MiSTer FPGA Forum

Woule be nice. Is it possible to make?

FM Towns is possible, GameCube is not.

Does nobody want fm towns too? Its an intersting system like the sharp x68000

FM Towns is interesting but is pretty niche, the X68000 core still has a ways to go too. Unfortunately there are only a finite number of developers with finite resources and esoteric Japanese systems require a lot of research and hands on with the hardware.

Whilst I'm sure a lot of people here would love FM Towns and Towns Marty it's a massive undertaking.

6th generation hardware (Gamecube, PS2, Dreamcast etc.) are out of the scope of the DE10-Nano (not enough logic elements/memory bandwidth).

Unfortunately, even with the logic elements, the GameCube is probably too complex for hobbyists to implement. During the 90s, CPUs implemented two massive leaps in design: superscalar pipelines and out-of-order execution. These advances let the CPU stay a lot busier, and therefore get more work done, even at the same clock speed. The GameCube CPU implemented both, for comparison, the PS1 and N64 CPUs are in-order and use far shorter pipelines.

The downside is that they’re far more complex than older in-order CPUs, and/or those with shorter pipelines. That complexity means that, even if you had an infinitely large FPGA, it’s still going to take a lot longer to design the core. It’d be a full-time job for a team of developers with a valuable skill set.

Lowe0 wrote: ↑Tue Oct 22, 2024 3:05 pm

Unfortunately, even with the logic elements, the GameCube is probably too complex for hobbyists to implement. During the 90s, CPUs implemented two massive leaps in design: superscalar pipelines and out-of-order execution. These advances let the CPU stay a lot busier, and therefore get more work done, even at the same clock speed. The GameCube CPU implemented both, for comparison, the PS1 and N64 CPUs are in-order and use far shorter pipelines.

The downside is that they’re far more complex than older in-order CPUs, and/or those with shorter pipelines. That complexity means that, even if you had an infinitely large FPGA, it’s still going to take a lot longer to design the core. It’d be a full-time job for a team of developers with a valuable skill set.

Preach, brother. I've been trying to explain why the PS2 isn't happening for exactly the same reasons for years.

Plus once the consoles use frame buffers the advantages of FPGA evaporate. Gamecube is near perfect on Dolphin.

??? There already is an FM Towns core?

viewtopic.php?t=104&start=60

Very early though. Not really usable yet.

Work has never stopped, the latest build still no good? FMTowns_20240409.rbf

FPGA64 wrote: ↑Tue Oct 22, 2024 3:48 pm

Plus once the consoles use frame buffers the advantages of FPGA evaporate. Gamecube is near perfect on Dolphin.

Why does this line keep being repeated everywhere? Have you used Dolphin? Dolphin has quite a lot of glitches and accuracy work has almost stopped. They're mostly working on the GUI now. Performance problems are obviously worse too compared to something like Snes9x. There is certainly merit in making a Gamecube FPGA core.

Yes I use Dolphin on a high end Gaming PC and its near perfect. I dont see any issues with it at all. With SW emulation you need a decent PC, its never going to be fun on an Android device.

softtest9 wrote: ↑Tue Oct 22, 2024 10:31 pm

FPGA64 wrote: ↑Tue Oct 22, 2024 3:48 pm

Plus once the consoles use frame buffers the advantages of FPGA evaporate. Gamecube is near perfect on Dolphin.

Why does this line keep being repeated everywhere? Have you used Dolphin? Dolphin has quite a lot of glitches and accuracy work has almost stopped. They're mostly working on the GUI now. Performance problems are obviously worse too compared to something like Snes9x. There is certainly merit in making a Gamecube FPGA core.

The Gamecube graphics chip contained 51 million transistors, for comparison the graphics chip of the PSX contained 1 million transistors and the graphics chip of the N64 was made of 2.6 million transistors. Reverse Engineering such a chip is an impossible task, the only way I can think of is using high resolution die shots that are converted 1:1 into the FPGA but I doubt any FPGA available today would be big enough for something like that.

niallquinn wrote: ↑Tue Oct 22, 2024 4:52 pm

??? There already is an FM Towns core?

viewtopic.php?t=104&start=60

Please work further on this core. Many people would be gratefull.

As has been stated the Author works as when they feel they want to. Japanese systems always take the longest because the tech docs are in Japanese. It will be done if and when and asking for it to happen wont make it occur any faster,

FPGA64 wrote: ↑Wed Oct 23, 2024 12:00 pm

It will be done if and when and asking for it to happen wont make it occur any faster,

People expressing interest in something (cores in this case) can absolutely be a motivation for the developer and result in resuming work, faster completion, etc.

akeley wrote: ↑Wed Oct 23, 2024 1:24 pm

FPGA64 wrote: ↑Wed Oct 23, 2024 12:00 pm

It will be done if and when and asking for it to happen wont make it occur any faster,

People expressing interest in something (cores in this case) can absolutely be a motivation for the developer and result in resuming work, faster completion, etc.

and can have exactly the opposite effect too.

FPGA64 wrote: ↑Wed Oct 23, 2024 2:35 pm

and can have exactly the opposite effect too.

If you say so, but it goes against common sense and most people's experiences.

And this kind of gatekeeping can keep people from engaging with the community, especially on a niche forum which is already pretty quiet as it is.

akeley wrote: ↑Wed Oct 23, 2024 1:24 pm

FPGA64 wrote: ↑Wed Oct 23, 2024 12:00 pm

It will be done if and when and asking for it to happen wont make it occur any faster,

People expressing interest in something (cores in this case) can absolutely be a motivation for the developer and result in resuming work, faster completion, etc.

I don't really think that's the case at all. It's a developer wanting to scratch an itch, VERY few are "community-motivated" and a great many have abandoned projects altogether as a result.

rhester72 wrote: ↑Wed Oct 23, 2024 5:03 pm

I don't really think that's the case at all. It's a developer wanting to scratch an itch, VERY few are "community-motivated" and a great many have abandoned projects altogether as a result.

Even if some developer is just "wanting to scratch an itch", how can expressing interest in his work be harmful?
Can you point me to examples of "great many" devs who have abandoned projects because some people were interested in them?
And what is your basis for saying "VERY few" are community-motivated? (even though I did not directly nor indirectly imply that, so not sure why did you put it in quotes)

I'm sorry, but unless you can prove such pronouncements with some very well documented stuff then it's just your say so, which goes totally against normal human behaviour and common sense.

Oh my goodness. Best of luck with that approach. I'm out of this discussion.

rhester72 wrote: ↑Wed Oct 23, 2024 6:28 pm

Oh my goodness. Best of luck with that approach.

So questioning and asking for examples proving your vague, sweeping claims is a no-no, some sort of faux pas?
I'd say it's some interesting logic on display here, but logic has left the building a few posts ago.

One thing I can agree on though, is that it's time to abandon thread and put this exchange under "yet another topsy-turvy forum moment".