Aftermarket SNES Cartridge Expansion Hardware

[turboxray]

Obviously subjective as to where to draw the line, especially nowadays where you can put tons of rom on cart (with even the most simplest of mappers), what is considered cheating a "lil bit" and what's considered "fake"?

I always figured whatever was used as official addons, are definitely legit (even if you have to approximate it with new hardware). There's nothing cheating or fake about having an SA-1 in your projects.. it's just period correct for late gen SNES games (and I'm sure it wasn't cheap haha). You could even do a bit of extended rom (assuming it overlapped with N64 era). For the same reason, I consider having 2megabytes of ram for a CD game for PCE is legit as well (Arcade card).

I draw the line on stuff like MSU1 being touted as a SNES-CD equivalent. It might have been inspired by the SNES-CD (and that's what I remember when it was being talking about in design phase), but I clearly remember the final consensus being "why limit ourselves to 90s tech". So I would never count it as such; it doesn't have the limitations that are inherit CD game development - no matter how much you pretend otherwise. Seek times, sloooow data transfer rate, limited cart ram (to hold code and data), only being able to load either data or stream audio - not both. Those things strongly impact game design. Having developed software/games for CD games.. those things matter to me. Especially when you had to really optimize for all those things. MSU1 throws that ALL out of the window. Don't get me wrong: I think the MSU is great for rom hacks (and that really was its intended purpose from what I remember). But having said that, if you did make a more 'authentic' SNES-CD "chip"... would "borrowing" other systems' addons/capabilities be considered legit? If the Arcade card has 2megabytes of ram, why not the same for a more authentic SNES-CD setup? I think that's valid. Tho I think in the SNES case, it had more addons chips in cart than any other system... so it doesn't mean much on the SNES side. But hey, SA-1 in a PCE or MD.. or NES? That last one feels weird.

If you're worried about gaming audience.. I think most gamers won't care as long as it runs on a SNES. The only people judging you are mostly other devs hahah. So I guess at the end of the day, it's whatever you feel comfortable with using. As a gamer, and not a retro dev, I know that I personally like retro dev that mostly adheres to the original console limits. I'm not opposed to some sort of fantasy/alt-world addon chip as long as it's very much period specific (and I mean that in a "makes sense in a commercial" way.. as in cost). If a game is coming at me with crazy stuff in the cart (even if be it "possible" in the 90s), as a gamer, I'm not really interested. If breaks the "suspension of disbelief" so to speak, not really interested. That said, without trying to sound too hypocritical, I give the NES much more freedom in this area than the 16bit console retro dev hahah. Don't ask me why (doesn't make logical sense).

Anyway, if someone does their best to avoid mappers, addons, mass roms.. for snes retro dev and it looks great - I most likely WILL be hyped for that release. There is something to be said about development along original "common" constraints. It's pretty damn attractive.

[lidnariq]

But hey, SA-1 in a PCE or MD.. or NES? That last one feels weird.

SA-1 in a NES would be amazing. Goofy, but amazing.

After all, Seta was planning on launching their ARM-accelerated Shogi game for both SNES and NES - and they got as far as lot check.

[Oziphantom]

It seems like everybody's talking about extreme high-end hardware meanwhile when I created the thread, I was thinking more along the lines of some sort of embedded processor that isn't anywhere near as powerful as whatever NVidia or AMD's pumping out or even Intel's stuff.

Like maybe an ARM chip (not the GBA's) or something, though of course even that would still be very powerful for hardware meant to interact with 90s hardware.

Okay we add an Arm2, can't use it to interact with the SNES in a native way so its a coprocessor and its just as awkward to program for as the 65816, we could get a bit more power out of it sure, there is a game that does. But at this point I feel its a move sideways in terms of development time/the game you can make. Adding a SH2 would be funny though.

[Kannagi]

I agree with Oziphantom, for me the debate is irrelevant, the SNES must work in stock.
That's the goal of making retro games, to make a machine work again by exploiting its hardware.

If the goal is that your SNES has just one output on the screen.
Bought a rasberry PI or a SNES mini and make your games in it without any constraints, without exploiting any hardware.

[Oziphantom]

they do still make SH2 with a video output engine and tonne of internal ram.. sadly not 5V though https://www.renesas.com/us/en/products/ ... ontrollers

[mannes]

However, in the context of shoving multi GHz co-processors into a retro machine to heavily bypass that machine's limitations? You are no longer making a game for that system. In fact, the SuperFX only gets a passing grade because Nintendo released it. Oziphantom was asking "why?". If you need a multi GHz processor to run your game then you are either a terrible programmer, making a 'modern' game with retro aesthetics or want something to serve as anti-emulation DRM.

Or maybe you want to do something cool that pushes system limits like NES droom. This might be the first post I've seen yet that hints at logically or sensibly hating the NES Franken-kart needed to mount a Raspberry Pi but the guy who did it has only gotten positive recognition for doing it. In the same way, the SFX2 was Nintendo approved and everybody has clapped.

My problem is that I cannot justify dropping $300 on a FX PAX PRO that utilizes a bunch of emulation of special chips but doesn't emulate something like the R3000 or a SA-1 at 100-500MHz. If by some miracle a new discovery is made on the SNES that allows me to port Powerslave, I'm going to need more horse power to do it. And I will likely have to handle bitmap to CHR in real-time which can be very expensive. It only alienates developers and hobbyists who might be interested in doing something new or spicy when in reality bothering to emulate 20 special SNES chips but not giving access to the developer to overclock even one is somehow seen as ok. It's about as ridiculous as a governor on a sport bike except this one in particular set to 20MPH when it could be at least 160.

[mannes]

Okay we add an Arm2, can't use it to interact with the SNES in a native way so its a coprocessor and its just as awkward to program for as the 65816, we could get a bit more power out of it sure, there is a game that does. But at this point I feel its a move sideways in terms of development time/the game you can make. Adding a SH2 would be funny though.

To be fair while I hate ARM when compared to something like MIPS, it makes about as much sense as MIPS by the end of the day. I'd prefer wrestling with MIPS or a overclocked R4300 compared to ARM, but both architectures usually just have the software compiled for it anyway in C/C++, like StarFox or the other SFX2 games.

Point is an ARM chip even if it wasn't energy friendly, would be a huge help over nothing and would solve my problems with tinkering on a classic console if I wanted to try something crazy like super botw.

[lidnariq]

My problem is that I cannot justify dropping $300 on a FX PAX PRO that utilizes a bunch of emulation of special chips but doesn't emulate something like the R3000 or a SA-1 at 100-500MHz. If by some miracle a new discovery is made on the SNES that allows me to port Powerslave, I'm going to need more horse power to do it.

The problem is two halves:
1- If you're not dealing with all of the challenges of programming for the SNES, philosophically is it still a SNES game? That includes dealing with the 65816 and a plausible-in-1995 coprocessor. The logical extreme of this is the SGB, and the aftermarket GBA-in-a-SNES devices that use the SNES only for power and controllers. Are all Game Boy and GBA games thus SNES games? Surely not.

1b- As an even more extreme example of this, the N64 has more than enough DMA bandwidth to update the framebuffer at 60Hz. If you put in a coprocessor that's just drawing whole framebuffers (possibly even streamed via wifi), is that an N64 game?

2- A lot of what makes retrodev interesting is dealing with the real limits of what the hardware can do. If you're just giving up on the 65816, was it still an interesting challenge?

[creaothceann]

Or maybe you want to do something cool that pushes system limits like NES Doom [...] the guy who did it has only gotten positive recognition for doing it

"There's no cheating.... except for a tiny computer that's 4000x more powerful than the original hardware of the NES...."
"I didn't cheat" - Literally 15 seconds later... - "Here's how I cheated"
"That sounds exactly like cheating to me."

My problem is that I cannot justify dropping $300 on a FXPAK PRO that utilizes a bunch of emulation of special chips but doesn't emulate something like the R3000 or a SA-1 at 100-500MHz

The FXPAK PRO uses an Altera Cyclone IV; there is also at least one R3000 core for it. So theoretically you could buy an FXPAK PRO and replace the firmware with your own, which might involve replacing the on-board storage.

handle bitmap to CHR in real-time which can be very expensive

From my (limited) understanding an FPGA should allow the implementation of a bit shuffle matrix that can convert the format in a single clock cycle.

It only alienates developers and hobbyists who might be interested in doing something new or spicy when in reality bothering to emulate 20 special SNES chips but not giving access to the developer to overclock even one is somehow seen as ok

Developers mainly wanted to run their code on the main CPU, because adding chips adds cost = lowers sales = lowers profit, unless it reduced piracy enough. In fact this actually might be the main reason why the new Doom version has a custom chip...

Older emulators used to allow overclocking which can be nice for fixing slowdowns (and possibly increasing difficulty) in addition to making existing games more unstable. Newer emulators can increase the CPU's VBLANK time.

There's nothing wrong with "doing something new or spicy", the issue is in still calling it a SNES game.

[turboxray]

Or maybe you want to do something cool that pushes system limits like NES droom.

That's not "pushing" the NES. The NES isn't even doing anything. That project was interesting.. for all about 10 seconds. People love a good party trick. But that's about all that's worth.

This might be the first post I've seen yet that hints at logically or sensibly hating the NES Franken-kart needed to mount a Raspberry Pi but the guy who did it has only gotten positive recognition for doing it. In the same way, the SFX2 was Nintendo approved and everybody has clapped.

The two aren't equivalent at all. At all.

My problem is that I cannot justify dropping $300 on a FX PAX PRO that utilizes a bunch of emulation of special chips but doesn't emulate something like the R3000 or a SA-1 at 100-500MHz. If by some miracle a new discovery is made on the SNES that allows me to port Powerslave, I'm going to need more horse power to do it. And I will likely have to handle bitmap to CHR in real-time which can be very expensive. It only alienates developers and hobbyists who might be interested in doing something new or spicy when in reality bothering to emulate 20 special SNES chips but not giving access to the developer to overclock even one is somehow seen as ok. It's about as ridiculous as a governor on a sport bike except this one in particular set to 20MPH when it could be at least 160.

I think you forgot the "/s"

[tepples]

The problem is two halves:
1- If you're not dealing with all of the challenges of programming for the SNES, philosophically is it still a SNES game?

So long "all of the challenges" doesn't include LARPing the nontechnical challenge of getting approval of your project (the studio, the game pitch, and the final ROM) from the company authorized to make the CIC.

[Nikku4211]

The problem is two halves:
1- If you're not dealing with all of the challenges of programming for the SNES, philosophically is it still a SNES game?

So long "all of the challenges" doesn't include LARPing the nontechnical challenge of getting approval of your project (the studio, the game pitch, and the final ROM) from the company authorized to make the CIC.

Pretty sure he just means all of the challenges of just working with the SNES hardware by itself in general, and not any of the challenges of dealing with a publisher or console manufacturer in the 1990s, or even the challenges of dealing with the exact tools that actually existed back in the 1990s.

How would you even LARP that now anyway? If we're LARPing, then it's time for me to dress up as my DnD character.

[Oziphantom]

My problem is that I cannot justify dropping $300 on a FX PAX PRO that utilizes a bunch of emulation of special chips but doesn't emulate something like the R3000 or a SA-1 at 100-500MHz. If by some miracle a new discovery is made on the SNES that allows me to port Powerslave, I'm going to need more horse power to do it. And I will likely have to handle bitmap to CHR in real-time which can be very expensive. It only alienates developers and hobbyists who might be interested in doing something new or spicy when in reality bothering to emulate 20 special SNES chips but not giving access to the developer to overclock even one is somehow seen as ok. It's about as ridiculous as a governor on a sport bike except this one in particular set to 20MPH when it could be at least 160.

The SA-1 already has bitmap-char conversion done in hardware. It is done by its DMA engine.

[Oziphantom]

but both architectures usually just have the software compiled for it anyway in C/C++, like StarFox or the other SFX2 games.

The SFX games were not written in C or C++.

Point is an ARM chip even if it wasn't energy friendly, would be a huge help over nothing and would solve my problems with tinkering on a classic console if I wanted to try something crazy like super botw.

What is the point of doing something like botw on a SNES, if you are targeting a SNES you should do something that is for the SNES. For example you don't have any analouge sticks, so moving in a 3D world with finesse on a DPAD is "not fun".

[93143]

What is the point of doing something like botw on a SNES

To see how far you can push the hardware.

I was actually brainstorming such a game a while back. Not a port of BOTW itself, of course (I'd want at least an N64 to tackle that). But a huge 3D open-world game with realistic time-of-day, dynamic weather, seasons and agriculture, astronomically correct skies, realistic lighting and shadowing, reflective translucent water waves, rustling leaves, unlimited draw distance...

All of this was to run on the Super FX. Assisted by an MSU1, of course, but that's just storage and I don't consider that quite as hard a constraint. I was also considering the feasibility of using a period-accurate CD-ROM attachment instead...

The game would have been pretty rough to look at, windowboxed with double-wide pixels (except for a few nearby NPCs and auxiliary layers like rain or the night sky) and heavy LoD on the terrain and foliage. Frame rate would have been single digits for sure, at least anywhere near relatively calm water... Using hardware sprites for the most immediately relevant NPCs would have been absolutely necessary to make fights playable, since even with normal-mapped lighting and foreground masking on the Super FX, they could update far faster than the main framebuffer.

Making this game on a modern SoC would be much less interesting.

Making something else on SNES would be much more practical...