MiSTer FPGA Forum

meepowl wrote: ↑Thu Jan 19, 2023 10:52 pm

I'm new to the forum and my first reaction to reading this discussion was to think how refreshing it is to see something relatively controversial being discussed almost entirely without rancour. It's a pleasant change from some other retro communities out there.

There's clearly a bit of scepticism from some about this being raised by the multisystem team. I take the point but having bought one for one of my MiSTer setups I really don't think it's the sort of thing you'd get into selling if you weren't committed to MiSTer or were principally seeking to make money. I think the obvious way to short circuit this issue though would be to make it an article of faith that any new iteration of MiSTer was built around a commercially available development board (as was proposed). That way no entity can have undue influence on the project.

That said, I'd have thought the most serious, and at times controversial, downside of the MiSTer implementation is the limitation of attaching SDRAM via GPIO? Do we not run into the same issue with the proposed Trion board, or any other dev board?

The original post said that there's work to do to dive into the toolchain. I'd see that as as big a priority as finding a board with the right performance:price profile. I just spent two days solid getting Quartus 17 running on Ubuntu. It's not the worst thing I've ever used for development but it's also a world away from the sorts of tools I use at work or for other personal projects. It would be nice to see a lower barrier to getting involved in MiSTer development (the development work itself should be the hard bit). This is especially important if there ever were to be a push beyond fourth-gen: we'll need more people working in parallel.

Finally, purely personally, I like the idea of an offshoot of the project using simpler dev boards to implement one system at a time. At the other end of the spectrum, I also like the idea of something which would allow more hardware tinkering, whether that's cartridges or peripherals or any other sort of addon. Not because it's needed, just because it's fun.

Thanks so much for the feedback. It's always difficult to try reaching out to any community, I'll try to give a little more background just in case it makes a difference to how we may be perceived and what our intentions are.

We (Heber) are not a large company, we are a team of 16, but have a 39 year history of developing and manufacturing electronics for gaming markets worldwide, using FPGA based technology in many of our designs over the years. We also have a significant number of people in the company who like and enjoy gaming & retro technology. We (Heber) are part of the Multisystem team that includes support from Neil (RMCRetro) and The Retro Group.

We are always open with our plans, we initially developed the Multisystem I/O board and for the last 2 years have manufactured these to a high quality using branded components and ISO9001:2015 manufacturing methods directly in the UK. I personally designed the open-source 3D printed Multisystem enclosure and we print these here too. These products are never going to be the lowest possible cost, but we are proud of quality, support and all the team here and most importantly we all enjoy doing it. This year (2023) we will get to the point where the initial investment in the design, development, test tooling and set-up costs will just start to break-even. Last year we employed another engineer to work on further product developments, some of these are Multisystem and Retro Gaming. We are investing in people, ideas and Yes, we would very much like to design and manufacture products, electronics and retro gaming tech for an interested customer base/market, This is exactly what we do, it's no secret.

We are willing to invest what could be a further £150k or more in new projects/developments this year, it would be nice if we can continue to recover costs and breaking even at some years in the future. Not looking for any guarantee, it's always a risk doing any new development, we just don't want to go the opposite direction to the community, rather work together and end up with something really neat that people want.

We have also helped enable two museums to start and grow over the last 3 years, RMCRetro's classic computer and gaming history museum is now in the same building as us and we have all jointly enabled a classic Arcade museum (Arcade Archive) on the bottom floor of our building.

I/We will always be open and try to answer any questions, concerns or try to collaborate on ideas to the best of our abilities, both commercial and technical that anyone has. We are not asking for anything, just trying to slowly build some trust in a project we really enjoy being part of. That's it, that's the plan.

Richard.

ericgus09 wrote: ↑Fri Jan 20, 2023 4:35 am

These talks come up about once a year and the all end the same way.. once people realize the actual "cost" to make a Mister 2.0 .. the project usually dies and fades away.. I suspect the same will happen here.

We really understand the cost and it'll take time and effort, but I do hope we can build a momentum of desire and interest in doing something.

How does anything get moving in an 'adhocracy' community - interesting ideas and enough support to make it happen.

I am all about new hardware. I got my first DE-10 in 2018 and have enjoyed the project. I support as many developers as I can on Patreon as well.

Here are a few shortcomings I personally have about the current MiSTer.

1). The menu system. Just to be clear, this one I have no issues with it. I like simple and fast (i have also done UNIX/Linux administration for over 2 decades, so i live on the cli). However My wife and when i have friends over and the kids, they struggle a bit. If i pop on the raspberry pi with retroarch, they can do everything. Perhaps new hardware could help with that (But i would agree it would have to be able to boot real fast like the current MiSTer).

2). No USB 3 or ability to connect a drive internally besides USB2. SD cards suck, 10 years of using raspberry pi's has taught me that. I minimize writes and still have odd failures. Having a faster way to connect drives would be great, especially with PSX and Saturn. Most peoiple will say just use network sharing, but there are silly things like the refusal to add NFS to the kernel. I do not understand this on any level. I dislike samba shares. We can argue that is a personal issue. This is not hardware, but the refusal to add it when issues are open on github is strange.

3). Limitations. On the PSX core you can have accurate sound with 2 RAM modules or analog out. This is not a huge deal to me and I cannot tell the difference but another board may be able to overcome limitations like this (and perhaps other unknown limitations). The option to allow this would be swell. Ensuring the new board has a lot of future proofing built in (like the awesome useless port on the bottom of the NES, haha). Please do not get hung up with the 2 RAM sticks point for the PSX core here, that was just an example of a possible limitation.

4). Formfactor. I have 2 MiSTer's, one on a multisystem, the other on an IronClad board. This makes them a lot easier to find a decent case. The issue i have with the misterstack (which i used to roll) is it is really light and easy to knock over and with ports on all sides can be a bit strange to get set up and look good in the living room. Having an industry standard single board with a mATX or other standard form factor would be sweet and make cases more accessible without needing a second board to plug the de-10 into.

5). Ability to use real carts. Yes this is controversial, but some of us like this. I will concede this is probably not even feasible with what Multisystem is proposing without it being an abomination like the retron consoles.

6). Open source, open standards. Everything open source (we wish). The build process, even using a RiscV chip instead of arm perhaps. This is a pipe dream, but as much as possible. I do understand a company has to be able to make R&D back though, so maybe opensource designs after X years or after making so much money back.

I will admit these are probably personal or impossible and can all be argued. However that is some of what I would like to see in a custom FPGA based board for retro gaming above what the MiSTer does.

Also in the back of my mind I think about all the cores being ported to the pocket. I do have an Analogue NT and noir and pocket and would be untruthful if I said i didn't like them. But Analogue as a company, does everyone really trust them, how about their communications skills with the community? The pocket, aside from not running all the cores (of course they have a rudimentary PSX core running, but who knows to what extent), with its dock is somewhat close to replacing the MiSTer. To me it lacks good controller support and video processing and I do not like having to dock something (my switch has not left its dock in over 2 years, i could live with a consolidated switch ).

Could a company like Multisystem who is actively trying to work with the community to see what they want and if something better could be built be worse? I personally can imagine Analogue, with the popularity of the pocket and openFPGA building the exact system Multisystem is talking about, cornering the market and not listening to the community. They already have the cores in place and more are being ported by the day.

Anyway there are my thoughts, maybe useless, maybe not. Just stuff to think about. In the meantime I am still happy with my MiSTer and all the peripherals I have with it and will still support as much as i can on Patreon. And if it does not change I will still be happy.

The way I see it, the only major drawback of the DE-10 Nano as a MiSTer target is its HDMI 1.x port. A board with a HDMI port capable of 4K would be great for enabling scaler output to 4K, which seems to me that is the only aspect where MiSTer feels lacking.

(With that said I'm not sure about how much extra load it would put on the FPGA to do 4K scaling, and if that would still be feasable? That might still be the bottleneck even with a more powerful HDMI port.)

When looking at any new platform you need to look at features, price and availability / support cycle. Right now the DE-10 nano is lacking some extra GPIO but has a very cheap price and a long support cycle and is generally available(current world situation withstanding). So What I see any alternative so far may have the extra GPIO but is way more expensive and will be a gamble support wise as who knows when it will be discontinued. The proposed board here does not check enough of the boxes and doesn't realistically give a decent amount of extra features for the cost and could put the project in jeopardy if it is discontinued. Right now the DE-10 nano works and it works well I would rather wait for further advancements in the fpga market to go to a further board. What I would rather see is side projects using MiSTer cores to port to cheaper FPGAs for a clone system projects and taking the cores and possibly making hardware asics based on the cores to replace crappy chinese asics like those NES or genesis on a chip that are just awful. I would rather see the community branch out and use this platform to create new things than a drop in replacement for the DE-10 nano because realistically it won't be able to hold more advanced cores than the Playstation or Saturn we have now. We are at the limit tech wise.

So my advice it to build something around the Speed Tang go cheaper and with better designs and make portables, or uniquely designed games systems or clone consoles and push the FPGA systems in a different direction using established cores on the MiSTer as your starting point. When the tech improves then we can go and talk about a MiSTer 2.0. Best of wishes on this journey I just respectfully don't think this is the time yet to go forward with this proposal.

multisystem wrote: ↑Fri Jan 20, 2023 9:58 am

ericgus09 wrote: ↑Fri Jan 20, 2023 4:35 am

These talks come up about once a year and the all end the same way.. once people realize the actual "cost" to make a Mister 2.0 .. the project usually dies and fades away.. I suspect the same will happen here.

How does anything get moving in an 'adhocracy' community - interesting ideas and enough support to make it happen.

To answer that its easy, someone actually "does it" .. that usually ends any "discussion" ..

The DE10-Nano is currently $225 Base price, is this conversation worth re-looking at? Because that's incredibly high compared to what it was originally, and it's not even an upgrade. Should we look for something better at the same price point, or something equivalent at a lower one?

GinGear wrote: ↑Mon May 01, 2023 12:02 am

The DE10-Nano is currently $225 Base price, is this conversation worth re-looking at? Because that's incredibly high compared to what it was originally, and it's not even an upgrade. Should we look for something better at the same price point, or something equivalent at a lower one?

Is there actually anything better than a DE10-Nano at the same price point of $225? Anything more than $250 (10% price increase) is not at the same price point unless it includes some nifty upgrades.

While the DE10-Nano is higher than it was, the current price of $225+shipping is still a great price when compared to the expenses of original hardware. That depends on the system, but these days the prices are growing to (1) acquire decent condition original hardware and controllers, (2) get that original hardware modded for modern use (power, video, recapping, etc.), and (3) purchase a flash cart. A single system can easily exceed the $225 of the DE10-Nano.

i think the community has to be driven with a very tangible incentive:

porting to the Analogue Pocket? obviously desirable given the hardware

porting to another random "alternative" FPGA ? if it's HALF the price of a DE10 and can do 16-bits generation i can see that happening, otherwise, the whole porting effort sounds kinda futile ?

porting to a way more capable FPGA to guarantee N64 accuracy and perhaps some other systems like MMX-class PC etc ? mayyyybe that could be a motivation, and honestly THAT's what i could call a MiSTer 2.0, but again, not with a 500+€$ board either

Nickie wrote: ↑Mon May 01, 2023 12:51 pm

porting to the Analogue Pocket? obviously desirable given the hardware

It's far less capable hardware than the current MiSTer platform so it feels like a step backward, however it's handheld and cheaper so I can understand why some people get excited I guess

Nickie wrote: ↑Mon May 01, 2023 12:51 pm

porting to another random "alternative" FPGA ? if it's HALF the price of a DE10 and can do 16-bits generation i can see that happening, otherwise, the whole porting effort sounds kinda futile ?

The SNES core in many ways would be more demanding than most cores, so at minimum it should be a roughly equivalent FPGA to the current one used in the DE10-Nano.

Nickie wrote: ↑Mon May 01, 2023 12:51 pm

porting to a way more capable FPGA to guarantee N64 accuracy and perhaps some other systems like MMX-class PC etc ? mayyyybe that could be a motivation, and honestly THAT's what i could call a MiSTer 2.0, but again, not with a 500+€$ board either

Personally, $500~ is a good target price for an upgraded system. Plenty of people have spent that much on their whole MiSTer setup after add-on boards, case, etc...

akator wrote: ↑Mon May 01, 2023 11:55 am

GinGear wrote: ↑Mon May 01, 2023 12:02 am

The DE10-Nano is currently $225 Base price, is this conversation worth re-looking at? Because that's incredibly high compared to what it was originally, and it's not even an upgrade. Should we look for something better at the same price point, or something equivalent at a lower one?

Is there actually anything better than a DE10-Nano at the same price point of $225? Anything more than $250 (10% price increase) is not at the same price point unless it includes some nifty upgrades.

While the DE10-Nano is higher than it was, the current price of $225+shipping is still a great price when compared to the expenses of original hardware. That depends on the system, but these days the prices are growing to (1) acquire decent condition original hardware and controllers, (2) get that original hardware modded for modern use (power, video, recapping, etc.), and (3) purchase a flash cart. A single system can easily exceed the $225 of the DE10-Nano.

Better in terms of the SoC-FPGA's capabilities? Sure, the Kria KV260 for $249. However, there's no way to hook up SDRAM to that so a ton of cores wouldn't work if they were ported over. It also doesn't have VGA out, etc... There's a lot of disadvantages to it because it's more "modern" in many ways and designed specifically for machine vision development. Finally, it's software is annoying as shit to work with (I've messed with a little, it's very broken). The linux images that come with it barely work because the cryptography module on it has a near-broken implementation.

aberu wrote: ↑Mon May 01, 2023 4:08 pm

however it's handheld and cheaper so I can understand why some people get excited I guess

it's literally the one FPGA handheld so yes, excited people would be

aberu wrote: ↑Mon May 01, 2023 4:08 pm

The SNES core in many ways would be more demanding than most cores, so at minimum it should be a roughly equivalent FPGA to the current one used in the DE10-Nano.

the DE10 handles Saturn and perhaps soon a good chunk of N64 so how does SNES look like such a demanding core to the DE10?

aberu wrote: ↑Mon May 01, 2023 4:08 pm

Personally, $500~ is a good target price for an upgraded system. Plenty of people have spent that much on their whole MiSTer setup after add-on boards, case, etc...

people have spent that much with accessories, that supposedly you'd also have to pay EXTRA just the same with the new platform unless there's one somewhere that doesn't need RAM over the GPIO, has enough USB ports and other video-outputtable ports ... so i don't think you can compare one option with all accessories with a hypothetical barebones one

For what it's worth and speaking as someone with virtually zero knowledge of the actual layers here, I think you should work from the software side before you worry about which exact chip to pivot to. A great deal of work has accreted for the DE10 and the Cyclone V in general, and even if you have a very viable chip you are not going to see a 100% transfer-rate for let's say, every arcade core. IMO, the discussion should surround the concept of making the framework more portable, because every incrementation that is made equates to man-hours being lost to time, even if it's some back-alley game only one guy cared enough about to write a core for. In my idealist view, the MiSTer of the future is not a board or a project of hardware constraints but a documentation project like MAME, which is easily portable to both the devices of today and of the future. In a hypothetical world where the newest hotness in FPGA comes out every 5 years, having a massive schism at each interval just feels colossally wasteful.

To people who have worked on the project, I'd ask, what is the bottleneck for porting to another chip as I am not super familiar with the Quartus toolsets vs Trion Efinity vs Xilinx whatever. I think a port to a Xilinx board which is designed to make the existing base more flexy is a much more interesting and forward-looking proposition than jumping wholesale to a specific product, and gives the project a solid footing in both of the largest silicon vendors. How tightly integrated to the DE10 is the Mister Framework as it exists today, and what tools could be used to serve as an abstraction layer to make it easier to do an in-place substitute? Could we cut down the framework to it's core and provide BSPs for individual development boards, so that in the future porting to a new board is a matter of configuration, more than of total overhaul? I think that is a much more interesting proposition.

Again, this is speaking as someone with zero experience writing HDL so I may be making horrible assumptions. Please educate me if so! I'd love to learn more about this, but life gets in the way of time spent on learning exercises.

I've long thought that the best option would be developing a FPGA hat for a Raspberry Pi. A daughterboard with FPGA and dedicated SDRAM that could run everything MiSTer can run plus a SNAC port to ensure zero input lag. The RPi would serve as the "HPS" for the system, doing all the OS and data handling tasks that don't need to run on the FPGA, and outputting low-latency video. It would use the Raspberry Pi's often-overlooked Secondary Memory Interface for maximum throughput between the FPGA and the main board. Presumably, support for this FPGA board would be added to existing RPi retro emulator platforms, so the FPGA will emulate whatever platforms it can support and the rest will run in software emulation.

Such a board could be priced around what the De10-nano is now, and could be attached to folks' existing Raspberry Pis.

i don't think video would be as low-latency if the RPI has to fetch it back through the memory interface and buffer it into it's own display system tho, that sounds like a bit of a thing to overcome
plus i suppose that hat should have a VGA-type analog output of it's own too

One dude is working on getting mister and mister cores on other FPGA´s/systems like risc instead of arm.
And it will mostly be converted automatically cutting much time for the dev´s.
See Bobs interview with him about it here, it was really interesting!

The horse is not dead yet, it is still being ridden. Unfortunately, the area is very fast-moving, so it makes little sense to plan "too early". It should also not be forgotten that most of them have invested a lot in their Mister. You don't like to throw that away just like that. Personally, I am of the opinion that a further development later should be a pure FPGA module with Ram as a basis, for which you could design any carrier boards.
Similar to the CM4 from the Pi. Here, however, I would rather rely on a "big" manufacturer than on a small one. I just wonder who is going to do this to themselves to manage this. Designing the hardware for it, on the other hand, is easy and will be finished at some point. But managing such a project, supervising it, etc. is actually not feasible for a company, because then you have to calculate quite differently. Therefore, we should all be happy as long as we still have the "Mister" and its creator.

I think one thing people often forget is the escalating complexity of writing cores for more advanced systems. There's a point of rapidly diminishing returns (and I'd argue strongly that's the N64) that beyond which you're way beyond the bounds of what can be accomplished by a single (or even handful) of developers realistically. Pentium, for example, carries an AWESOME degree of complexity.

rhester72 wrote: ↑Thu May 04, 2023 11:33 pm

I think one thing people often forget is the escalating complexity of writing cores for more advanced systems. There's a point of rapidly diminishing returns (and I'd argue strongly that's the N64) that beyond which you're way beyond the bounds of what can be accomplished by a single (or even handful) of developers realistically. Pentium, for example, carries an AWESOME degree of complexity.

I agree .. and the only thing i've heard people banter about and I don't even know if its a viable "thing" yet, is using decapped chips and AI to help figure out what the logic on the chip is doing .. and to me this still seems a bit off .. I think some people are fiddling with this but I don't think its wholly viable at this point, but thats my best guess at how you would go beyond the current limit of where we are rapidly approaching.

Jotego seems to play with a thing called Data-Storm.

I think the real successor will be when Intel announces a true successor to the Cyclone V SoC line, which they haven't yet. The cyclone series is all about low cost/low power applications. The Cyclone 10 line doesn't have an SoC-FPGA yet, it's possible they may make one with a combined Cyclone 10 GX architecture with an SoC with probably 200k+ LE's. But the time this will happen is a big question mark because the market for something like that is pretty limited. Low-cost+low-power is a different market than something that needs 200k+ LE's.

Nickie wrote: ↑Mon May 01, 2023 6:52 pm

the DE10 handles Saturn and perhaps soon a good chunk of N64 so how does SNES look like such a demanding core to the DE10?

The amount of logic it consumes due to the number of enhancement chips is massive. Ao486 is also huge. the PSX core is now the largest core, but it also has savestates and tons of features that pushes the limits which the SNES core cannot have due to platform limitations.

Nickie wrote: ↑Mon May 01, 2023 6:52 pm

aberu wrote: ↑Mon May 01, 2023 4:08 pm

Personally, $500~ is a good target price for an upgraded system. Plenty of people have spent that much on their whole MiSTer setup after add-on boards, case, etc...

people have spent that much with accessories, that supposedly you'd also have to pay EXTRA just the same with the new platform unless there's one somewhere that doesn't need RAM over the GPIO, has enough USB ports and other video-outputtable ports ... so i don't think you can compare one option with all accessories with a hypothetical barebones one

Yeah, part of the reason why an upgraded platform can potentially bring down the costs is a single board with all of these things, rather than needing 3-4 separate PCB's manufactured as add-ons. But the reason why a custom solution will cost extra from someone like the OP is because they don't have electronics company partners likely giving them discounts the way Terasic gets with their educational board whose primary purpose is as a promotional tool for Intel/Quartus in the educational and hobbyist realm.

Sipeed just came out with a new Tang Nano 20K FPGA at https://www.aliexpress.com/item/1005005581148230.html looks pretty interesting. 20,000 logic elements puts it closer to the level of the original MiST project. Still, a tiny MiST with built in HDMI and USB-C for power would be enough for a lot of people's portable needs.

Looks like NES is already working fairly well:
https://twitter.com/nand2mario/status/1 ... 7003006978

Genesis was mentioned in the comments as a possibility. Will be interesting to see how this progresses.

aberu wrote: ↑Fri May 05, 2023 2:17 pm

I think the real successor will be when Intel announces a true successor to the Cyclone V SoC line, which they haven't yet. The cyclone series is all about low cost/low power applications. The Cyclone 10 line doesn't have an SoC-FPGA yet, it's possible they may make one with a combined Cyclone 10 GX architecture with an SoC with probably 200k+ LE's. But the time this will happen is a big question mark because the market for something like that is pretty limited. Low-cost+low-power is a different market than something that needs 200k+ LE's.

That is the big issue both Intel and now AMD with Xilinx seem to focusing more on the pro and industrial markets with their FPGAs and even combining them with their other IPs. AMD like Intel after the Altera buyout has raised pricing by 25%
Intel has not really done much with Altera since the buy out.

I cant see Terasic dropping the price of the DE-10 back to where it was two years ago as they have found a nice niche for a product line that seemed just to gather dust on college storeroom shelves, At least supply has started to improve with Digikey having multiple restocks in the past month

The hobby market for FPGAs was always a promotional tool for Intel, it's never been a serious investment with any serious profit margins. It's to get people used to using Intel's products so when they get into the professional space it's what they are used to.

The various chips in the newer gen stuff from what we have adds more to the time and effort needed to get a decent core in place. Those projects will definitely take longer to develop, and a considerable time investment for core devs.

A few of us are scraping the bottom regarding really obscure computer cores, and just finding technical docs on those is difficult itself. The cores get so far along that parts start to work, then running something on them highlights more work that can be cumbersome to debug and resolve. I can imagine the larger dev projects hitting more walls along the way too.

One other thing is that MiSTer has come a long way with addons, case configurations etc that make the whole experience enjoyable. Such things will all need to be created too, as I dont expect they will work on any replacement board for the existing MiSTer we have.

aberu wrote: ↑Fri May 26, 2023 1:32 pm

The hobby market for FPGAs was always a promotional tool for Intel, it's never been a serious investment with any serious profit margins. It's to get people used to using Intel's products so when they get into the professional space it's what they are used to.

We have had FPGA based retrogaming devices for 20 years though look at the C-One, Minimig, Turbo Chameleon etc this was before the Intel buyout, Cyclone V and DE-10 but this is when pricing was more attractive.

Intel's college program really didn't do great things for the DE-10 as when supply started to dry up MiSTer hardware sellers were buying cases of unused DE-10s from colleges that was sitting on shelves gathering dust.

Terasic has now found a nice outlet for the DE-10 and even though we are seeing supply improving (Digikey has had more stock again in the past couple of days) I doubt we will see pricing returning to what it was sadly

DE-10 is very likely the cheapest logic for $ out there since it's subsidized.

I'd be interested in repurposing a Kria K26 or Ultra96 w/ some kind of low-latency RAM expansion as those are low-ish cost and have 2x the logic resources of the DE-10.

Well sipeed is apparently targeting a mister replacement with the gowin arora v 138k + Andes risc v cores

Their dev tooling is kind of meh, and Altera LEs aren't directly comparable to gowin LES, but it does look quite interesting for half the price and 138k le.

Would need a carrier with HDMI TX (12.5gbps serdes so 4k120 works lol) and low latency ram unless there's a direct to logic bus vs axi

But that's relatively easy to do: maybe HYPERRAM/rldram and ti tdp1204 on a USB c pd carrier with breakouts for controller gpio.

Promising and much cheaper than kria. Supposed to release in August.

What I have found interesting is supply for the DE-10 has been improving, Mouser for example have 342 in stock but sales are slowing down, I noticed this with the last couple of batches Digikey had recently too

Mouser has sold just 15 worldwide in four days, so they are not flying off the shelves anymore

I just noticed that the Efinix Ti 180 FPGA is finally available on Digikey for $55 in small quantities (80 units).

I think there's a place for a System-on-Module (SoM) ecosystem like the Raspberry Pi 4 CM. It may be possible to match the capability of the DE-10 with a cheaper board that's also portable. I think there's a lot of interest in a cheaper and/or portable system, especially one that's open-source. It doesn't have to be about making something more powerful. As others have pointed out, that's a dead-end road for many reasons.

I imagine a hand-held with a Raspberry Pi Zero 2 W, a SoM socket, and a dock connector. If you're just after light emulation, no expensive SoM is needed, but if you want the full MiSTer experience, you plug in a module with an FPGA with its own SRAM. If you want to play a multi-player game or use an arcade box, plug it into a dock.

I know a lot of this isn't the most original plan in the world, but it seems like it might be worth the effort.

Edit: I know Efinix offers a cheaper Trion (non-Ti) family of FPGAs that goes up to 120K LEs for as little as $27 on Digikey, however; I have my doubts whether a design that takes up around 80% of a 110K LE Cyclone V could fit in a 120K LE Trion because the feature set and device design are so different.

The feature overlap between the Cyclone V and the Trion Titanium families is much greater, so it might be possible to make due with the 120K LE version of the Trion Titanium, but that chip is only marginally cheaper ($49 vs $55) than the 180K LE version, so I figure better safe than sorry.