VOGONS


First post, by commodore256

User metadata
Rank Newbie
Rank
Newbie

I don't know where to put this, but I figured this community would be interested. Since the patents on the Pentium MMX have expired, would anybody be interested a Single Board Computer like the Raspberry Pi, but uses an IP Core ASIC designed from Dosbox/Bochs algorithms? I'd get it just for the novelty if it was under $60 and maybe some industrial clients would be interested though I think they would want ISA slots. ASICs maybe expensive to make for new 14nm stuff, but 95nm is pretty cheap now. If the IP Core was there, it would cost almost nothing to make the ASICs, but there would be other stuff like the various components and assembly and licensing of newer ports like an SD card slot or HDMI.

Reply 1 of 13, by keenmaster486

User metadata
Rank l33t
Rank
l33t

I tried this idea out on the community before too: Open-Source Retro x86 PC: Feasiblity report?
(I think there's another thread somewhere too)

Basically the consensus is: while many of us would buy it, we're pretty much the only ones and we're not enough to make it economically feasible.

It'd have to be a small run like with the Dreamblaster or something. But this is far more complicated than the Dreamblaster.

World's foremost 486 enjoyer.

Reply 3 of 13, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

I still think you should look into existing cores if you're interested. Vortex86, for instance. That way you just have to spin a motherboard, not a full CPU and chipset.

Also, there might be niche industrial applications.

All hail the Great Capacitor Brand Finder

Reply 5 of 13, by Jorpho

User metadata
Rank l33t++
Rank
l33t++
commodore256 wrote:

I think for something as old as 180nm, you could break even from selling a couple thousand units.

If it was that easy to do, someone would have done it already. There's enough money to be made for those industrial applications.

commodore256 wrote:

DOS word processors work entirely adequately with DOSBox (or vDosPlus) and you will have a hard time convincing someone to shell out extra money to do the exact same thing on custom hardware.

Reply 7 of 13, by vladstamate

User metadata
Rank Oldbie
Rank
Oldbie

The problem with the old thread (and this one as well) is that we do not have a clear target. What I mean is people mentioned doing an XT board but 3 posts down someone says 386/486. Now this thread mentions Pentium MMX. There are very large differences (both in implementation and usage) between those systems. If the window of what we want keeps moving it is hard to identify what components people want and what is the feasibility of building such device. If all we say is "I want is something to play Wing Commander" that is not enough description to build anything.

For example:

4Mhz vs 266Mhz
ISA only vs VLB vs PCI
no FPU vs FPU vs MMX

In my opinion building something around a 486 platform would give enough breath to support all the DOS era games plus some win9x ones.

YouTube channel: https://www.youtube.com/channel/UC7HbC_nq8t1S9l7qGYL0mTA
Collection: http://www.digiloguemuseum.com/index.html
Emulator: https://sites.google.com/site/capex86/
Raytracer: https://sites.google.com/site/opaqueraytracer/

Reply 8 of 13, by keenmaster486

User metadata
Rank l33t
Rank
l33t
vladstamate wrote:

In my opinion building something around a 486 platform would give enough breath to support all the DOS era games plus some win9x ones.

This is what I've always thought. There is a pretty complete project out there to replicate the 486 on a FPGA. It even has a rudimentary Sound Blaster 2.0 built in, and VGA graphics iirc.

World's foremost 486 enjoyer.

Reply 9 of 13, by kpuchatek

User metadata
Rank Newbie
Rank
Newbie

New Pentium class PC working with DOS would be nice. But SW licensing can be killer. Without BIOS, DOS such machine is dead. Success depends on features and end user functionality.

Reply 10 of 13, by Jorpho

User metadata
Rank l33t++
Rank
l33t++
kpuchatek wrote:

New Pentium class PC working with DOS would be nice. But SW licensing can be killer. Without BIOS, DOS such machine is dead. Success depends on features and end user functionality.

There are a couple of open-source BIOS ROMs out there which are used by Bochs and PCem, if I'm not mistaken. (I have no idea how easy it would be to adapt them to real hardware, though.)

Reply 11 of 13, by Kisai

User metadata
Rank Member
Rank
Member
commodore256 wrote:

I don't know where to put this, but I figured this community would be interested. Since the patents on the Pentium MMX have expired, would anybody be interested a Single Board Computer like the Raspberry Pi, but uses an IP Core ASIC designed from Dosbox/Bochs algorithms? I'd get it just for the novelty if it was under $60 and maybe some industrial clients would be interested though I think they would want ISA slots. ASICs maybe expensive to make for new 14nm stuff, but 95nm is pretty cheap now. If the IP Core was there, it would cost almost nothing to make the ASICs, but there would be other stuff like the various components and assembly and licensing of newer ports like an SD card slot or HDMI.

I think right now, the best option is a FPGA for a 8088/80286/80386 + Sound Blaster/Pro/16 + CGA/EGA/VGA perfect-pixel doubling to HDMI. Because there are so many variations in x86 cores, soundblaster clones and video boards, an upgradable FPGA is more viable. That said, if the goal was a clock-adjustable Pentium+MMX level ASIC that can be paired with maybe a cheaper FPGA that the video/sound can be reprogrammed that might have some interest. Right now a Pentium or Pentium MMX would not be within reach of a FPGA.

Mainly there are two communities that would be interested. A "GOG for DOS" board that would make the DOS games more like a console that can be used on conventional monitors and televisions, this would basically need to be something that the CPU can be clocked independently of the ISA bus. The other community would be the developer/educational market who is more interested in working within the confines of older systems.

But there is something of a problem with trying to create such things. Unlike say an actual console which has very little variation, there is no way to connect real ISA hardware to such things since old ISA hardware operates on 5V while many new FPGA type devices operate at 1.8V now. So you end up adding a bunch of level shifters. So just the CPU would not be enough. A bunch of parts would need to be made, and while I'd be personally interested such a thing, I really doubt there is enough interest to mass-produce ASIC's.

Let's say for example, you build a Pentium-MMX compatible core, and a North Bridge/SouthBridge that supports PCI and ISA. The next part you need is the Video (VGA, DVI/HDMI,) which could be another stackable board, and then a storage board(USB storage, SD-card), and a sound board(S/PDIF in/out, Analog stereo output and input.) Like these could all just be FPGA's with different IO connectors to allow changes. But a complete "GOG" ASIC would be pretty hard create something that works for everyone, and that is what I think is going to happen. Like right now with dosbox, people switch out Adlib OPL2 to OPL3, to General Midi or MUNT(MT-32), cloning a MT-32 on a FPGA would be interesting but it would likely need a much larger chip, and still has the problem of the ROM's.

What we have right now with DOSBOX is a bit more flexible for the purpose of playing old games, but we also can't replicate many of the edge cases (eg 8088MPH) because many of those require the original timing quirks, and largely the problem is people trying to use DOSBOX for things that are neither DOS (eg Windows), or games (eg word processors, SCADA management devices, etc.)

HDMI output/scan-doubling can be driven by a FPGA. Pretty much everything can be done with a single FPGA, but you don't really have a lot of space on cheap FPGA's. Like it costs about $100 for a FPGA that can emulate 8-bit and 16-bit computers http://harbaum.org/till/mist/index.shtml (see https://github.com/mist-devel/mist-board/wiki … FPGA%20Projects ), http://www.fpgaarcade.com/replay/ but when you start emulating multi-chip computers and consoles it rapidly requires something more complicated and expensive.

Like a 486 SOC http://opencores.org/project,ao486 for a FPGA, note that it requires as many logic cells for the CPU as it does for the sound. (Cyclone IV E development board, which costs $595.) But the performance of that 486 core is closer to a 30Mhz system. A Pentium FPGA http://www.eecg.toronto.edu/~yiannac/docs/fpga07.pdf requires more logic elements but they could only operate it at 25Mhz.

So ultimately I believe what would be the most useful would be to develop a complete x86 core and multiple cheap expansion boards to do the i/o rather than trying to to stuff it all into one ASIC and end up with less compatibility than software emulators.

Reply 13 of 13, by Jorpho

User metadata
Rank l33t++
Rank
l33t++
Azarien wrote:

ISA only vs VLB vs PCI

No real slots are needed.

Seems to me one of the biggest reasons people keep using old systems is so they can keep using their old Voodoos and ISA sound cards. Of course, I suppose you could integrate such features into this theoretical board, except choosing sound hardware that will satisfy everyone seems to be a hopeless cause.