FPGA Based Videogame System

[SegaSnatcher]

 

You mean it's okay for you and others to slag off MiSTer, but when they do that then it's a nasty inferiority complex? Strange logic

 

I do not really have a horse in this race, but this thread does sometimes "reek" of Analogue fanboyism, the last couple of pages or so regarding MiSTer show that quite well. Seems to me that some people who invested lots of dosh in its products would hate to see an open source, much cheaper solution to do well, hence some heavy rationalizing going on here and anti-MiSTer condescension/sniping too. That on top of the general hype train and the 1337 club feel, which also may or may not be quite "douchey". So, not saying of course that the other side should act nasty but there are definitely two sides to this story.

 

While I personally would much rather buy MiSTer, I do not have any sort of problem with the existence of Analogue's consoles and wish Kevtris himself well. But some opinions posted here really are cringeworthy. Is it really so hard to understand that some folk may prefer to spend ~200$ and have heaps of open source cores available, with very little tinkering involved, than buy Nt Mini for 3 (or 4) times the price? Comparing MiSTer to Super NT/Mega does not even make much sense since they're one-core offerings. And believe it or not, not everybody is forever locked in the SMS/SNES/NES/Megadrive world, there are people very much interested in microcomputers and less popular consoles too.

 

And the added bonus here is some really intense - 0-lag proselytizing. Now we all know that lag is bad for you, mmkay, but seriously...dismissing everything that is not "0" as straight up garbage is just heavy-eyeroll worthy. As someone who plays mostly on a RPi+CRT, with resulting 4-6 (?) frames of lag, does not feel it and yet manages to do quite well in assorted High Score clubs, I can attest to that. Sure, we all should strive to minimalise that unwelcome malady, but that "I'm sensitive and refuse to play on anything <0" humblebragging can get quite tiresome. Unless people like SS here start coming up with arcane equations which involve even the baked in lag, or dissing 2-day old cores, then it's quite hilarious actually.

 

tl;dr: live and let live, and don't turn this Analogue thing into a snobbish ivory tower. It'd be much better if we all just could get along, different markets for different people. Otherwise, don't be surprised if people sneer back at you too.

It doesn't have to be "0" lag necessarily, but 2 frames is just too much when most TVs inherently have 1 - 2 frames of lag already baked in. In its current form I personally don't see how MiSTer is superior to say a software emulation solution, especially with the run ahead latency feature. The only way to reduce lag on the MiSTer is to buy the IO board and run the analog out to an OSSC which is a $200 investment on top of the $130 DE-10 + however much the IO Board costs. Its great that you can still find enjoyment playing games with a lot of added latency, myself and some others simply can't.

Edited December 3, 2018 by SegaSnatcher

[Andromeda Stardust]

 

It's perfectly fine to hold your own opinion about open sources vs closed as Sorg did. The spirit of open source is that anyone (or someone) can eventually continue the legacy of someone (or people) prior as far as R&D. Research, code and resources won't exactly die with a creator or be locked behind a company forever. However, it's Kevtris' prerogative to do whatever he wants to do with his projects & code, and everyone will just have to deal with it. This was posted in May:

 

http://www.atari-forum.com/viewtopic.php?t=29415#p344366

 

 

There's a very high level of jealousy & disdain towards Kevtris/Analogue that I've seen from some (not many) people who have jumped onto the MiSTer project, even as end-users. Mainly due to cores not being released to open-source/ported to MiSTer. People definitely consider any core or recreation of a chip by Kevtris to be much more accurate than MiSTer's current offerings. I don't think Sorgelig even dislikes Kevtris at all, he only hates the general idea of closed source projects and is very committed to his idea of and dream of open source (which is interesting discussion in and of itself in the last few days).

Sorry, but Kevtris has the right to do with his IP what he will. If he doesn't want to share, it is within his right not to do so. Analogue makes a fine product and I will continue buying them. Mister in it's current state is about where Nesticle and other emulators were in the mid to late 90s. Immature and unfinished. That doesn't mean you can't enjoy and unfinished product though.

 

When you buy pro, you get quality. When you go free, you get a mixed bag.

[ijor]

I will elaborate about the technical issues of video lag later, just wanted to reply to this in the meantime ...

 

It doesn't have to be "0" lag necessarily, but 2 frames is just too much ... The only way to reduce lag on the MiSTer is to buy the IO board and run the analog out to an OSSC ...

 

Video lag is not exactly two frames, it oscillates between one and two frames, This is a direct consequence of using triple buffering at different refresh rates. You do have to add input lag as well, but this seems to be not as significant in comparison. I have no intentions to claim this is good or bad, just wanted to be accurate.

 

If you insist, you can reduce most of the video lag with the new MiSTer scaler. The option is still not enabled however, and you would need to recompile the core. Yeah, I can understand the pain, just saying it is possible if you insist. Again, will elaborate later about the technical details.

[SegaSnatcher]

I will elaborate about the technical issues of video lag later, just wanted to reply to this in the meantime ...

 

 

Video lag is not exactly two frames, it oscillates between one and two frames, This is a direct consequence of using triple buffering at different refresh rates. You do have to add input lag as well, but this seems to be not as significant in comparison. I have no intentions to claim this is good or bad, just wanted to be accurate.

 

If you insist, you can reduce most of the video lag with the new MiSTer scaler. The option is still not enabled however, and you would need to recompile the core. Yeah, I can understand the pain, just saying it is possible if you insist. Again, will elaborate later about the technical details.

 

Just curious, why are some people working on MiSTer so against the idea of doing the adjust to 60hz trick to not have to rely on a frame buffer and reduce lag? Now, I'm fully aware some consoles like 2600 and GB need a framebuffer, but for those that don't it seems pointless to drag them down with framebuffers.

[ijor]

Just curious, why are some people working on MiSTer so against the idea of doing the adjust to 60hz trick to not have to rely on a frame buffer and reduce lag? Now, I'm fully aware some consoles like 2600 and GB need a framebuffer, but for those that don't it seems pointless to drag them down with framebuffers.

 

Well, I have absolutely nothing against it, quite the contrary. The main reason that this is not implemented yet, is probably because Sorgelig doesn't give a sh* about lag. Right or wrong, he believes it is a non issue. And the current development towards lower latency is being pushed pretty much despite him.

 

The other reason is that MiSTer allows the user complete freedom to specify the video output parameters. You are not locked to a particular refresh rate or pixel clock. That is great because it is always better to use the natural resolution and refresh rate of your particular monitor. Otherwise the monitor might perform yet another scaler, and yet add more lag. This and other issues makes more difficult, but certainly not impossible, to implement the method you mention.

Edited December 3, 2018 by ijor

[Sho]

Well, Kevtris never owned a Genesis when it initially came out and only played a few games on it through the years which was mainly Sonic games, yet now he is knee deep in Mega SG development.

 

True enough! I guess all that matters really is skill involved, which is very high level for both people.

[Sho]

Sorry, but Kevtris has the right to do with his IP what he will. If he doesn't want to share, it is within his right not to do so. Analogue makes a fine product and I will continue buying them. Mister in it's current state is about where Nesticle and other emulators were in the mid to late 90s. Immature and unfinished. That doesn't mean you can't enjoy and unfinished product though.

 

When you buy pro, you get quality. When you go free, you get a mixed bag.

 

I was saying the exact same thing for the most part, lol.

 

Eventually the quality of the MiSTer might reach Analogue stuff for some of the more popular cores. There are also merits to having other cores that Kevtris may never get to, like the Arcade and some of the computers. It also seems like the MiSTer may never get to do things such as play games directly off of physical media and some of the more polished settings like music players and things like that. There's stuff for people on both sides. I think they can both co-exist and thrive in the community.

Edited December 3, 2018 by Sho

[Forsaken]

 

Do you think that's the case for the majority of games? . I mean obviously lag can affect some game's difficulty significantly like Battletoads, Punch-Out, and for speedrunning: Mario.. but, it doesn't really affect things like Shadowgate, Lolo, and Dragon Warrior, etc. But what about the in-between games? I mean does a couple frames lag really affect lower reflex games like Kirby, Zelda, Dr. Mario, or even Duck Tales for example?

Well if you don't play games where reaction time matters, then no, input lag doesn't matter for you. Input lag just slows your reaction time.

 

Variable input lag is the worst, as that will mess you up even if you try to anticipate and hit early, and will mess you up on rhythm type games.

[Newsdee]

There is a bit of history to explain some of what is seen that I think is worth mentioning.

I'd like to offer a bit of perspective from somebody who has been following FPGA emulation since 2013 or so, and also owns a few Analogue products including the original NT (non FPGA).

 

FPGA emulation is not new at all. People have been at it since the early 00s but it was held back by bad blood between some developers trying to make a quick buck, e.g. taking collaborative projects for their own commercial products or copycat implementations offering a partial/subpar experience to the original (I won't give names but one can find those by some googling). Some end products out of these were known and had moderate success, but people had no idea there was an FPGA project behind it (note: these are not Kevtris/Analogue products, just to be clear)

 

Then something changed... slowly some developers started making available completely open source cores for others to see and license them for reuse. Perhaps their idea was to sell hardware instead (i.e. it would have to run on a custom board made for it), such as the Minimig (Amiga clone), OneChipMSX (MSX clone), and the TC64 (C64 clone/expansion) to name a few. Other probably did it for "the community" since the core only worked on a development board (Torlus, MikeJ of FPGA Arcade, etc.).

 

In 2013 the MiST came out as an evolution of the idea because the FPGAArcade boards were not available (and more expensive). Till Harbaum came up with a PCB that was relatively cheap (much much cheaper than Terasic dev boards at the time), and that led to a small explosion of cores being ported to it. For a while cores weren't very compatible but worked well enough and had a flurry of activity.

 

MiSTer started with an idea from Sorgelig to take a (newly released) cheap Terasic board and port MiST to it. The newer and more powerful FPGA allowed cores that the MiST couldn't fit, and the board was easier to source.

One major driver for his idea was also to provide decent HDMI output; MiST provides VGA out but frequencies are often difficult to upscale since there is no framebuffer (btw that means the MiST has zero lag, but there are caveats of the tradeoff).

 

Basically the MiSTer is the current iteration and convergence of multiple FPGA core projects done in the past years by many different people.

From that perspective (among others) the work that Sorgelig has done is truly amazing as he had to unify a lot of disparate things, and now finally it is paying off by more developers collaborating in the core framework for it.

 

Kevtris/Analogue has also done something amazing but of a different nature: they came up with a viable commercial FPGA retro gaming product, which seemed impossible even back in 2013, at least to me.

Edited December 4, 2018 by Newsdee

[cacophony]

It doesn't have to be "0" lag necessarily, but 2 frames is just too much when most TVs inherently have 1 - 2 frames of lag already baked in. In its current form I personally don't see how MiSTer is superior to say a software emulation solution, especially with the run ahead latency feature. The only way to reduce lag on the MiSTer is to buy the IO board and run the analog out to an OSSC which is a $200 investment on top of the $130 DE-10 + however much the IO Board costs. Its great that you can still find enjoyment playing games with a lot of added latency, myself and some others simply can't.

 

idk, the thought of a tiny, silent, $160 device that can potentially outperform a giant, loud, expensive PC is pretty compelling to me but I guess everyone has different priorities. There are already a number of cores that are considered perfect, and many more will come over time. Run ahead is a neat idea, but even the devs of that feature consider it totally inaccurate.

Edited December 4, 2018 by cacophony

[Newsdee]

For those complaining about lag, you may want to consider a MiST. It has less cores but it outputs to a native resolution at the original frequencies and can be easily hooked up with a proper cable to a PVM or SCART TV.

It can also accept two Atari or Amiga-style joysticks by DB9 so there is no input lag whatsoever from the USB encoder. There is also the possibility of making a special cable for Component out (YPbPr) so an upscaler isn't needed.

 

The MiSTer, via the I/O board, can also offer the same outputs options.

Or just wait that people finish working on the new upscaler, it is progressing nicely. All that work will keep progressing as long as there is developer interest, which has been growing.

 

As to the comparison with an emulation board? Aside from the price/performance ratio (which is great), it's also much easier to setup since there is no need to bother with an OS.
I'm definitely getting much more use of this than say a Retro Freak (which feels laggy as hell in comparison) or the NES and SNES minis. RPis feel way too fiddly in comparison. But that's just my opinion; others may be fine with it.

Edited December 4, 2018 by Newsdee

[phoenixdownita]

You do can make a MiSTer like board with a smaller FPGA, A5 or A4, and most cores would still run fine. Probably almost all of them with an A5. There are even some ports to other similar Terasic boards with a smaller FPGA. But the FPGA device must be of the SOC type, having an Arm Cortex. You can't use a device from the Cyclone V E family regardless of the size.

 

In anycase, every core must be recompiled, even if the pinout is exactly the same.

They are cheaper but still no close to the subsidized price of the Terasic Dev Kit:

 

For the A4:

https://www.mouser.com/Semiconductors/Programmable-Logic-ICs/FPGA-Field-Programmable-Gate-Array/_/N-3oh9p?Keyword=5CSEBA4U23&FS=True

 

 

 

For the A5:

https://www.mouser.com/Semiconductors/Programmable-Logic-ICs/FPGA-Field-Programmable-Gate-Array/_/N-3oh9p?Keyword=5CSEBA5U23&FS=True

 

 

So as you can see it's neigh impossible even with just the cheapest Cyclone V SE A4 to likely build a whole MiSTer (once you include a custom PCB, USBs, HDMI [with licenses or can't use the connector, see the mess with the F18A mark II here on the forums], SD card, programming flash etc...etc...).

Note I am not even sure the cheapest SE A4 listed there with a slower 8 speed grade and single core ( 5CSEBA4U19C8SN ) makes the cut but let's assume it does, it alone is ~77 US$ + taxes .... the HDMI consortium asks for 10K for a yearly license plus royalties .... so unless the numbers go up and quickly it's not something one can commercially do just because.

The Terasic Dev Kit packs a punch in a very cheap solution, it's hard to beat it with even the smaller SE A4 level.

SE A5 starts at around the cost of the whole DE10-nano board ... so yeah really can't make it in a commercial setting and be competitive.

 

 

My argument is not against MiSTer (I keep on considering buying it), it's just to point out that if it gets popular the DE10-nano will either start to cost more or prohibit the usage in that setting (I believe they are already not usable in a commercial capacity, but I am not sure), either way nothing as cheap and commercially viable can come out of it for now (for better or for worse). There's a reason why the original MiST with a less powerful FPGA, no HDMI (and related due royalties) costed 200+US$.

 

So in the end it's not clear to me how much a "commercial version" (with say an ad-hoc board) would actually cost .... if it ever reaches that stage. I also believe that reaching that stage is necessary to its success which I truly wish for. In the end choice/competition has the consumer winning (within reason, a race to the bottom tends to result in crappy products nobody really wants but that is not the case here luckily).

[SegaSnatcher]

 

idk, the thought of a tiny, silent, $160 device that can potentially outperform a giant, loud, expensive PC is pretty compelling to me but I guess everyone has different priorities. There are already a number of cores that are considered perfect, and many more will come over time. Run ahead is a neat idea, but even the devs of that feature consider it totally inaccurate.

Thats the keyword, potentially. Yes, that could very well happen, but I'm thinking in the present. Trust me nobody wants to see MiSTer improve more than me. Once some of the console cores get a bit more mature and there is a low lag mode via HDMI out, I'll definitely be interested. I might still be interested if I end up buying an OSSC and then I could just use the analog out.

[RetrogamerX]

They sure do, I've witnessed it on the discord chat so far. I still am trying to figure out what I ever did to them to result in such negativity and hate from them.

 

Jealousy is a beautiful thing.

 

Who the hell cares what they think. You’re doing awesome work Kevtris.

[ijor]

The Terasic Dev Kit packs a punch in a very cheap solution, it's hard to beat it with even the smaller SE A4 level.

SE A5 starts at around the cost of the whole DE10-nano board ... so yeah really can't make it in a commercial setting and be competitive.

 

My argument is not against MiSTer (I keep on considering buying it), it's just to point out that if it gets popular the DE10-nano will either start to cost more or prohibit the usage in that setting (I believe they are already not usable in a commercial capacity, but I am not sure), either way nothing as cheap and commercially viable can come out of it for now (for better or for worse). There's a reason why the original MiST with a less powerful FPGA, no HDMI (and related due royalties) costed 200+US$.

 

So in the end it's not clear to me how much a "commercial version" (with say an ad-hoc board) would actually cost .... if it ever reaches that stage. I also believe that reaching that stage is necessary to its success which I truly wish for. In the end choice/competition has the consumer winning (within reason, a race to the bottom tends to result in crappy products nobody really wants but that is not the case here luckily).

 

Yeah, we know that it is impossible to compete with Terasic pricing. There are no restriction that I know for using the board for commercial purposes, unless of course you use the even lower pricing for students. I don't think the license or the pricing will change just because of the MiSTer. The only risk is that at some point Intel might decide to end the University Program and will stop providing subsidized devices to Terasic. Doesn't seem this will happen anytime soon though.

 

But I agree that a commercial version built from scratch, without using a Terasic board, is very unlikely. Somebody perhaps would sell the whole combo packed on a nice case, that's all, I'm afraid.

[Andromeda Stardust]

Man... I wanna get this lol.

 

https://www.analogue.co/editions/ghostly-super-nt/

 

Just for the sake of it lol, Ordering it from Ghostly directly, shipping was 12.28.

 

Now I'll have to sell some stuff because I have a black one (being used), sealed US color scheme, and now this one. /shrugs

I got the email yesterday; looks quite badass. However I would question the merits of buying this model over the standard flare Super NT if you already own one, unless you're buying it for collectability or the "cool" factor.

 

One thing that concerns me however: will the custom boot screens get overwritten if the firmware is upgraded or replaced? Because I imagine fans would be disappointed if this were the case.

 

If the stock firmware does not overwrite the custom boot screen, then the firmware upgrade would have to have the Ghostly boot code already present (assuming it overwrites everything in the console), and some hardware identifier would determine the console type and resultant boot screen.

 

Thus, it may be possible to hack a standard issue Super NT to display the Ghostly boot menus.

[Andromeda Stardust]

That’s the real problem down the road, IMO. Will HDMI get deprecated 20 years from now and replaced with some crazy new thing?

 

That said, a good OSS core will at least be a usable starting point for anything new on that front, though. Once you have a good, well-debugged core, you can focus more on how the PPU outputs are being handled and less on fixing the various esoteric edge cases that devs relied on for their particular games.

I don't think we have to worry about that much. While the transition from analog to digital had some pretty heavy growing pains, most digital standards maintain backwards compatibility. For instance, Super Audio CD players played CDs. BluRay players play DVDs, and in fact require a dual laser assembly with both a 650nm red laser (for DVDs and CDs) and a 405nm Blue/Violet laser for reading BluRay discs. 4k or UHD BluRay players play DVDs and BluRay as well.

 

Passive adapters exist to easily and painlessly convert HDMI into other standards such as DVI and Display port. USB3 ports can function as HDMI outputs as well.

 

So I think you will find that legacy digital standards such as HDMI will be supported for a very long time to come, even if the form factor of the plug changes.

[Andromeda Stardust]

Well, Kevtris never owned a Genesis when it initially came out and only played a few games on it through the years which was mainly Sonic games, yet now he is knee deep in Mega SG development.

When did he say this specifically?

[cacophony]

Thats the keyword, potentially. Yes, that could very well happen, but I'm thinking in the present. Trust me nobody wants to see MiSTer improve more than me. Once some of the console cores get a bit more mature and there is a low lag mode via HDMI out, I'll definitely be interested. I might still be interested if I end up buying an OSSC and then I could just use the analog out.

 

Many of the cores are already superior to the best software emulation at this point, it's just a matter of what cores are important. But tbh I've been enjoying pretty all the console cores. Just knowing it's hardware emulation is a plus to me. Might be purely psychological but it feels more authentic

Edited December 4, 2018 by cacophony

[Jurai]

Look at the bright side guys, we'll all be long dead before playing our consoles becomes a *real* problem

[Jurai]

I got the email yesterday; looks quite badass. However I would question the merits of buying this model over the standard flare Super NT if you already own one, unless you're buying it for collectability or the "cool" factor.

 

One thing that concerns me however: will the custom boot screens get overwritten if the firmware is upgraded or replaced? Because I imagine fans would be disappointed if this were the case.

 

If the stock firmware does not overwrite the custom boot screen, then the firmware upgrade would have to have the Ghostly boot code already present (assuming it overwrites everything in the console), and some hardware identifier would determine the console type and resultant boot screen.

 

Thus, it may be possible to hack a standard issue Super NT to display the Ghostly boot menus.

 

I like the artists that ghostly is using, but since I got my sd2snes I honestly just boot straight to cart now

[Andromeda Stardust]

They are cheaper but still no close to the subsidized price of the Terasic Dev Kit:

 

So as you can see it's neigh impossible even with just the cheapest Cyclone V SE A4 to likely build a whole MiSTer (once you include a custom PCB, USBs, HDMI [with licenses or can't use the connector, see the mess with the F18A mark II here on the forums], SD card, programming flash etc...etc...).

Note I am not even sure the cheapest SE A4 listed there with a slower 8 speed grade and single core ( 5CSEBA4U19C8SN ) makes the cut but let's assume it does, it alone is ~77 US$ + taxes

the Super NT and upcoming Mega SG both run on a Cyclone V, do they not? How does the included FPGA compare pricewise?

HDMI consortium asks for 10K for a yearly license plus royalties .... so unless the numbers go up and quickly it's not something one can commercially do just because.

So exactly how does Analogue, RetroUSB get around using HDMI on their devices? Last I checked, anyone can go to Mouser or Digikey and buy HDMI connectors in bulk.

 

Maybe getting the HDCP license is only to use the HDMI logo (which isn't featured on the AVS or Super NT AFAIK) or to generate and/or display protected video streams requires the ten grand per year?

 

Or if the FPGA already has the HDMI driver built into it, or some other IC chip within the console has a built in encoder, than the chip manufacturer has already paid the HDMI license.

 

I really don't know if the license can apply to a circuit on the device and not the device itself. But there is no general protection against using HDMI ports for other uses which do not meet the standard.

 

One company did a kickstarter for a universal controller converter, to allow to plug everything into a PC. The pentagon shaped dongle had a USB on one side and four HDMI ports that plugged into custom controller adapters.

[SegaSnatcher]

When did he say this specifically?

Are you not on the Discord? He told people this a bunch of times on the Classic Gaming channel.

 

"and I never played the Genesis because I never owned one. I had an SNES."

Edited December 4, 2018 by SegaSnatcher

[SegaSnatcher]

 

Many of the cores are already superior to the best software emulation at this point, it's just a matter of what cores are important. But tbh I've been enjoying pretty all the console cores. Just knowing it's hardware emulation is a plus to me. Might be purely psychological but it feels more authentic

From my understanding only some of the PC cores might be. None of the console cores are better in terms of accuracy than the best software emulators.

[kevtris]

When did he say this specifically?

He is right, I never had a genesis. I had a SNES though. Friend and I swapped systems for a month or two; I got his genesis+sonic and he got my SNES+super mario world. Up until this year, about the only things I played on it were sonic games.