If I'm not mistaken, there have been persistent rumors of some sort of emulation box that will run DOSBox (presumably under a specialized Linux distribution). That will be more than enough to captivate a large segment of the market that might be interested in a retro DOS PC, and it would be considerably cheaper than something more directly hardware-based.

Another, much smaller but undoubtedly more valuable segment would be industrial users seeking to replace their aging parts, but anything priced for that market would probably be inaccessible to the typical hobbyist.

You could look at Vortex86 as a basis, perhaps.

I think the biggest problem with a retro DOS PC is "what era do you pick"? There are plenty of people who have interest in ONLY 8088/CGA gaming. Plenty of others who don't care about that at all, but are all in to late 386/VGA gaming. Then there are late Pentium/early 3D accelerator gamers.

C64, ZX Spectrum, even Apple II and similar era are all "single-spec" systems. Yeah, some might have more memory, or even slightly better graphics (late Apple II vs early Apple II) but the CPU is the same, and by and large games will work on *ANY* model (RAM being the big limiting factor on some, like Apple II.) My 1989 Apple II copy of "Where in History is Carmen Sandiego" will run on a circa-1978 Apple II Plus, but my MS-DOS copy will not run on a circa-1981 IBM PC.

And the reverse is true as well, Maniac Mansion will run on a fully-upgraded Apple IIgs, but many early PC games are unplayably fast on even a 386.

That said, I wouldn't mind seeing some sort of modern custom-made "vintage DOS game machine" that had new enough parts to run the last generation of DOS game, but was speed-configurable to run even original DOS games. Bonus points for having a VGA connector *AND* a composite-CGA-out plug. Double bonus points for some hardware hackery to allow for CGA composite over the VGA!

The system should also have a modern instant-boot SSD, along with at least a 3.5" drive, bonus points for a 5.25" drive as well. Jackpot bonus points for a USB port up front that allows a flash drive with floppy disk images to be mounted as if they were floppy disks.

Anonymous Freak wrote:

I think the biggest problem with a retro DOS PC is "what era do you pick"? There are plenty of people who have interest in ONLY […]
Show full quote

I think the biggest problem with a retro DOS PC is "what era do you pick"? There are plenty of people who have interest in ONLY 8088/CGA gaming. Plenty of others who don't care about that at all, but are all in to late 386/VGA gaming. Then there are late Pentium/early 3D accelerator gamers.

C64, ZX Spectrum, even Apple II and similar era are all "single-spec" systems. Yeah, some might have more memory, or even slightly better graphics (late Apple II vs early Apple II) but the CPU is the same, and by and large games will work on *ANY* model (RAM being the big limiting factor on some, like Apple II.) My 1989 Apple II copy of "Where in History is Carmen Sandiego" will run on a circa-1978 Apple II Plus, but my MS-DOS copy will not run on a circa-1981 IBM PC.

And the reverse is true as well, Maniac Mansion will run on a fully-upgraded Apple IIgs, but many early PC games are unplayably fast on even a 386.

That said, I wouldn't mind seeing some sort of modern custom-made "vintage DOS game machine" that had new enough parts to run the last generation of DOS game, but was speed-configurable to run even original DOS games. Bonus points for having a VGA connector *AND* a composite-CGA-out plug. Double bonus points for some hardware hackery to allow for CGA composite over the VGA!

The system should also have a modern instant-boot SSD, along with at least a 3.5" drive, bonus points for a 5.25" drive as well. Jackpot bonus points for a USB port up front that allows a flash drive with floppy disk images to be mounted as if they were floppy disks.

This. There are just too many PC eras to cover. You'd need to decide on a specific era and hardware spec, or else do everything in an FPGA and let the user choose their own soft core. How are you planning on getting software on it? What would you use for video output? If you want a turnkey product, similar to the C64 in a joystick that Jeri Ellsworth came up with, then you need an OS and software. Forget Windows or anything with copyright or IP issues. Think more like FreeDOS and GEM. Then what about games? If you want to have several games, even at GOG prices, you're talking about $6 each, unless you are expecting the user to supply their own software. An FPGA product with FreeDOS and GEM and maybe 5 games is going to run at least $150. Your target market is either going to already be running things in DOS box on a modern system or they will already have built their own retro system out of used parts. So of the already small market for retro PCs, I see you selling maybe 100 units. As someone else already mentioned, if you want to make money you'd be better off selling to industrial customers. But then you price yourself so high that the hobbyist would never buy your product.

As much as I appreciate the spirit of your idea and have even thought of doing something similar myself, I think as a business it is a non-starter.

Yup. This is something you would have to do because you enjoy it. The retro PC market just isn't large enough.

We could use a ATX compatible motherboard with some sort of large scale adjustable CPU, and fully functional ISA slots, that will be something for a start. Next level will be to develop an universal video card. Then a sound card. That wil be it. CPU instructions are the same, just speed. And you can run DOS with DDR4 RAM also 😉

I have large collection of floppies and hard drives and ide dvd-rw-s, but it think i will not miss them, if i cand run all it form a SD card or USB stick.

With so few players in the x86 CPU market, I'm not sure what you propose using. Vortex86 is one of the few which may, possibly be not so heavily cache dependent like almost any other current production CPU.

I want real old hardware, the one I grew up with, tinkered with and what I couldn't afford back then so I can tinker with it now... For the rest there is already DosBox 😉

Look at ZX-UNO, it's a FPGA board that was thought as a mini ZX Spectrum. They have some interesting ideas about hardware and software:

- Do you need more than one hardware spec? It allows you to use up to 9 cores (although the first one must be the ZX Spectrum core). Note that a core may define ANY hardware, as consoles like Sega Master System or NES.

- What about the BIOS? It comes shipped with OpenSE, a public domain ROM for Spectrum. The owner is free to load any other ROMs, even copyrighted (note: ZX Spectrum has not a BIOS, but a ROM that includes BASIC).

- What about the software? It comes with some public domain games, but it has a nice SD slot. You may load your own software using real tapes or copying files into an SD card.

So, it can emulate ANY system (if it can be loaded into the FPGA) and doesn't need to pay royalties. It can't be used for most 16 bit systems (FPGA too small), but you can get some ideas.

With enough RAM and an FPGA big enough you could make XT+CGA, AT+EGA, 386+VGA and 486+SVGA cores, so you'll cover most eras. Also, you could make Amiga, ST and Mac cores...

There's a BIG difference in the FPGA firmware chops required to create a core for the ZX spectrum vs a 1990 era PC. Even creating a CGA/EGA/VGA video core would be a massive job. The timing requirements and logic complexity is just a whole 'nother level.

You'd need an ATX motherboard with ISA, PCI and AGP, full DOS and win98 support and a fully unlocked single core CPU that scales from 286 to pentium 4 performance, from software or BIOS. It's totally doable. I guess the most compatible chipsets would be either the i865 + PCI to ISA bridge or a VIA P4/K8 chipset that naively supports ISA. The CPU could be a modified netburst P4 or Pentium M with unlocked multiplier witch allows for L1 and L2 caches to be fully disabled and would run on slow FSB as well (preferably 33, 66 and 100MHz). If AMD decided to make a retro CPU a die-shrinked Athlon 64 or Athlon XP unlocked single core would do - but I guess the simplest way to do it would be to use an old chipset and CPU combo - say modified versions of the VIA CX266 chipset + a VIA CPU soldered to the board.

I don't really think there's a market for it tough. Personally I prefer authentic hardware from back in the day.

Very small market. Good luck.

thank <imaginary person> there's no market. I prefer buying boards with cpus for 10eur than putting 100s for those old pieces. My NES/SNES Collection is already expensive enough 🤣

gdjacobs wrote:

There's a BIG difference in the FPGA firmware chops required to create a core for the ZX spectrum vs a 1990 era PC. Even creating a CGA/EGA/VGA video core would be a massive job. The timing requirements and logic complexity is just a whole 'nother level.

There already are FPGA boards/cores of Amiga and Atari ST, and I guess both are more complex than the usual XT (and some ATs).

The main ideas exposed on my post were:
- That the FPGA approach may be possible. I didn't said that it would be easy.
- Your computer don't need to use copyrighted firmware/software.

About the market... it won't sold well. Their main market would be nostalgic people, but if it is expensive most people will continue using Dosbox.

My take: almost anyone who would want a retro PC would also want to source their own parts and build it themselves. That's part of the experience.

Anonymous Freak wrote:

There are plenty of people who have interest in ONLY 8088/CGA gaming.

I have a hard time believing that. I mean, there are barely any examples of such games on GOG.com. (Of course, one could argue that's because such games are so old it might be too difficult to track down whoever holds the rights. But then, if there was any kind of money to be made, that would not necessarily be an obstacle.)

I have built my retro PC around that exact target, 8088/CGA.

There's a TON of old games from that time period, CRPGs, graphic adventures, platformers, and the like.

I think making a hardware based recreation of a retro-pc-clone-something-machine is too big a mouthfull.
Amiga, Atari and C64 are easy, compared to something like a pc clone. Too many standards in excistance.

It has to be a narrow focus point, else go the software way.
Raspberry PI or other small computers are a good starting point.
Take the software suite "Retro-PI", and you have a nice starting point.

https://retropie.org.uk/

Systems, supported in Retro-PI:
https://retropie.org.uk/documentation/systems-in-retropie/

If this is not a desired solution, then go for real hardware, like a SS7 system.

As others have pointed out the market is small a niche. The other problem is that even though the market does exist it is mostly made up of exclusively DIYers. Kinda like trying to sell fully built muscle cars to guys that build their own muscle cars. You might sell some...but not many. I am willing to bet that most of us would rather spend more money on parts that we got to choose ourselves and put it together even if it cost more money then buying a full system that some other professional has certified as functional. If we did mess with full systems it's generally someone elses junnk that they threw out.

Basically there are plenty of people who are into this. But, most of the time we are looking to get buy on the cheap and are only willing to pay premiums for a small select amount of hard to find pieces.

The one market I do see is reproduction or retro compatible components such as the dreamblaster S1/X1 and that Gravis project that someone is working on. I don't think full systems will sell much unfortunately.