MiSTer FPGA board

[Syntax]

It does work natively fine. But a permanent hub for the MiSTer or Pi3 would be nice.

[orange808]

Been whispering about this in private messages for months now. Lots of people agree; it's not just me.

Does anybody else think sorgelig's abrasive and dismissive attitude is hurting MISTer?

[maxtherabbit]

Been whispering about this in private messages for months now. Lots of people agree; it's not just me.

Does anybody else think sorgelig's abrasive and dismissive attitude is hurting MISTer?

He's the one who has produced the vast majority of the project, so he can have any attitude he wants IMO

[orange808]

Been whispering about this in private messages for months now. Lots of people agree; it's not just me.

Does anybody else think sorgelig's abrasive and dismissive attitude is hurting MISTer?

He's the one who has produced the vast majority of the project, so he can have any attitude he wants IMO

Yep. He sure did.

Did you see how he greeted marqs today?

[maxtherabbit]

Been whispering about this in private messages for months now. Lots of people agree; it's not just me.

Does anybody else think sorgelig's abrasive and dismissive attitude is hurting MISTer?

He's the one who has produced the vast majority of the project, so he can have any attitude he wants IMO

Yep. He sure did.

Did you see how he greeted marqs today?

I did not. I'm not exactly trying to go to bat for a shitty attitude, just pointing out the absurdity of saying he's hurting the project when he essentially is the project

[Wolf_]

Yea he has this thing where he gets super upset if anyone mentions improving the scaler lag so it is no wonder he got upset at marqs. He seems to think lag is necessary and just pumping the native clock&refresh rates out like the ossc does is blasphemy. Personally I think the project would be a lot better off if he only did updates and someone else like SmokeMonster had control of the main build and what direction it takes.

[orange808]

We got the same reaction to LL Cool Joy.

[rama]

I recommend accepting the designers decision on the HDMI scaler.
For one, the project is open. I'm sure it can be customized by others to provide the lesser lag options.
Second, the project has a lag-free high quality analog output.
This can be digitized by low lag (scanline lag) solutions such as many $5 VGA to HDMI adapters.

If the pressure keeps up on the developer to change the design he firmly decided on, I can easily see him loosing interest in it all.
It would be great if people could just respectfully disagree with him. This is the basis for constructive improvement.
Don't haggle about how many frames are acceptable lag, and certainly don't go into arguments over the topic.
It's enough to state that you're looking for least lag over things like tearing, and then we can proceed in accomplishing it.

[Kez]

the absurdity of saying he's hurting the project when he essentially is the project

It's not that much of a reach to say being unwelcoming when someone knowledgeable tries to get involved is hurtful to the project. Sorge certainly does a huge amount of work on the project, but a lot of work is also being done by many other people.. not to mention much of Sorge's work porting the cores builds upon work done by yet more people.

If the pressure keeps up on the developer to change the design he firmly decided on, I can easily see him loosing interest in it all.
It would be great if people could just respectfully disagree with him. This is the basis for constructive improvement.

In general his reactions to any feedback/comments that are not aligned with his personal philosophy are disinterest or occasionally antagonism, however when people actually make something that works well he merges it with the main branch without any fuss. I certainly haven't heard of anyone's work being turned away, even if his previous comments on the issue have been negative.

[Galgomite]

If I can change the subject back to the MiSTer itself— FX chip games now play on the snes core! That’s a feature that will probably never appear on the Super NT.

[Wolf_]

If I can change the subject back to the MiSTer itself— FX chip games now play on the snes core! That’s a feature that will probably never appear on the Super NT.

Definitely never. It doesn't have the power.

[sparksterz]

That’s a feature that will probably never appear on the Super NT.

But why would it? Nintendo's design decision was to have expansion chips in the carts. Not to have the console support them. If the Super NT and Mega SG are representations of the console hardware itself then I think having the console emulate the needs of a cartridge is not really part of the design philosophy. That's why the SD2SNES emulates the expansion chips in its FPGA.

[Wolf_]

That’s a feature that will probably never appear on the Super NT.

But why would it? Nintendo's design decision was to have expansion chips in the carts. Not to have the console support them. If the Super NT and Mega SG are representations of the console hardware itself then I think having the console emulate the needs of a cartridge is not really part of the design philosophy. That's why the SD2SNES emulates the expansion chips in its FPGA.

I think you're a bit confused. Nintendo put expansion chips in carts so they could release games that were more demanding than the technology at the time they released the snes allowed. This way they could have more advanced games than their competitors on the current generation. The sd2snes has effect chip support on the cart because it is designed to be used with a snes console so if it didn't support the chips then the console couldn't run them. Obviously if you are playing the actual cart it will have the effect chips but if you don't and still want to play the game then you need to have it supported via something else.

1080p hdmi output isn't part of the design philosophy of the original snes any more than effect chip support is and the carts won't last forever (Thankfully the sd2snes should last a lot longer) so having a way to still play those games in a hardware accurate way is pretty important. Of course if Analogue actually gave a crap about preservation they would port all of Kevtris' other cores over to a console they actually still sell and not just let them rot on copies of the nt mini being scalped on ebay so I wouldn't expect them to fuss about effect chip support. (especially when the fpga they released the snes on isn't powerful enough for some of them).

[sparksterz]

I think you're a bit confused. Nintendo put expansion chips in carts so they could release games that were more demanding than the technology at the time they released the snes allowed. This way they could have more advanced games than their competitors on the current generation. The sd2snes has effect chip support on the cart because it is designed to be used with a snes console so if it didn't support the chips then the console couldn't run them. Obviously if you are playing the actual cart it will have the effect chips but if you don't and still want to play the game then you need to have it supported via something else.

1080p hdmi output isn't part of the design philosophy of the original snes any more than effect chip support is and the carts won't last forever (Thankfully the sd2snes should last a lot longer) so having a way to still play those games in a hardware accurate way is pretty important. Of course if Analogue actually gave a crap about preservation they would port all of Kevtris' other cores over to a console they actually still sell and not just let them rot on copies of the nt mini being scalped on ebay so I wouldn't expect them to fuss about effect chip support. (especially when the fpga they released the snes on isn't powerful enough for some of them).

I don't think I relayed my point properly. That's the exact point I was trying to make. The expansion chips are the responsibility of the game in this case. Just like Virtua Racing on the Genesis using the SVP chip. I wouldn't expect that the Mega SG would work with just the rom as the rom is dependent on an implementation of that chip which in this case lived in the cartridge of that game.

Now of course when you're talking a MiSTer core, I totally understand...there's nothing else to take on that responsibility if the main FPGA doesn't have the overhead to support it as well as run the system.

[Wolf_]

I think you're a bit confused. Nintendo put expansion chips in carts so they could release games that were more demanding than the technology at the time they released the snes allowed. This way they could have more advanced games than their competitors on the current generation. The sd2snes has effect chip support on the cart because it is designed to be used with a snes console so if it didn't support the chips then the console couldn't run them. Obviously if you are playing the actual cart it will have the effect chips but if you don't and still want to play the game then you need to have it supported via something else.

1080p hdmi output isn't part of the design philosophy of the original snes any more than effect chip support is and the carts won't last forever (Thankfully the sd2snes should last a lot longer) so having a way to still play those games in a hardware accurate way is pretty important. Of course if Analogue actually gave a crap about preservation they would port all of Kevtris' other cores over to a console they actually still sell and not just let them rot on copies of the nt mini being scalped on ebay so I wouldn't expect them to fuss about effect chip support. (especially when the fpga they released the snes on isn't powerful enough for some of them).

I don't think I relayed my point properly. That's the exact point I was trying to make. The expansion chips are the responsibility of the game in this case. Just like Virtua Racing on the Genesis using the SVP chip. I wouldn't expect that the Mega SG would work with just the rom as the rom is dependent on an implementation of that chip which in this case lived in the cartridge of that game.

Now of course when you're talking a MiSTer core, I totally understand...there's nothing else to take on that responsibility if the main FPGA doesn't have the overhead to support it as well as run the system.

People are working on the ability to add cart support to the MiSTer right now but the issue isn't carts it is what do people do when those carts die or are stupidly expensive? Analogue claims to be preservationists but they don't offer support for running the roms of effect chip games or port their existing cores to anything they are still currently selling. It's a sickening waste of quality fpga cores tbh.

[sparksterz]

People are working on the ability to add cart support to the MiSTer right now but the issue isn't carts it is what do people do when those carts die or are stupidly expensive? Analogue claims to be preservationists but they don't offer support for running the roms of effect chip games or port their existing cores to anything they are still currently selling. It's a sickening waste of quality fpga cores tbh.

I mean, those consoles are intended to run the carts though...I guess my point is just at odds with the fact that you want their devices to do more than they currently do so I'll drop it, we don't have to agree.

Regardless I'm glad to see MiSTer cores providing implementations of these so that using a combination of FPGA in cart form and system form is no longer necessary.

[thrasherx]

Looks like there's some good discussion over here. I run https://www.misteraddons.com and will be popping by from time to time to say hi and hopefully add to the conversation.

[James-F]

Bought my stuff from you thrasherx on misteraddons.com, great service.

[ASDR]

Hm, the 2 frame lag seems really bad. For me the main two points of FPGA clone consoles are electric compatibility with original accessories (controllers, carts, other add-ons) and latency. Since there aren't exactly SNES controller and cartridge ports on the machine and latency is worse than SNES+OSSC (or even FM with its 1 1/2 frames latency), kind of a bummer. I'd personally always prefer the option to just nudge the clock of the core a bit till the output is standard 60Hz or 59.94Hz or whatever the display wants.

[Wolf_]

Hm, the 2 frame lag seems really bad. For me the main two points of FPGA clone consoles are electric compatibility with original accessories (controllers, carts, other add-ons) and latency. Since there aren't exactly SNES controller and cartridge ports on the machine and latency is worse than SNES+OSSC (or even FM with its 1 1/2 frames latency), kind of a bummer. I'd personally always prefer the option to just nudge the clock of the core a bit till the output is standard 60Hz or 59.94Hz or whatever the display wants.

Well the LL board is in development and nearing completion. So when that gets released you will be able to use basically every controller with the MiSTer natively without lag.

From there I imagine one day it will have the option of either using the native refresh rate out with ossc levels of lag or an adjusted clock rate for 0 lag but with the timing slightly tweaked.

[James-F]

Hm, the 2 frame lag seems really bad. For me the main two points of FPGA clone consoles are electric compatibility with original accessories (controllers, carts, other add-ons) and latency. Since there aren't exactly SNES controller and cartridge ports on the machine and latency is worse than SNES+OSSC (or even FM with its 1 1/2 frames latency), kind of a bummer. I'd personally always prefer the option to just nudge the clock of the core a bit till the output is standard 60Hz or 59.94Hz or whatever the display wants.

The default option for HDMI scaler is triple buffered vsync hence 2 framas of lag, but there is a new option for zero lag which is work in progress and almost complete.
Since the MiSTer supports many core unlike SuperNT, it will not have an exact 60.00Hz option for each and every core.

[ASDR]

Sure, seems like the hardware would be capable of most of this and everything is of course very actively developed. I was just thinking out loud in terms of "could this replace my OSSC + SFC/MD1/PCE today".

Why doesn't every core have the option to just output 60Hz? I know very little about hardware development - would this literally just be tweaking a clock speed to get the video refresh to a standard value or does it involve a whole bunch of development work to get all the timing/signals/synchornization etc. changed?

[paulb_nl]

You can change the clock speed to change the refresh rate to 60Hz but there is no benefit in doing so unless your display does not support the original refresh rate at all.

It makes no difference for lag.

[orange808]

Sure, seems like the hardware would be capable of most of this and everything is of course very actively developed. I was just thinking out loud in terms of "could this replace my OSSC + SFC/MD1/PCE today".

Why doesn't every core have the option to just output 60Hz? I know very little about hardware development - would this literally just be tweaking a clock speed to get the video refresh to a standard value or does it involve a whole bunch of development work to get all the timing/signals/synchornization etc. changed?

There are lots of different visions of what it should be.

Personally, I don't want that from an FPGA console. I want a real replacement for my hardware.

I want an FPGA system that reproduces the frame rate the consoles put out. I don't want it underclocked. I don't want it overclocked. I don't want frame rate conversion.

I want the real system video output available over analog and HDMI. Scaling over HDMI is fine, but I want authentic frame rate.

Some kind of compatibility features are okay, but I don't want or need them.

[ASDR]

You can change the clock speed to change the refresh rate to 60Hz but there is no benefit in doing so unless your display does not support the original refresh rate at all.

It makes no difference for lag.

But at least currently the scaler in the MiSTer has the 2-frame lag for framerate conversion, right?

There are lots of different visions of what it should be.

Personally, I don't want that from an FPGA console. I want a real replacement for my hardware.

I want an FPGA system that reproduces the frame rate the consoles put out. I don't want it underclocked. I don't want it overclocked. I don't want frame rate conversion.

I want the real system video output available over analog and HDMI. Scaling over HDMI is fine, but I want authentic frame rate.

Some kind of compatibility features are okay, but I don't want or need them.

I'd also much rather have displays stop being pieces of shit and just deal with any refresh rate I throw it them. But if an odd refresh rate causes no-signal, tearing or lag inducing frame rate conversion than I'd much rather nudge the core to 60Hz. I'll notice the former issues but I doubt I'd ever be able to tell the 0.3Hz or whatever speed difference.

[paulb_nl]

But at least currently the scaler in the MiSTer has the 2-frame lag for framerate conversion, right?

Yes the default setting is triple buffer to 60Hz for compatibility. Even if you change the core to run at 60Hz there is still a buffer required to adjust to standard HDMI timings. That is what the setting vsync_adjust=2 does. It outputs at original refresh rate and uses a single buffer for low lag. So unless your display does not accept refresh rates other than exact 60Hz there is no need to change the refresh rate.

There is still tearing sometimes with the low lag setting so it is not quite ready yet.

[thrasherx]

Honored to be part of a Racketboy article. Did anyone else see this? http://www.racketboy.com/retro/mister-f ... eservation

[orange808]

On one hand, MISTer is fantastic. On the other hand, the hype train is starting again. This is getting out of control.

Is FPGA the future? Probably. Is it ready for prime time? No. Is MISTer powerful enough to deliver the level of emulation you believe is happening under the hood? No.

It's emulation. It's not a real transistor level reproduction of the real hardware. It's still using hacks to cover a variety of behaviors that aren't organically produced from a deep reproduction of the original hardware. It's still emulation.

Everyone flamed byuu, but his core points were all true.
http://archive.li/ypRuv

It's all true.

We don't even have a proper transistor level 6502 running on expensive FPGA hardware. If you like to tinker, I'll provide a link. Please lurk if you can't contribute to development. I'm not trying to upset the noise to signal there.
http://forum.6502.org/viewforum.php?f=1 ... fbc6255f24

Furthermore, I now prefer WinUAE to minimig on MISTer for Amiga. Beam racing makes all the difference. So, I'm choosing the software emulation there. At best, minimig can only save me a couple milliseconds of latency and FPGA isn't inherently more accurate. Cycle accuracy is about timing--not accurate reproduction of the hardware behavior. WinUAE manages to tackle the obstacles that Windows creates very well--and the software emulation can compete with any and all emulation solutions.

MISTer is fantastic, I love my MISTer, and I plan to keep it around for a long time. But, we're really hyping this.

The first true reproduction of a machine in FPGA will probably be the Atari 2600 and even that is a long road. That will be multiple expensive FPGA's "wired" together.

I love MISTer, but the hype is getting bigger than the reality.

[Wolf_]

On one hand, MISTer is fantastic. On the other hand, the hype train is starting again. This is getting out of control.

Is FPGA the future? Probably. Is it ready for prime time? No. Is MISTer powerful enough to deliver the level of emulation you believe is happening under the hood? No.

It's emulation. It's not a real transistor level reproduction of the real hardware. It's still using hacks to cover a variety of behaviors that aren't organically produced from a deep reproduction of the original hardware. It's still emulation.

Everyone flamed byuu, but his core points were all true.
http://archive.li/ypRuv

It's all true.

We don't even have a proper transistor level 6502 running on expensive FPGA hardware. If you like to tinker, I'll provide a link. Please lurk if you can't contribute to development. I'm not trying to upset the noise to signal there.
http://forum.6502.org/viewforum.php?f=1 ... fbc6255f24

Furthermore, I now prefer WinUAE to minimig on MISTer for Amiga. Beam racing makes all the difference. So, I'm choosing the software emulation there. At best, minimig can only save me a couple milliseconds of latency and FPGA isn't inherently more accurate. Cycle accuracy is about timing--not accurate reproduction of the hardware behavior. WinUAE manages to tackle the obstacles that Windows creates very well--and the software emulation can compete with any and all emulation solutions.

MISTer is fantastic, I love my MISTer, and I plan to keep it around for a long time. But, we're really hyping this.

The first true reproduction of a machine in FPGA will probably be the Atari 2600 and even that is a long road. That will be multiple expensive FPGA's "wired" together.

I love MISTer, but the hype is getting bigger than the reality.

There is a difference between hardware simulation and emulation. You are correct that one is not more inherently accurate than the other, however if you get the hardware simulated well enough (which also takes considerably less resources to do than to emulate it) then you can run the original code and have near perfect accuracy which has already been achieved for multiple fpga systems.

I agree the hype train is looking pretty far down the road talking about things like the Sega Saturn but there are many new systems I could see coming out in the next couple months that people would freak out about like the SNES enhancement chips, Neo Geo, TG-CD, and Game Boy Advance. And in a year or two we'll probably see the PS1 covered. That's all well deserved hype imo.

Edit: Not to mention the arcade cores in the works, some of which on their own cost well over 5x the price of a MiSTer.

[orange808]

WinUAE exists. It's just not getting hyped on YouTube.

So, your argument about more development (on a given hardware platform) eventually matching achievements of other projects (on the same platform) and reproducing real hardware behaviors can be applied directly to software emulation as well.