Next Generation FPGA Hardware's Potential for Existing MiSTer Cores?

[KremlingKuthroat19]

About once a quarter or so, everyone talks about the next FPGA platform and what cores we can get for new and more advanced board. We know that a future board will likely cover the N64 and DS and that any hardware beyond that could be a bit difficult.

However, this thread's purpose is to discuss how existing cores could benefit from new hardware. Let's say for sake of discussion that a new board was 2-3 times more powerful than the current MiSTer at 200-300K LEs and 2-3x the memory and bandwidth. How would existing cores benefit? Here are a few thoughts:

Sega Genesis/Sega CD and 32X could all be merged into one core as the infamous "tower of power". This merging would be beneficial because the Sega CD/32X exclusive games would be playable. The current MiSTer could fit potentially 32X and Genesis into one core, but the Sega CD wouldn't fit. Also, there's not enough memory I believe for the Sega CD/32X games to run on MiSTer, so these games could be added.

The PSX core could have correct perspective on textures. Don't want to put words in his mouth, but I also believe Robert said that 2x resolution could be possible on a bigger board and that increasing the resolution 2x would cost the current MiSTer 125% of its logic gates. If there's a board twice as big, I suppose we could get resolutions up to 480p or 720p, but I'm just speculating so correct me if I'm way off with this even being possible.

The Game Boy and Game Boy Advance 2P cores could probably just fit in their original cores so they could be merged back into 2 cores instead of 4. Also, I wonder if we could have 4P modes for each of these cores since the Game Boy has a 4P adapter, even though it only supports about 10 games, and the GBA has many 4P games such as Mario Kart Super Circuit, Doom 2, etc. That'd be crazy to have 4P support on GBA if possible.

Another thing that'd benefit tremendously is save states. Save states take up a ton of logic, so adding them to cores that are pushing the FPGA limits currently that don't have save states such as the SNES, Genesis, and TurboGrafx cores could be a lot more feasible since space will be less of a problem, but I don't want to make adding save states to existing cores sound trivial.

I mostly play console games, but these are my ideas of how existing console cores could benefit from advanced FPGA hardware. Any ideas from you guys about console, computer or arcade cores benefitting or am I way off base?

[arromdee]

Voodoo graphics in ao486?

[aberu]

The the ST-V arcade hardware would be possible on a new platform like you are describing. Cave1k. Big advancements for the ao486 core could happen. A big increase in BRAM and onboard SRAM would be a welcome upgrade to help support these sorts of things.

One day I suppose, but the de10-nano is more than sufficient for our needs for now personally.

[dmckean]

Voodoo graphics in ao486?

No. You really need a Pentium for all those games that support Voodoo graphics.

[pgimeno]

Maybe not Voodoo, but real OPL3, a 487 coprocessor and some outstanding bugfixes that are now unfixable would become possible for ao486. Perhaps some more advanced arcades would become possible too.

The DS maybe, the N64 definitely not. This has been said in other threads.

[KremlingKuthroat19]

The point of this thread is to discuss how existing cores can benefit from new hardware. It's not about requesting new cores. There's plenty of other threads about that

[Bas]

Minimig PPC? One can dream right?

[pgimeno]

Not sure if that would be possible with a better CPU in the SoC and hybrid emulation.

[kolla]

Stuff I sort of miss…

- 68882 for TG68 cores
- 56k DSP
- Minimig RTA (AHI)
- virtual ethernet connection in the core to the Linux side, not just serial.
- codec for mpeg-1, mp3, mpeg-2/h262, mpeg-4/h264 etc (MHI, mpeg.library mm)
- crypto/TLS accellerator (speed up AmiSSL)

[Mirest]

It is possible to capture static screenshots across all cores on the DE10-Nano, though, MiSTer is missing a similar simple way to capture/record video and sound without external gadgetry. Screen recording is common on modern video game consoles, mobile devices, and desktop computers through various software. Bringing some or all of that capability to this platform (possibly on different FPGA hardware) may expand the audience or increase involvement, while providing a more enjoyable and comfortable experience for those with other machines who wish to easily share here and elsewhere in a somewhat familiar yet still unique way. Even if it would be limited by time or size per recording, a feature such as this could also serve as a bug fixing tool for developers.

[Nioreh]

System link games on PSX with 2 instances running simultaneously.

[AmanoJacu]

The the ST-V arcade hardware would be possible on a new platform like you are describing. Cave1k. Big advancements for the ao486 core could happen. A big increase in BRAM and onboard SRAM would be a welcome upgrade to help support these sorts of things.

One day I suppose, but the de10-nano is more than sufficient for our needs for now personally.

With Saturn being possible on the de10, I thought the ST-V would be as well, is it not?

[KremlingKuthroat19]

The the ST-V arcade hardware would be possible on a new platform like you are describing. Cave1k. Big advancements for the ao486 core could happen. A big increase in BRAM and onboard SRAM would be a welcome upgrade to help support these sorts of things.

One day I suppose, but the de10-nano is more than sufficient for our needs for now personally.

With Saturn being possible on the de10, I thought the ST-V would be as well, is it not?

The ST-V is an almost identical copy of a Saturn with a cartridge slot as opposed to a CD drive. Since CD drives themselves take up a ton of logic, the ST-V would likely be an easier fit on the De-10 than the Saturn itself from what I've heard.

[dmckean]

The the ST-V arcade hardware would be possible on a new platform like you are describing. Cave1k. Big advancements for the ao486 core could happen. A big increase in BRAM and onboard SRAM would be a welcome upgrade to help support these sorts of things.

One day I suppose, but the de10-nano is more than sufficient for our needs for now personally.

With Saturn being possible on the de10, I thought the ST-V would be as well, is it not?

The ST-V is an almost identical copy of a Saturn with a cartridge slot as opposed to a CD drive. Since CD drives themselves take up a ton of logic, the ST-V would likely be an easier fit on the De-10 than the Saturn itself from what I've heard.

It's another fast memory bus on a core that's already starved for bandwidth even when utilizing dual ram.

[aberu]

You know what, I was mixing up ST-V with the Sega Model 2, my bad. The Sega Model 2 is almost certainly not possible on the DE10-Nano, ST-V is a distant maybe due to bandwidth concerns, but might be possible if a lot of unneeded pieces of the saturn hardware are cut out and the core is slimmed down in the future, or optimized. Who knows on that one.

Dual SDRAM actually stresses timings a bit more, but whether or not that is a problem depends on the constraints of what you are working on.

[Newsdee]

MiSTer is missing a similar simple way to capture/record video and sound without external gadgetry.

That's much harder than doing screenshots, because latency and storage speed will be stressed and it might not work well.

It's better to use an external device that can capture HDMI or VGA output and acurately reflect the native speed of MiSTer.

[Mirest]

Another thing that'd benefit tremendously is save states. Save states take up a ton of logic, so adding them to cores that are pushing the FPGA limits currently that don't have save states such as the SNES, Genesis, and TurboGrafx cores could be a lot more feasible since space will be less of a problem, but I don't want to make adding save states to existing cores sound trivial.

If there is want for it, and technological allowance, core developers could pair save states with gameplay rewind and fast-forward. Great for speedrunning practice.

[throAU]

Maybe not Voodoo, but real OPL3, a 487 coprocessor and some outstanding bugfixes that are now unfixable would become possible for ao486. Perhaps some more advanced arcades would become possible too.

The DS maybe, the N64 definitely not. This has been said in other threads.

Most games that require/benefit from an FPU are Pentium based. About the only game that ran reasonably on a 486 that even used FPU was Falcon 3.0 with high fidelity flight model turned on.

For gaming, little point enabling hardware FPU without both the cpu and FPU being of Pentium class.

And for that era also having a GPU.

A big step for the pc core(s). The next tier of hardware for gaming probably worthwhile is something like Pentium 100-200/voodoo

[Malor]

They might be able to extend the ST core to cover Falcon hardware, but doing the DSP would be a huge job if nobody's done it yet.

As others have said, it would be a huge blessing for the AO486 core, which has a bunch of unfixable bugs because it's too large. They could probably make it work with many more games.

[Longtime4321]

I love the idea of an FPGA N64! Emulation for the platform has never been good enough to run every game, so an fpga n64 would be a godsend! That being said I will not fund or support any new MiST project because I just built my mister this year :p

Edit: haha this thread is for existing cores. Anyways, yeah the Genesis tower of power is certainly a plus and I would really like to do 4 player FPGA gaming.

[Mirest]

Optional (via OSD) custom or official active/animated or passive/static bezels, overlays, and underlays could be possible.

[thisisamigaspeaking]

I was looking at an SDR with a 301K LE Cyclone V and thinking wouldn't it be nice to have that chip for MiSTer. The SDR board is $850 though, which is actually a pretty fair price for it I think.

[Malor]

From what I've gathered over the last few months, it sounds like the switch to the MiSTer board (away from MiST) was to get access to the high-speed HDMI scaler the Nano offers. Any next generation board would need a scaler at least as good, or it might end up being a pain in the butt like the MiST was, with multiple output methods and having to manually switch all the time.

301K LE or not, if the scaler sucks, I think Sorgelig wouldn't be interested even if it was cheap.

[grizzly]

Not only the scaler, the ARM cpu that controls USB/SD card/network is a big part too.
That save much of the FPGA to what the FPGA does best.

[thisisamigaspeaking]

From what I've gathered over the last few months, it sounds like the switch to the MiSTer board (away from MiST) was to get access to the high-speed HDMI scaler the Nano offers. Any next generation board would need a scaler at least as good, or it might end up being a pain in the butt like the MiST was, with multiple output methods and having to manually switch all the time.

301K LE or not, if the scaler sucks, I think Sorgelig wouldn't be interested even if it was cheap.

Not saying we should use an SDR for a MiSTer... (sorry if this wasn't addressed at my prior post)

[Jae099]

System link games on PSX with 2 instances running simultaneously.

Great idea!!!!!

[dmckean]

Quad everything!!

[jca]

I just saw this talk about the scaler just now. It seems there is some confusion about the scaler.
The scaler is written in HDL and is part of the framework. It does not really have anything to do with the Nano besides the fact that HDMI is directly connected to the FPGA and that the nano has enough LEs to implement the scaler alongside the rest of the cores