Converting a Core into Real Hardware
Converting a Core into Real Hardware
This question is more of a curiosity than anything else. Has anyone converted a MiSTer core into real hardware? For example, build an NES clone console from the NES core. I don't intend on purchasing anything like this. Just curious.
Re: Converting a Core into Real Hardware
If that's what you mean, someone put a MiSTer Setup in the Shell of a Neo Geo X: viewtopic.php?t=1377
Re: Converting a Core into Real Hardware
Well if you got about $500K - 1 million+ USD you might be able to run a few hundred thousand ASIC chips from your FPGA core (these are of course minimums in terms of cost and run numbers) .. but I dont know anyone who has that kind of spare cash just at the ready, plus you have to of course find a chip fab company that will partner with you etc.. then of course there are the sticky issues of legality of if you even have the rights to do it.. and if you arent clear and prove IP rights allowing you to make the chips, no fab will even talk to you.
Re: Converting a Core into Real Hardware
Thanks for the info. What got me thinking this is all the clone consoles using bad quality SOCs or 6502 clone chips.
I see that this has been brought up before.
viewtopic.php?t=3334
I see that this has been brought up before.
viewtopic.php?t=3334
Re: Converting a Core into Real Hardware
Design patents on these retro consoles expired long ago. Patents wise, the IP right holders can't touch you.ericgus09 wrote: ↑Tue Jan 11, 2022 12:14 am Well if you got about $500K - 1 million+ USD you might be able to run a few hundred thousand ASIC chips from your FPGA core (these are of course minimums in terms of cost and run numbers) .. but I dont know anyone who has that kind of spare cash just at the ready, plus you have to of course find a chip fab company that will partner with you etc.. then of course there are the sticky issues of legality of if you even have the rights to do it.. and if you arent clear and prove IP rights allowing you to make the chips, no fab will even talk to you.
Keep in mind there 3 areas of IP law, Patents (physical designs), Copyright (anything written or artistically created, that includes software) and Trade marks (I.E. this is my Stamp, Logo, makers mark, ect).
You are free to make ASICs of the hardware of the NES now, Nintendo could try to take you to court, but most judges would see the patent expired and a motion to dismiss the case will be made. Its how Rretro USB and Analogue Interactive can make their clone FPGA consoles without legal reprisal.
The Catch 22 is Copyright, that would include any BIOS or Boot Files, OS/Kernel Data, Firmware Code, Micro-code, ect.
It is my great regret that we live in an age that is proud of machines that think and suspicious of people who try to.
Re: Converting a Core into Real Hardware
Yes true..darksakul wrote: ↑Tue Jan 11, 2022 1:13 amDesign patents on these retro consoles expired long ago. Patents wise, the IP right holders can't touch you.ericgus09 wrote: ↑Tue Jan 11, 2022 12:14 am Well if you got about $500K - 1 million+ USD you might be able to run a few hundred thousand ASIC chips from your FPGA core (these are of course minimums in terms of cost and run numbers) .. but I dont know anyone who has that kind of spare cash just at the ready, plus you have to of course find a chip fab company that will partner with you etc.. then of course there are the sticky issues of legality of if you even have the rights to do it.. and if you arent clear and prove IP rights allowing you to make the chips, no fab will even talk to you.
Keep in mind there 3 areas of IP law, Patents (physical designs), Copyright (anything written or artistically created, that includes software) and Trade marks (I.E. this is my Stamp, Logo, makers mark, ect).
You are free to make ASICs of the hardware of the NES now, Nintendo could try to take you to court, but most judges would see the patent expired and a motion to dismiss the case will be made. Its how Rretro USB and Analogue Interactive can make their clone FPGA consoles without legal reprisal.
The Catch 22 is Copyright, that would include any BIOS or Boot Files, OS/Kernel Data, Firmware Code, Micro-code, ect.
Again I dont think any of the major chip fabs would even talk to you with the slightest whiff of this kinda thing in the air .. while you are right, they really are mostly very risk-averse and will likely decline your request from the start simply to avoid even the slightest problems with the law legitimate or otherwise.. No one wants to have to go thru an expensive legal defense even if you are in the right. Its just too costly.
-
- Posts: 97
- Joined: Wed Sep 08, 2021 12:09 pm
- Has thanked: 25 times
- Been thanked: 13 times
Re: Converting a Core into Real Hardware
The cost is certainly the main reason why Nintendo preferred to choose a SOC design for its own minis. Although tens of millions of have been made here.
Re: Converting a Core into Real Hardware
There has been thousands of ASICs that are NOC (Nintendo on a Chip) that already been made.ericgus09 wrote: ↑Wed Jan 12, 2022 3:32 amYes true..darksakul wrote: ↑Tue Jan 11, 2022 1:13 amDesign patents on these retro consoles expired long ago. Patents wise, the IP right holders can't touch you.ericgus09 wrote: ↑Tue Jan 11, 2022 12:14 am Well if you got about $500K - 1 million+ USD you might be able to run a few hundred thousand ASIC chips from your FPGA core (these are of course minimums in terms of cost and run numbers) .. but I dont know anyone who has that kind of spare cash just at the ready, plus you have to of course find a chip fab company that will partner with you etc.. then of course there are the sticky issues of legality of if you even have the rights to do it.. and if you arent clear and prove IP rights allowing you to make the chips, no fab will even talk to you.
Keep in mind there 3 areas of IP law, Patents (physical designs), Copyright (anything written or artistically created, that includes software) and Trade marks (I.E. this is my Stamp, Logo, makers mark, ect).
You are free to make ASICs of the hardware of the NES now, Nintendo could try to take you to court, but most judges would see the patent expired and a motion to dismiss the case will be made. Its how Rretro USB and Analogue Interactive can make their clone FPGA consoles without legal reprisal.
The Catch 22 is Copyright, that would include any BIOS or Boot Files, OS/Kernel Data, Firmware Code, Micro-code, ect.
Again I dont think any of the major chip fabs would even talk to you with the slightest whiff of this kinda thing in the air .. while you are right, they really are mostly very risk-averse and will likely decline your request from the start simply to avoid even the slightest problems with the law legitimate or otherwise.. No one wants to have to go thru an expensive legal defense even if you are in the right. Its just too costly.
Most fab houses aren't even going to look that closely to see what you are having made, as long as it works to specifications
It is my great regret that we live in an age that is proud of machines that think and suspicious of people who try to.
- Newsdee
- Top Contributor
- Posts: 830
- Joined: Mon May 25, 2020 1:07 am
- Has thanked: 98 times
- Been thanked: 209 times
Re: Converting a Core into Real Hardware
The C64 DTV was a good example of the pitfalls of trying to make an ASIC clone ( https://www.c64-wiki.com/wiki/C64DTV ).
When it came out, I figured it would be reused widely, like the NES SoC that powers so many cheap retro game products.
It has sold well (600K+ units according to the above) yet there aren't any more commercial products based on it.
According to this crowdfunding page for The C64:
https://www.indiegogo.com/projects/thec ... -console#/
When it came out, I figured it would be reused widely, like the NES SoC that powers so many cheap retro game products.
It has sold well (600K+ units according to the above) yet there aren't any more commercial products based on it.
According to this crowdfunding page for The C64:
https://www.indiegogo.com/projects/thec ... -console#/
I suppose the ASIC did not sell well enough to allow for a v2 that fixed some of the problems. I also suspect that for TheC64 they didn't have hardware designer onboard, so going the FPGA or ASIC route would have been even more expensive than just programming an emulator for some commonplace phone SoC.When we created the C64DTV we faced similar challenges. Creating Commodore hardware that will accurately run the BIOS and software is a challenge, whether that’s a custom ASIC or an off the shelf programmable FPGA, it requires a considerable amount of time, effort and resources. And we learned from that. We can improve on the limited expandability of the C64DTV!
-
- Posts: 97
- Joined: Wed Sep 08, 2021 12:09 pm
- Has thanked: 25 times
- Been thanked: 13 times
Re: Converting a Core into Real Hardware
I think one has little to do with the other. The C64DTV was actually developed by Jerry Ellsworth and is believed to have been a by-product of the C-One development. A SOC kit was used with TheC64. There isn't much left of the version with the module slot shown in the campaign.
- Newsdee
- Top Contributor
- Posts: 830
- Joined: Mon May 25, 2020 1:07 am
- Has thanked: 98 times
- Been thanked: 209 times
Re: Converting a Core into Real Hardware
Seems like this guy Darren is the common link. He was working at IronStone Partners which got the C64 rights and licenses from Tulip to develop the DTV. Perhaps he was the marketer / "idea person" behind the DTV, whereas Jeri Ellsworth was part of the hardware development team. Apparently he worked on licensing C64 games on Wii Virtual console as well. (source: posts on various forums, and c64 wikis)
In any case, for whatever reason TheC64 is indeed a SoC. These people were aware of ASICs / FPGAs from the DTV development, but for some reason did not go with that option. My guess is that the ASICs would have need "reprinting" (or whatever the term is to produce new units), and that rehiring a hardware designer to review the ASIC logic was not desirable, especially after (successfully) dabbling with emulation on the Wii Virtual console.
I understand it costs a lot of money do a print of ASICs - even though the individual cost is small, you have to make tens of thousands to justify the setup cost of the machinery. And that is a huge bottleneck unless you are super rich AND a fan of retro computing
![Smile :)](./images/smilies/icon_e_smile.gif)
Re: Converting a Core into Real Hardware
by this logic emulation is running on "real hardware" which for the purposes of using the term emulation makes the statement very misleading
-
- Core Developer
- Posts: 537
- Joined: Sun May 24, 2020 9:30 pm
- Has thanked: 19 times
- Been thanked: 143 times
Re: Converting a Core into Real Hardware
Well, yes and no… because a computer is not a console, and there are a huge number of abstraction-level variances in the operating system. Software emulation (as commonly implemented) is “real hardware” in the sense that it runs on electricity though.
However, remove the operating system abstractions, and make base operations happen in real time (rather than speed up/slow down), and simplify the circuitry so that it doesn’t dwarf the original machine, including entire subsystems which didn’t exist on the original, and I’d be inclined to agree that it is “real hardware” in the same way as a FPGA. Check out the BBC Micro emulator running on a RPi Pico for that.
https://www.hackster.io/news/graham-san ... 4c078a6a9a
But for the FPGA implementations, I’m not sure what makes you think there is enough of a tangible difference to make a distinction (when implemented properly, they are signal-equivalent).
-
- Top Contributor
- Posts: 1911
- Joined: Wed May 27, 2020 1:59 pm
- Has thanked: 145 times
- Been thanked: 454 times
Re: Converting a Core into Real Hardware
Here are some examples:
https://www.retrorgb.com/fpga-based-ym2 ... ement.html
https://www.cast-inc.com/blog/chip-repl ... r-example/
Millogic achieves milestone for sales of drop-in FPGA based obsolescence substitute devices
March 16, 2019 - Millogic sales of pin-compatible FPGA emulators of obsolete components has hit a milestone of 10,000 units. Our emulator assemblies have qualified as alternative source devices for processors and controllers that our customers have no longer been able to source. These production quantities have been installed in Medical, Defense and Industrial products, where the finished product life exceeds the commercial availability of some components. These solutions are particularly viable in industries with extensive and expensive qualification processes, such as FDA approvals or some defense industry requirements, where qualification of an alternate source for a component is much shorter than the qualification of a new design.
-
- Core Developer
- Posts: 537
- Joined: Sun May 24, 2020 9:30 pm
- Has thanked: 19 times
- Been thanked: 143 times
Re: Converting a Core into Real Hardware
Yeah, so… this is just pulling apart words and ascribing pedantic meanings to them… in that sense, they are all “emulators” and all “real hardware”. Which is fine, and literally true.
But in effect you are asking for an all-new lexicon to be created to describe any gaps/differences, which I am not against, but also not going to propose, as most people already have a clear idea of what “emulator” means to them (connotation), despite its denotative meaning.
But I guess you could also say - getting back to the original poster’s topic/question - that I was, in my own way, asking for a more precise definition of what they meant by “real hardware”.
But in effect you are asking for an all-new lexicon to be created to describe any gaps/differences, which I am not against, but also not going to propose, as most people already have a clear idea of what “emulator” means to them (connotation), despite its denotative meaning.
But I guess you could also say - getting back to the original poster’s topic/question - that I was, in my own way, asking for a more precise definition of what they meant by “real hardware”.
Re: Converting a Core into Real Hardware
The MiSTer implementations of real hardware do have enough constraints/limitations that you wouldn't generally consider them fully equivalent, like in the sense that you can connect all the original media readers or any peripherals that would have worked. I wouldn't sell someone an Amiga and give them a MiSTer instead although I do understand your pointdshadoff wrote: ↑Sat Jan 22, 2022 2:16 pmWell, yes and no… because a computer is not a console, and there are a huge number of abstraction-level variances in the operating system. Software emulation (as commonly implemented) is “real hardware” in the sense that it runs on electricity though.
However, remove the operating system abstractions, and make base operations happen in real time (rather than speed up/slow down), and simplify the circuitry so that it doesn’t dwarf the original machine, including entire subsystems which didn’t exist on the original, and I’d be inclined to agree that it is “real hardware” in the same way as a FPGA. Check out the BBC Micro emulator running on a RPi Pico for that.
https://www.hackster.io/news/graham-san ... 4c078a6a9a
But for the FPGA implementations, I’m not sure what makes you think there is enough of a tangible difference to make a distinction (when implemented properly, they are signal-equivalent).
Re: Converting a Core into Real Hardware
I guess I should have said, converting a core into an ASIC or something like that.
Example: Convert the NES core into CPU and PPU replacement chips that can be used to refurbish a broken NES.
Example: Convert the NES core into CPU and PPU replacement chips that can be used to refurbish a broken NES.
Re: Converting a Core into Real Hardware
I can't find the link, but I'm pretty sure someone used one of jotego's Yamaha cores (jtopl, jt12, jt51) to build a module that plugs into an original system.
-
- Core Developer
- Posts: 537
- Joined: Sun May 24, 2020 9:30 pm
- Has thanked: 19 times
- Been thanked: 143 times
Re: Converting a Core into Real Hardware
Quite a few of the cores have the key chips broken down that way, and work hard to have the same signals, timing, and transitions. (Though not all do this.)
These days, the real problem would be the required level-shift for any system which runs as 5V logic, or buffering required for CMOS-to-bipolar transition.
(FPGAs are low-power CMOS which basically can drive 3.3V as the maximum - so would need level-shift chips - and CMOS isn't meant to drive the current used by bipolar TTL - which was generally phased-out in the mid-1980s due to power consumption).
These days, the real problem would be the required level-shift for any system which runs as 5V logic, or buffering required for CMOS-to-bipolar transition.
(FPGAs are low-power CMOS which basically can drive 3.3V as the maximum - so would need level-shift chips - and CMOS isn't meant to drive the current used by bipolar TTL - which was generally phased-out in the mid-1980s due to power consumption).
- Newsdee
- Top Contributor
- Posts: 830
- Joined: Mon May 25, 2020 1:07 am
- Has thanked: 98 times
- Been thanked: 209 times
Re: Converting a Core into Real Hardware
Here's a recent chat with Jeri Ellsworth where she gives quite a bit of detail about converting an FPGA design for the C64 DTV into an ASIC, and some other fun behind-the-scenes anecdotes of the DTV project: https://youtu.be/C2hLIiogQrY?t=272
tl;dw takeaways relevant for this thread:
tl;dw takeaways relevant for this thread:
- It costs millions to do an ASIC wafer and send it to print
- The DTV only happened because it had a toy company behind it
- Original interest was because they tried to use the NES SoC but it was not fit for purpose
- Going from FPGA design to an actual ASIC chip required a whole extra set of skills - they barely pulled it off on time given tight deadlines
- She still has ~250,000 new DTV chips somewhere... the company didn't know what to do with it and did not want to produce more
- Nowadays it is much much cheaper to emulate