FPGA Based Videogame System

[Schizophretard]

but support digital out as well as analog (for who still cares about it)

 

According to this thread's poll it is the majority that still cares about it.

[Keatah]

What's stopping them now to create/add them to existing powerful development boards?

 

https://www.arrow.com/en/products/deca/arrow-development-tools

65US$ for 50K LE. HDMI connector already in, SD card reader already in, 512MB of SDRAM in etc...etc....

 

Is it the aluminum milled case? Is it the fact that they can design and implement a perfect arcade core (to the transistor level no less) but can't design a PCB to interface a joypad? What is it?

 

I mean seriously the only thing I think those "experts" seems to be waiting for is for someone else to prepare the hardware (ports and all) that resembles something users may want in their home but then again some are fine with an RPi in plastic case, sell said HW to as many people as possible with an initial set of kick-ass cores (build an audience), open up the architecture (sort of document/prepare an "SDK" for the HW) so they can jump in the winner bandwagon.

 

Yes. Why indeed.. An excellent point.

 

In taking a simple viewpoint, I wouldn't want to be re-inventing the wheel. It'd seem like a hollow pursuit. AND there really is no standard for an FPGA console unlike an R-Pi or (even better) an x86 box full of software emulators. Not yet. One cannot simply take a core/file and load it into another rig. It has to be redone for each flavor of hardware. Each FPGA console is going to live and die by its own merits - whatever combination of hardware and software it comes with. FPGA is still niche. Like pinhole niche. And in only sections of the pinhole, off to the side, under some debris niche. I don't mean that in a bad way. Nor can anyone simply step in and start making additions and improvements to what's already out there. It mostly has to be done by the system's designer.

 

What I personally find annoying, and you may or may not agree, is that many cores are incomplete. They're just enough to run most games and demos. And if an 8-bit computer is being simulated, doing archiving work or something more serious/technical, you're pretty much S.O.L. There aren't many enriching extras and tie-ins to the host/pc environment. No debugger, no special peripherals, no import/export niceties, none of that.

 

But if you look at Software Emulation it's all over the place, prevalent, pervasive, and everyone knows about it. And they're likely using it even if they don't know about it. The bandwagon is turbocharged and rolling. Software Emulation seems to have an endless amount of extras and cool features.

 

Stella, the venerable VCS emulator, is in the final stages of getting an internal makeover. These last years have seen it become a team effort. Even new bank-switch schemes are being developed and vetted on it. So, as accuracy and detail in emulators (of all kinds) increase and move forward, the time involved in the work seems to increase exponentially. Getting the last bugs out can take months.

 

And MAME has a list of contributors now numbering in the hundreds.. But that's a whole other beast.

 

So, I think that two things are missing from the FPGA arena:

 

1- Teamwork and collaboration. As the old saying goes, two pairs of eyes are better than one. New ideas & new features coming from every angle. Many people looking at the same problem.. The list of benefits goes on.

 

2- Installed base of hardware. Between Android, MacOS, Linux/Windows/x86, and R-Pi, there are millions upon millions of platforms out there and ready to go. MAME alone sees 25,000 downloads the first week a new version comes out.

 

---

 

Many folks cite the accuracy and exactness and consistency FPGA simulation offers. It really comes down to who is doing the programming - because there are highly accurate Software Emulators as well. And for the emus that aren't quite as perfect as the original hardware, that's ok. I'm willing to give up a couple percentage points to gain all the convenience and versatility they offer. The classic case of the good outweighing the bad.

[Zonie]

FWIW, the current versions of MAME suck. What I was using in 1996 was much easier to deal with.

[phoenixdownita]

....

 

So what you're really saying is that Kev has no business letting other people poke around his hardware?

Nope all I said is that it's his choice and he already spoke, given the amount of support he would have to undertake he doesn't want to bother.

The second thing I said is that if in order to support these other "core developers" he needs to do work that he didn't plan to it's no dealbreaker if he doesn't do anything more than what's needed to support his cores his way.

 

And if I may why do you want someone else to devote resources into making the do all of FPGA arcades at home? The MAME of FPGAs so to speak.

I believe that he can if he wants to but he also has a job, there's already arcade cores for MiST so just buy that one, the "arcade core experts" can use that too, right?

[Andromeda Stardust]

I don't think that Kevtris is intentionally making the NT Mini and/or Zimba3000 unhackable. It's just the instruction language for FPGAs is so kryptic that there's maybe a few dozen individuals on the planet who have the skillsets to write one who also have an interest in preserving video games through FPGA. Any of those few people who come out of the woodworks would probably produce their own hardware as Kevtris has done, rather than reverse assemble someone elses.

 

To produce a third party core for Zimba / NT_Mini, you must not only reverse assemble the device you plan to emulate, but also must reverse assemble the NT Mini or Zimba 3000 itself, and that is a tall order.

 

PS, Kevtris you rock!

[kevtris]

 

Yes. Why indeed.. An excellent point.

 

In taking a simple viewpoint, I wouldn't want to be re-inventing the wheel. It'd seem like a hollow pursuit. AND there really is no standard for an FPGA console unlike an R-Pi or (even better) an x86 box full of software emulators. Not yet. One cannot simply take a core/file and load it into another rig. It has to be redone for each flavor of hardware. Each FPGA console is going to live and die by its own merits - whatever combination of hardware and software it comes with. FPGA is still niche. Like pinhole niche. And in only sections of the pinhole, off to the side, under some debris niche. I don't mean that in a bad way. Nor can anyone simply step in and start making additions and improvements to what's already out there. It mostly has to be done by the system's designer.

 

What I personally find annoying, and you may or may not agree, is that many cores are incomplete. They're just enough to run most games and demos. And if an 8-bit computer is being simulated, doing archiving work or something more serious/technical, you're pretty much S.O.L. There aren't many enriching extras and tie-ins to the host/pc environment. No debugger, no special peripherals, no import/export niceties, none of that.

 

But if you look at Software Emulation it's all over the place, prevalent, pervasive, and everyone knows about it. And they're likely using it even if they don't know about it. The bandwagon is turbocharged and rolling. Software Emulation seems to have an endless amount of extras and cool features.

 

Stella, the venerable VCS emulator, is in the final stages of getting an internal makeover. These last years have seen it become a team effort. Even new bank-switch schemes are being developed and vetted on it. So, as accuracy and detail in emulators (of all kinds) increase and move forward, the time involved in the work seems to increase exponentially. Getting the last bugs out can take months.

 

And MAME has a list of contributors now numbering in the hundreds.. But that's a whole other beast.

 

So, I think that two things are missing from the FPGA arena:

 

1- Teamwork and collaboration. As the old saying goes, two pairs of eyes are better than one. New ideas & new features coming from every angle. Many people looking at the same problem.. The list of benefits goes on.

 

2- Installed base of hardware. Between Android, MacOS, Linux/Windows/x86, and R-Pi, there are millions upon millions of platforms out there and ready to go. MAME alone sees 25,000 downloads the first week a new version comes out.

 

---

 

Many folks cite the accuracy and exactness and consistency FPGA simulation offers. It really comes down to who is doing the programming - because there are highly accurate Software Emulators as well. And for the emus that aren't quite as perfect as the original hardware, that's ok. I'm willing to give up a couple percentage points to gain all the convenience and versatility they offer. The classic case of the good outweighing the bad.

Honestly I was fairly disappointed with the existing crop of cores for the other FPGA boards. The computer cores (Amiga, etc) seem to be well represented and refined, but cores for videogame consoles themselves seemed to be primitive at best. Someone would start on a core, then when it got too difficult just abandon it. Zero followup ever happened. I am not knocking anyone here- it's just the reality of the situation as I saw it at the time. A couple years later it might be better- I haven't checked in quite awhile.

 

Do remember that hardware solutions aren't going to have a lot of "software niceties" like debuggers and such. At least not without a lot of work. The best core is going to work exactly like a real system, warts and all. Software emulators are definitely worlds better for code development IMO than a hardware solution (be it a real system or an FPGA one). The FPGA's claim to fame is you can get much closer accuracy/lag vs. a software solution.

 

I don't know how well collaboration will work with FPGA stuff. Sure it can be done, but you have to find two people who share the same interest in systems/hardware and have the same FPGA board and tools. It definitely works in a corporate environment- you're paying 2 or more people to work together on a single project so collab works great there. When it comes to hobby projects, it's more like herding cats. And yes I've made a lot of contributions to MAME and MESS over the years. Both technical information and hard to do binary code extraction from mask ROM microcontrollers using totally undocumented factory test modes.

 

 

 

Nope all I said is that it's his choice and he already spoke, given the amount of support he would have to undertake he doesn't want to bother.

The second thing I said is that if in order to support these other "core developers" he needs to do work that he didn't plan to it's no dealbreaker if he doesn't do anything more than what's needed to support his cores his way.

 

And if I may why do you want someone else to devote resources into making the do all of FPGA arcades at home? The MAME of FPGAs so to speak.

I believe that he can if he wants to but he also has a job, there's already arcade cores for MiST so just buy that one, the "arcade core experts" can use that too, right?

 

Pretty much this. I don't have time to do it, not with my "real" job and paid work I do after my day job. I will be full up on work until some time next year at this rate, unfortunately.

 

I don't think that Kevtris is intentionally making the NT Mini and/or Zimba3000 unhackable. It's just the instruction language for FPGAs is so kryptic that there's maybe a few dozen individuals on the planet who have the skillsets to write one who also have an interest in preserving video games through FPGA. Any of those few people who come out of the woodworks would probably produce their own hardware as Kevtris has done, rather than reverse assemble someone elses.

 

To produce a third party core for Zimba / NT_Mini, you must not only reverse assemble the device you plan to emulate, but also must reverse assemble the NT Mini or Zimba 3000 itself, and that is a tall order.

 

PS, Kevtris you rock!

 

I saw this hilarious venn diagram on nesdev the other day and it perfectly fits this situation I think...

 

 

In summary, you need to become the "Banana" :-)

[+eebuckeye]

Hi Kevtris..

 

When might we expect an update to the Mini NT? Thanks!

[kevtris]

Hi Kevtris..

 

When might we expect an update to the Mini NT? Thanks!

seeing how I'm working 10-12 hour days, 7 days a week until at least next year, figure next year some time.

[Wolf_]

seeing how I'm working 10-12 hour days, 7 days a week until at least next year, figure next year some time.

[+eebuckeye]

I understand but I'm bummed... I was waiting for the Intellivision cores and other cores to be finished. Also thought cartridge and joystick adapters were coming out.

[NE146]

In summary, you need to become the "Banana" :-)

 

Kev, is this you?

 

[kevtris]

loooool. I intend to do as little dying as possible.

[Zonie]

 

Kev, is this you?

 

 

https://www.youtube.com/watch?v=yMUW3ABGt_c

[bikerspade]

I upgraded my Hi-Def NES front-loader from 2.25 to 3.01 today. I use an Everdrive N8 to play FDS games. I notice that in the FDS version of Metroid, the sound effect is now missing when shooting open doors. I made sure that the FDS sound channel is enabled. It didn't seem to have this issue with firmware 2.25. Any ideas? With this latest update, the FDS sound effects seem to work ok for FDS Zelda 1 and when transforming into a ball in FDS Metroid.

Edited May 5, 2017 by bikerspade

[atmn]

seeing how I'm working 10-12 hour days, 7 days a week until at least next year, figure next year some time.

Take your time. Think people have some playing around with current cores do do yet.

 

Noticed thats the 16-bit stuff cores that are "on hold" on the confirmed releases. Are the 16-bit cores more challanging, even if its a primitive system like Intellivision?

[retro_fan]

[...]

 

The FPGA's claim to fame is you can get much closer accuracy/lag vs. a software solution.

 

[...]

Hmm... There seem to be some efforts lately to investigate and reduce the lag in software emulators, I wonder to what an extent it can be reduced.

[Newsdee]

Honestly I was fairly disappointed with the existing crop of cores for the other FPGA boards. The computer cores (Amiga, etc) seem to be well represented and refined, but cores for videogame consoles themselves seemed to be primitive at best. Someone would start on a core, then when it got too difficult just abandon it. Zero followup ever happened. I am not knocking anyone here- it's just the reality of the situation as I saw it at the time. A couple years later it might be better- I haven't checked in quite awhile.

(...)

I don't know how well collaboration will work with FPGA stuff.

There is a difference with many recent cores in that they are open source. While that doesn't guarantee they will get a follow-up, there have been many cases of cores "resurrected" and slightly improved by another developer. The Macintosh core for the MiST is a great example. Granted, thats a computer core.

 

I feel there are a few parts of the problem though (but this is just my impression). First every HDL designer is skilled enough to design their own board so they might not be happy with compromises done with somebody else's board. Then they might not be willing to open source for various reasons (same as many software developers), or even if they do there is no real incentive to spend the 80% of time needed to finish the last 20% of functionality, other than "online recognition" which is more or less important to different people.

 

TL;DR: consolized "expandable" hardware already exists with open sources, but still many developers won't develop for it for various reasons.

[Keatah]

Hmm... There seem to be some efforts lately to investigate and reduce the lag in software emulators, I wonder to what an extent it can be reduced.

 

Some emulators are already very good and lag by only 1 or 2 additional frames compared against the original console+cartridge. But now that attention is being focused there I believe improvements will be coming. How much? That's going to be on an emulator-by-emulator basis. I also suspect they'll come from controllers and their interfaces.

[Andromeda Stardust]

Honestly I was fairly disappointed with the existing crop of cores for the other FPGA boards. The computer cores (Amiga, etc) seem to be well represented and refined, but cores for videogame consoles themselves seemed to be primitive at best. Someone would start on a core, then when it got too difficult just abandon it. Zero followup ever happened. I am not knocking anyone here- it's just the reality of the situation as I saw it at the time. A couple years later it might be better- I haven't checked in quite awhile.

 

Do remember that hardware solutions aren't going to have a lot of "software niceties" like debuggers and such. At least not without a lot of work. The best core is going to work exactly like a real system, warts and all. Software emulators are definitely worlds better for code development IMO than a hardware solution (be it a real system or an FPGA one). The FPGA's claim to fame is you can get much closer accuracy/lag vs. a software solution.

 

I don't know how well collaboration will work with FPGA stuff. Sure it can be done, but you have to find two people who share the same interest in systems/hardware and have the same FPGA board and tools. It definitely works in a corporate environment- you're paying 2 or more people to work together on a single project so collab works great there. When it comes to hobby projects, it's more like herding cats. And yes I've made a lot of contributions to MAME and MESS over the years. Both technical information and hard to do binary code extraction from mask ROM microcontrollers using totally undocumented factory test modes.

 

 

 

 

Pretty much this. I don't have time to do it, not with my "real" job and paid work I do after my day job. I will be full up on work until some time next year at this rate, unfortunately.

 

 

I saw this hilarious venn diagram on nesdev the other day and it perfectly fits this situation I think...

 

 

In summary, you need to become the "Banana" :-)

+ infinity. Unfortunately, I only have one "like" to give, but perhaps an infinite number of people feel the same way I do. Doubtful, but there are probaly a billion earth-like planets in the universe, so i'll file this under "plausible."

[Andromeda Stardust]

seeing how I'm working 10-12 hour days, 7 days a week until at least next year, figure next year some time.

less time posting; more time working!

[Andromeda Stardust]

Take your time. Think people have some playing around with current cores do do yet.

LOL, U said "do-do"...

[Wolf_]

 

Some emulators are already very good and lag by only 1 or 2 additional frames compared against the original console+cartridge. But now that attention is being focused there I believe improvements will be coming. How much? That's going to be on an emulator-by-emulator basis. I also suspect they'll come from controllers and their interfaces.

According to nonmame (a web page that recommends the most accurate emulators per system) a snes emulator exists that now that is accurate to the system cycle of the original snes. That said though it is pretty system intensive on the computer that runs it as well as maybe possibly having some compatibility issues as it does not do everything exactly the same as the original system in terms of coding. Besides that the snes is not incredibly difficult to run and doing the same thing with say a gamecube would probably require triple the processing power or more where as fpgas will probably reach that stage significantly faster.

 

edit: just noticed the snes emulator shoutout was what was linked to in the post.

[Andromeda Stardust]

According to nonmame (a web page that recommends the most accurate emulators per system) a snes emulator exists that now that is accurate to the system cycle of the original snes. That said though it is pretty system intensive on the computer that runs it as well as maybe possibly having some compatibility issues as it does not do everything exactly the same as the original system in terms of coding. Besides that the snes is not incredibly difficult to run and doing the same thing with say a gamecube would probably require triple the processing power or more where as fpgas will probably reach that stage significantly faster.

 

edit: just noticed the snes emulator shoutout was what was linked to in the post.

Does the creator's username start with "byuu..."?

[Schizophretard]

+ infinity. Unfortunately, I only have one "like" to give, but perhaps an infinite number of people feel the same way I do. Doubtful, but there are probaly a billion earth-like planets in the universe, so i'll file this under "plausible."

I have been liking all of his posts and those are just to make up for all my likes for his opening post.

[cacophony]

Does the creator's username start with "byuu..."?

 

Here's the quote Wolf's referring to:

 

"bsnes accuracy is VERY close to 100%, with a 65816 CPU core that is - are you sitting down? - accurate to the clock cycle. "

(from http://nonmame.retrogames.com/)

Edited May 6, 2017 by cacophony