MiSTer FPGA Forum

Hello MiSTer community,

In an attempt to stimulate some discussion around future FPGA platforms/devices that could be implemented as an extension / expansion of the MiSTer eco-system, I would like to ask for feedback on the idea of using the Efinix Trion T120 as a starting point for ideas, discussions and the modifications that would be needed with the MiSTer platform.

Full open transparency here, if the MiSTer user/developer community can decide (or at least have the desire to explore) a new/alternative FPGA based platform, then we could step forward to design a custom FPGA module to have as an alternative to the DE10-nano / or go down a different branch if a change to architecture makes more sense for everyone involved.
We have started on several investigations for FPGA based families for recreation of retro hardware platforms, along with concept ideas based around multiple/modular architectures to allow a fully scalable range of 'platforms' (Console / computer / handheld / Arcade etc.).

Very happy to also have the discussion about how best to help with a design – full open-source hardware etc. - we just want to see some community driven project focus on an alternative to the existing DE10-nano as it’s showing its age and cost are still rising significantly, it’s starting to feel like a dead-end. Working with Core dev’s on this would be ideal, but we are honestly just trying to work out if that’s what the community wants or not? - Not trying to step on any toes here, just offering to help as a way to have something new in the hardware pipe-line so the MiSTer project keep on growing for everyone.

A form-fit-and-function ‘replacement’ DE10-nano style module has a lot of obvious benefits, along with the fact that the TRION family of FPGA’s can scale from smaller devices up to significant sizes, may allow different entry price points for smaller cores - a single Arcade core for example on a smaller dedicated device.

I was not intending this initial discussion to be about what could come next with regards to core development (32bit platforms etc.) but more to consider how to lower the price point and start discussions around how a different FPGA family could be implemented as an alternative within and for the MiSTer project. But feel free to discuss whatever you like, especially if you are an FPGA dev for MiSTer and have some feedback on the Trion Devices / or wish to suggest another family for discussion / consideration.

A few obvious questions to start the discussion –
TRION FPGA’s don’t have an SOC (ARM/Linux) side to them, this could be addressed in a different way (Similar to the original MIST) or a different hardware, embedded firmware architecture could be developed if needed.
Removing the SOC aspect from the FPGA, does have benefits too, so that ARM/Linux I/O side could be replaced with a modular replacement where different microcontrollers could allow a much wider range of use. For example, a very low power ARM part for use in a handheld or a generic ‘Raspberry Pi’ type of module just to do the O/S – USB – File System – network stuff etc.

All FPGA’s are different and they are not easy to compare from a resources standpoint, we also have not yet fully dived into the Efinix FPGA tool-chain but we do see that the TRION FPGA’s are starting to be used for many Retro based projects, like accelerators, ODE’s and similar modular hardware replacements for classis computers & Arcades.

The Efinix Trion T120 evaluation board is already available as a starting point for core evaluation and to assist with working out what the specification of a ‘MiSTer 2.0’ platform could take, what interfaces need to be defined and how we solve the lack of SOC (ARM/Linux) issue.
DDR3 Memory, HDMI, Ethernet, Multiple Oscillators, SPI, High speed MIPI interface, 100 GPIO Pins for almost unlimited expansion options.
In the BGA 324pin FPGA form factor the Trion T120/85 and 55 could all be fitted to allow for different price points and core sizes. Or multiple / modular FPGA’s could be used together to dedicate different core functions (CPU/Graphics for example).

Why am I posting this?
We (Heber) designed the MiSTer Multisystem carrier board for the DE10-Nano, it was just a tiny step but people seem to very much enjoy the concept of a ‘MiSTer console’. It uses the DE10-nano and that has had us thinking about alternatives since before even doing the Multisystem.

It is a little odd to post this as a commercial company into a community based open-hardware development. If the general reply is, we don’t want any help, then that’s okay. We are enthusiasts ourselves and would like to see new platforms discussed. We are offering free help, not trying to railroad the project in any way.

In reality, a project like this takes a massive level of work, we could offer to do some/all of the modular hardware design but there is no sensible way we would start porting cores or write a custom Linux/firmware OS etc.
I also don’t expect anyone to want to build on a closed source hardware platform, the DE10-nano was convenient, but it’s now not viable to replicate, due to cost and component restrictions.

Finally we also don’t want to go ahead and make something that the existing community does not want, or have no interest in using/developing/evolving, so instead of development in secret, how about some open-innovation.
It just seems like a good time to kick off something new and if we can help, even if it’s with ideas then that’s something worth having a discussion about out in the open.

Best Regards,
Richard.

I've seen the idea of a alternative board or MiSTer 2 thrown around before on the FB group. A cheaper alternative to the DE-10 would be nice but we might end up in the same situation with a alternative board

I've also seen Sorg state on FB there isn't much further to go with FPGAs in general in terms of retrogaming for various reasons

We have also seen people planning to fork the project for other boards in past and not much has come of it and that leads down a very entertaining rabbit hole

Grumpy-Old-Gamer wrote: ↑Tue Jan 10, 2023 3:14 pm

I've seen the idea of a alternative board or MiSTer 2 thrown around before on the FB group. A cheaper alternative to the DE-10 would be nice but we might end up in the same situation with a alternative board

I've also seen Sorg state on FB there isn't much further to go with FPGAs in general in terms of retrogaming for various reasons

We have also seen people planning to fork the project for other boards in past and not much has come of it and that leads down a very entertaining rabbit hole

Thanks for the feedback -
Keep it modular, using more modern tech / FPGA's and it's less of a dead end and not reliant on a single device/family supplier.

There is still plenty of scope for FPGA based systems to be used in many different ways - dedicated systems, handheld, Arcade hardware replacement. The future does not always have to be the goal of next-gen hardware recreation - Sorg is correct with that for now, but the scope of using FPGA's to replicate instant-on hardware platforms is still open to a very wide range of systems / applications.

Certainly don't want to fork the project unless a significant majority of core/dev's want to try something new/different - that was the entire point of the proposal to explore the Trion family or anything else rather than just going ahead and doing it.

Oh granted but there is nothing really new in using FPGAs in Retrogaming or even add ons.

We have seen some great projects over the years like C-One, Minimig, MiST, Turbo Chameleon, Ultimate 64, 1541-U2, FPGA SID etc

We have seen lots of smaller multi system FPGAs too and analogue pocket now has MiSTer core ports

All we really need is Terasic to release its current tight control on DE-10 supply and surely there is a point where software emulation becomes a more viable option especially with price of the likes of the Xbox series S or even Steam deck for a handheld option ?

Forgive me but isn't the heart of the Mister framework basically linux running on ARM? You can "easily" port the cores to another FPGA, but the Mister itself is written for a CPU and a scaling chip. So without that its hardly Mister anymore?

As expensive as the DE-10 nano is getting, it still seems to benefit from subsidies such that are hard to compete with. Just how low can this Trion get?

Not exactly:
The framework is in the FPGA and communicates with MiSTer on the Arm side so the framework would have to be ported together with MiSTer. Not an easy job. There is no scaler chip, the scaler is in the FPGA.

As I understand it, you folks (Heber) can build most or all boards of the complexity class we're presently dealing with. It seems to me, then, that the primary question becomes this: is there another FPGA similar to or stronger than the Cyclone V that you can get cheaply, in volume? Are any of the FPGA makers out there willing to price their products to move?

If you can get a strong FPGA on the cheap, then there's something to talk about. Otherwise, everything else is kind of irrelevant; there's no point in trying to find an existing board or design a new one if the central engine is expensive or in short supply. It would be much ado about nothing.

The project may work better in a year or so. Once the chip squeeze has passed, maybe someone out there can decisively beat the Cyclone V. At that point, you folks might be in the exact right spot to design and launch something that's better than what we have now.

But until that Unicorn FPGA shows up, cheap and available in significant quantity, nothing seems likely to happen.

FoxbatStargazer wrote: ↑Tue Jan 10, 2023 4:28 pm

Forgive me but isn't the heart of the Mister framework basically linux running on ARM? You can "easily" port the cores to another FPGA, but the Mister itself is written for a CPU and a scaling chip. So without that its hardly Mister anymore?

As expensive as the DE-10 nano is getting, it still seems to benefit from subsidies such that are hard to compete with. Just how low can this Trion get?

Yes and no, the current MiSTer framework, main, and Linux are very specifically written for the Cyclone V. The current code is not portable to other FPGA platforms as the MiSTer Main applications maps to the memory addresses of the FPGA directly.

I've been tinkering with re-writing MiSTer Main to create something that could allow for porting parts of the ecosystem to other platforms. My main goal has been to stop poking the FPGA's memory addresses directly and lean more on the drivers provided by the Linux kernel. This would involve using device tree overlays in Linux to load cores and attach the appropriate drivers in the Kernel.

IMO the most important question to ask in any discussion of potential new hardware platforms is "what do we actually stand to gain from new hardware?"

If we imagine a dream world where we can get Stratix GX 10M boards for the cost of a DE10 Nano, how far up the hardware ladder do you realistically think that could let us go? When fifth generation consoles are proving to require months to years of work by skilled developers, and sixth generation hardware is generally an order of magnitude more complicated to reverse engineer, document, and reimplement, how much work would it take to develop a core for newer platforms? Dreamcast's SH4, the smallest of the sixth generation CPUs, has more than 10x the transistor count of the SH2s in the Saturn. The others are 30-110% larger than that. While transistor count doesn't directly relate to how hard it is to implement in FPGA it can provide a ballpark idea of the complexity of a given processor.

More importantly, how far up the hardware ladder is it even possible to get running at full speed? I'm not a FPGA expert, but my understanding is that going above a few hundred megahertz even on high end chips starts to require careful design which may not be possible when the goal is to reimplement an existing architecture, and the absolute hard limit is still well below 1GHz. Depending on where that line is drawn some of the sixth generation consoles might just not be possible to implement at full speed on a current FPGA at all. Xbox's 733MHz Celeron seems like it would pose a major challenge, likewise for PS2's 400MHz DDR Rambus memory.

If my understanding is correct, even with a top tier FPGA and a generous sponsor funding skilled developers to fill in the blanks the best case scenario would be one more generation of hardware support. On top of that it's a generation which is mostly well supported by software emulation, depends less on exact hardware timing, and benefits most widely from HLE "upgrades" like increased render resolution, texture/model replacement, etc.

Beyond the possibility of a few newer cores, what else would we gain? I guess maybe if we could get some HDMI 2.1 transmitters the scaler could be upgraded to do 4K or beyond, and with a good hard ARM core maybe there's enough headroom there to bring MT32-pi type stuff onboard, but both of those things only appeal to certain subsets of the userbase.

Now take all that, and remember we're still in dreamland imagining we have a FPGA 100x the size of what we're currently using that realistically costs more than most of our cars. Any product that's going to be viable to try to produce for mere mortals in the next decade isn't going to come anywhere close to that.

It just really doesn't seem like there's a lot to gain, while there would obviously be substantial costs in hardware design, at least some effort porting, and the effects of fracturing the community.

Regarding the specific board you proposed, honestly I'm not sure how it would be an upgrade at all. It looks to have a roughly equivalent FPGA while lacking the ARM, at a price that's higher than even scalper prices for a DE10-Nano. I can see the hypothetical appeal in a port to a hardware family that scales down nicely, but aside from handheld builds where size and power efficiency matter the product is going to need to be either substantially cheaper or substantially better to be worth the trouble.

I guess one reason outside of future different cores, etc. is this:

How long is the DE-10 Nano expected to be available for? If availability has a known or likely end-date then plans either need to be afoot to port to a new platform or the project will die when hardware runs out (and eventually existing boards will die too). As I understand it the reason for choosing this board was low cost; if it isn't low cost any more or isn't available then that changes things?

Like anything else, FPGA hardware will improve and cost will come down over time, so even if its just to run the currently supported cores, at some point there will be an attractive FPGA for less money or with better availability.

wolrah wrote: ↑Tue Jan 10, 2023 10:46 pm

Depending on where that line is drawn some of the sixth generation consoles might just not be possible to implement at full speed on a current FPGA at all. Xbox's 733MHz Celeron seems like it would pose a major challenge, likewise for PS2's 400MHz DDR Rambus memory.

I would be really, really surprised if we ever even saw an accurate Pentium 1 chip, never mind the P3-class Celeron. The out-of-order execution on those chips would be fiendishly difficult to get right. That crap is serious chip wizardry.

A reasonably accurate 80486 is probably about as far as FPGA reimplementations will ever go. (ie, a much better version of the AO486 we already have.) Going further is not in the realm of hobbyist endeavor anymore. The originals took huge teams and tens of millions of dollars to create; expecting amateurs to recreate a workalike for free is just not reasonable. It wouldn't be as pricey as the originals, since they'd have the working spec already and working chips they could examine, but it would still be a massive effort that would take a lot more money than small hobbyist communities are ever likely to generate.

I don't do any FPGA design work, it's well past my capabilities, but just eyeballing the project as a whole, it seems like an FPGA with about twice the elements of the Cyclone V would probably cover the great majority of what amateurs can actually accomplish. Example: an FPGA that big would likely allow the AO486 core to be fixed up properly. It probably wouldn't have enough room for an accurate Pentium, but that recreation is just not going to happen, no matter how big FPGAs get.

The question of a new mister came in the past. The general consensus seems to be that anything more powerful than mister is useless because beyond the 32 bit era consoles became too complex.
However if i trust what i red on this forum here and there, it seems there is a few doable things that are not possible in the mister but doable in the development perspective.
I think of the DS core that is not possible on the mister but it's very close, only 2d and not 3d. Or the N64 core that a guy is developing for many years with a more powerful fpga. There should be similar things in the Arcade world as well i suppose, or as it said in the last post in the computer world.

So the big limit in what fpga can do will surely be the N64, so maybe we should think of a mister 2 with that in mind. My two cents.

I don't trust multisystem at all to lead an initiative. Do they contribute monetarily or with open hardware designs back to MiSTer? I a big open source advocate and would hate this beautiful project because the users where mislead by a salesman trying to sell is wares. I am sure they have an ulterior motives.

Malor wrote: ↑Tue Jan 10, 2023 5:37 pm

As I understand it, you folks (Heber) can build most or all boards of the complexity class we're presently dealing with. It seems to me, then, that the primary question becomes this: is there another FPGA similar to or stronger than the Cyclone V that you can get cheaply, in volume? Are any of the FPGA makers out there willing to price their products to move?

If you can get a strong FPGA on the cheap, then there's something to talk about. Otherwise, everything else is kind of irrelevant; there's no point in trying to find an existing board or design a new one if the central engine is expensive or in short supply. It would be much ado about nothing.

The project may work better in a year or so. Once the chip squeeze has passed, maybe someone out there can decisively beat the Cyclone V. At that point, you folks might be in the exact right spot to design and launch something that's better than what we have now.

But until that Unicorn FPGA shows up, cheap and available in significant quantity, nothing seems likely to happen.

We currently have designs using almost all of the major FPGA manufacturers and plenty or others using old / obsolete and EOL stock. During the pandemic it was only the Old / Obsolete and EOL stock that we could actually buy, and manufacture. To this day the main FPGA manufacturers still have not restored supply of their current parts. Most offer a sample service on 14 weeks for a few parts that allows prototypes and development, then anything up to 60 weeks for some level of 'volume' production, and that can often be a lot less delivered than ordered. They are desperate for designers to keep on designing in their parts, but cant supply parts for production on timescales of less than 30/40/60 weeks.

I do hope Intel starts to restore supply of their FPGA's but it's not going to start happening until the end of this year or 2024 and prices are still very high indeed. Some are trickling out, the Intel supply to Terrasic must still be very limited as we have not seen any stock at distributors for over 12 months now.

Almost exactly the same thing happened to the ARM microcontroller market as almost every electronics design had some sort of ARM device used, the STM32 was and still is on serious allocation and 40+ weeks delivery time.

All the spare stock and grey market parts were sold out in 2020/21. It's a very long waiting game or re-design to parts that are available.

All that background was just to say that Efinix Trion FPGA's are available, in samples, as development boards and also as high volume production and are much lower cost devices. This is the closest we have to a 'unicorn FPGA family' right now, there really are not many other viable options.

Intel never took Altera internal after the buyout and the chips are still fabbed on TSMC 28nm, with TSMC increasing 28nm capacity and it being a older node you would have expected supplies to have improved by now. Though we don't know Intel's wafer deal or how much Terasic is controlling supply while enjoying the higher prices.

Terasic has always had DE-10s for direct sale even when their website states zero as long as you can afford the higher price point which is not helped by shipping price and import duty

Sorgelig the MiSTer creator on Facebook group explained in the past about using improved FPGA and an issue of the amount work to create more advanced hardware cores.

multisystem wrote: ↑Tue Jan 10, 2023 2:25 pm

More to consider how to lower the price point and start discussions around how a different FPGA family could be implemented as an alternative within and for the MiSTer project.

If that is really the case why give the thread a clickbait title MiSTer 2.0?

When what you mean is investigate an alternate hardware platform to the existing DE10-Nano.

annette wrote: ↑Wed Jan 11, 2023 11:10 am

Sorgelig the MiSTer creator on Facebook group explained in the past about using improved FPGA and an issue of the amount work to create more advanced hardware cores.

As I said in my original post -
"I was not intending this initial discussion to be about what could come next with regards to core development (32bit platforms etc.) but more to consider how to lower the price point and start discussions around how a different FPGA family could be implemented as an alternative within and for the MiSTer project. But feel free to discuss whatever you like, especially if you are an FPGA dev for MiSTer and have some feedback on the Trion Devices / or wish to suggest another family for discussion / consideration."

The point for discussion around the Trion FPGA family is as much to allow wider use capabilities and lower/different price points - like having a dedicated single use ARCADE core in a smaller FPGA, or have a different version of MiSTer 2.0 for only classic computers that has a lot of physical interfaces, expansion etc. Or even finally a dedicated console arrangement so that we can have real cartridge connections and other peripheral's.

The current level of 32bit cores are already pushing FPGA technology, so it's more appropriate to work backwards down from PSX, not try to reach for ever more complex cores as the FPGA technology and costs are not in-sync or appropriate for implementing much past PSX and Saturn. I fully agree with Sorg but I would also like to see a wider use of FPGA's to recreate more of the classic computer and console technology - as a bare minimum having multiple baseboards for dedicated systems that included cartridge connectors, original controller ports and other expansion would be quite viable with a family like the Trion as they have many more times the I/O available. People may not want to use MiSTer like that, but it would allow it to happen if there was an interest.

It is the modular aspect and possible expansion of the MiSTer platform, I was trying to highlight.

My focus was on breaking the lock on the DE10-nano and widening the MiSTer platform, so we could have more fun developing more hardware, more cores and hopefully lowering the price point and increasing functionality.

I never intended to be discussing DS / N64 / PS2 and more powerful cores. - that's not the focus or idea here.

Maybe asking the community if they are interested in some of the benefits a new FPGA family could bring (after many other difficulties are also solved) would be more useful to the conversation -

Assuming the 'MiSTer 2.0' could be made into a sensible range of 'FPGA modules' to allow for some flexibility in use and implementation of a wider range of systems / baseboards / I/O boards -

Are people interested in physical interfaces for cartridges / peripherals and maybe dedicated configurations designed to run one family of cores/games /programs etc. - For example a dedicated base board for GEN/MD that also has cartridge and the expansion port to connect a real Sega CD along with all the standard benefits of HDMI video out, analogue out, real controller ports.

What about a small baseboard for a viable handheld based around FPGA emulation (yes, similar to the pocket, but more open/community driven from every aspect).

Maybe a dedicated computer baseboard that has many or the classic computer interfaces of the 80's & 90's - Parallel port, serial, disk drives, cartridge, floppy, keyboard - in a similar way to the MEGA65 (for example). - I'm highly interested in this aspect from a hardware preservation standpoint, but maybe that's not what other MiSTer users want?

If none of that is interesting and all people want is more powerful recreations of more modern systems, then this idea is not going to get off the ground.

annette wrote: ↑Wed Jan 11, 2023 11:27 am

multisystem wrote: ↑Tue Jan 10, 2023 2:25 pm

More to consider how to lower the price point and start discussions around how a different FPGA family could be implemented as an alternative within and for the MiSTer project.

If that is really the case why give the thread a clickbait title MiSTer 2.0?

When what you mean is investigate an alternate hardware platform to the existing DE10-Nano.

No, not really, it is significantly expanding on the scope and use of MiSTer. A DE10-nano 'replacement' is the bare minimum that's useful for current users. I would imaging even the DE10-nano replacement would ned to be backwards compatible, but also have multiple extra connectors for expanded baseboards, then if people want more capably systems, new base-board designs can be made in a modular system.

What's interesting is that users may see '2.0' as a step in FPGA power/capability, but I am trying to showcase the idea that a wider base of hardware expansion is possible in a modular way. That's still a very valid direction for 'MiSTer 2.0' if people are interested, which is the entire point of this thread.

Low cost targeted FPGA boards are a sensible idea. See the Sipeed Tang Primer 20K FPGA a NES for 42$

https://github.com/nand2mario/nestang

The Sipeed Tang Nano FPGA Development Board at 17$ is also interesting

multisystem wrote: ↑Wed Jan 11, 2023 11:58 am

My focus was on breaking the lock on the DE10-nano and widening the MiSTer platform, so we could have more fun developing more hardware, more cores and hopefully lowering the price point and increasing functionality.

Are people interested in physical interfaces for cartridges / peripherals

Maybe a dedicated computer baseboard that has many or the classic computer interfaces of the 80's & 90's - Parallel port, serial, disk drives, cartridge, floppy, keyboard - in a similar way to the MEGA65 (for example). - I'm highly interested in this aspect from a hardware preservation standpoint, but maybe that's not what other MiSTer users want?

The cores available on MiSTer are not tied to the project we already seen ported to other projects like Neptuno, Mega 65, Analogue pocket and a lot have come from other past projects C-One, Minimig, MiST etc.

I would say people interested in using original carts, floppies , CDs already have options on the market for a lot of the systems, these closed products can offer a more accurate experience compared to MiSTer, the Ultimate 64 for example.

Though we do see support for original floppies on the C64 core on MiSTer due to basic nature of the IEC interface
There is also devices like PolyMega for CD based systems

When you have full emulated cartridge, floppy, HD, optical drive support is there any need to support the physical media ? The archiving has mostly been done for the majority of systems by other people over the years

annette wrote: ↑Wed Jan 11, 2023 12:50 pm

Low cost targeted FPGA boards are a sensible idea. See the Sipeed Tang Primer 20K FPGA a NES for 42$

https://github.com/nand2mario/nestang

The Sipeed Tang Nano FPGA Development Board at 17$ is also interesting

tang.jpg

Very neat, thanks for the link.

If we had a range of FPGA / modules that could operate under the MiSTer project eco-system that would be nice to see and allow scope for lower cost dedicated platforms.

multisystem wrote: ↑Tue Jan 10, 2023 2:25 pm

...as it’s showing its age and cost are still rising significantly, it’s starting to feel like a dead-end.

It's too presumptuous to call it a dead-end at this point. Terasic still makes the boards, if you order from them directly you get them in a couple weeks (when it's not christmas time), the price went up, but it's still cheaper than getting a retrotink 5x + rom carts + a couple properly functioning consoles.

Regarding the Efinix Trion series of chips as a comparison however:

1. The T120 series has 5.4Mb of block RAM whereas the Cyclone V in the DE10-Nano has 6.191Mb. Not a huge difference, but there are a tiny handful of cores that wouldn't work as a result without significant modification or tradeoffs in the framework.

2. Intel's Cyclone V in the DE10-Nano has a different way of producing what counts as a logic element than the Efinix Trion T120. Each logic cell in the Trion T120 is a 4-input LUT with one register. The Cyclone V is more accurately counted up in number of adaptive logic modules, which are 8-input fracturable LUTs with four registers each, and 2 full adders. You can also configure 1/4 of the Cyclone V's ALM's as MLAB (a type of true dual port BRAM available on the Cyclone V). I feel like the Cyclone V is more feature rich and flexible when it comes to how the synthesis tools will effectively use what's available.

3. The Trion T120's got more multipliers, but the Cyclone V has the option to configure them in 9x9, 18x18, and 27x27 natively, whereas the T120 can only use them as 18x18 natively. This limitation will probably impact synthesis optimization significantly as things that can't be done natively in the multipliers might end up synthesizing into logic instead, which is costly. It really depends, a deep comparison would have be made.

4. Trion T120 is on a 40nm process node whereas the Cyclone V is on a 28nm process node, and that will impact some power efficiency and heat dispersion under load. Not a major issue but a comparison nonetheless.

5. The Cyclone V has more capabilities in terms of the PLL (clocks) configuration from what I can tell.

I have been looking at the Efinix chips for awhile and they are interesting, but I think the more interesting ones are the new Titanium series which are on a 16nm process node, are more powerful, have a ton more block RAM (up to 72.09Mb!!) and still cost efficient. I think the professional hardware builder cost-cutting mindset got the better of you there for a bit since the Trion's are so cheap

I think a more appropriate Efinix chip to look at as a project you could work on would be the Efinix Titanium Ti180 or Ti240. However...

https://www.digikey.com/en/products/det ... K/16731861

That dev kit is $850, and only 63 in stock at digikey. I also don't see the higher end Titanium chips for sale, only the lower end ones that have similar specs to the highest end Trion's.

We'll see what happens I guess.

annette wrote: ↑Wed Jan 11, 2023 12:50 pm

Low cost targeted FPGA boards are a sensible idea. See the Sipeed Tang Primer 20K FPGA a NES for 42$

https://github.com/nand2mario/nestang

The Sipeed Tang Nano FPGA Development Board at 17$ is also interesting

tang.jpg

I see this as the logical "budget option" for some users who just want to emulate one system in FPGA in a simple way, little one-off projects like this are a great alternative if all you care about is playing NES games. MiSTer is more than that though, so not directly comparable of course.

aberu wrote: ↑Wed Jan 11, 2023 2:34 pm

multisystem wrote: ↑Tue Jan 10, 2023 2:25 pm

...as it’s showing its age and cost are still rising significantly, it’s starting to feel like a dead-end.

It's too presumptuous to call it a dead-end at this point. Terasic still makes the boards, if you order from them directly you get them in a couple weeks (when it's not christmas time), the price went up, but it's still cheaper than getting a retrotink 5x + rom carts + a couple properly functioning consoles.

Just wanted to add I received mine direct from Terasic within a few days after ordering last week!

I had this exact conversation with myself a while back and basically came to the conclusion that it wouldn't be feasible to produce a successor using different hardware while keeping the project aligned with the current codebase and third party peripherals without creating some serious headaches. It would be way easier starting a new system from scratch and just porting existing cores to it. There could be some serious room for improvement in hardware allowing for some currently unviable cores to be made while also making it lean and very cost favourable.

Personally I went down the route of making a JAMMA adapter for the DE10, purely focussing on the arcade cores and their playability in arcade cabinets with CRTs. Decided to put a pause on the project though when the availability of DE10s for £110 vapourised due to the current silicon s-show. I have working PCB prototypes and was going to release the design files as open source, just really needed to do some finishing and cleanup. Plus I wasn't the biggest fan of the frontend so I went about coding a new one before realising it would be better to contribute work to a branch or a new platform entirely.

Having been a commercial embedded electronics engineer for a decade I've had some contacts in the FPGA industry tell me that things are not looking good for silicon availability. Combine that with chip life cycles and I'm amazed anything is getting made at the moment. At work we even joked about buying the worlds supply of DE10 nano boards to use in one of the projects I had been tasked to work on replacing the low availability Xillinx FPGA that it was using. That wouldn't have got very far!

The Efinix Trion looks interesting and I may investigate it myself. Cores that I'de be interested in making an FPGA platform for with suitable bandwidth, memory and logic capacity would likely be the Cave CV1K hardware, Sega Model 1 & 2 and Namco (Super) System 22 / 23. I'de also have a hand in coding such cores as I have previous 3D rendering engine experience. I just need to set aside some time once again.

I think the "budget" or "MiSTer Lite" route would be the place to start. I'm sure there'd be plenty of interest in a board that could just run Golden Age arcade games, for example—perfect for a dedicated cabinet setup.

One important thing to keep in mind early on is that HDMI is not a licence-free standard. Terasic are obviously handling that with the DE10-Nano, but any community-developed alternative would need to eschew HDMI in favour of DVI or DisplayPort to avoid running afoul of the law, and that could put a lot of noses out of joint.

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.

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.