there will be lots to do for FPGA developers and users that want to play with older systems. It's great already that MiSTer gives us alternatives to give a second life to old controllers and older TVs, while bringing a lot of QoL modern comforts.