SNES Doom experiment (UAC Max'd/Arena Mode)

[Roebloz]

Testing out with mixing Greg Hastings Paintball level design with SNES Doom...Pretty funky results occur.

 

[Dark Pulse]

Pretty soon, gonna need to see if I can get hold of those files. Randy seems to be pretty busy as his releases of stuff have been stopped for quite some time now.

[xX_Lol6_Xx]

Awesome!

[Redneckerz]

I dream of the day SNES Doom can load custom content and (limited) TC's become possible, essentially allowing people to make new games for the SNES using the Reality Engine.

 

GBADoom shows it, Doom 32X shows it. One day...

[Dark Pulse]

On 3/7/2022 at 2:09 PM, Redneckerz said:

I dream of the day SNES Doom can load custom content and (limited) TC's become possible, essentially allowing people to make new games for the SNES using the Reality Engine.

 

GBADoom shows it, Doom 32X shows it. One day...

Admittedly what makes this tough is the ROM limit. The SuperFX 2 can only address 16 MBit of ROM. Some sort of "extended" version of it would be needed to begin to approach that sort of stuff.

 

While you could technically increase the ROM size (The SNES can address up to 128 Mbit/16 MB, but only 117.5 MBit of that would be accessible for technical reasons), it'd be at a relatively slower speed. A more "normal" mapping could do ~95 MBit (12 MB) but with 48 Mbit (6 MB) at what's called FastROM speed.

 

But the SuperFX can still only address 16 MBit. You'd need to bank data in and out of that 2 MB window, or "extend" the SuperFX 2 to address more ROM.

[Kyle07]

Is this a port of an existing wad or is this a recreation? As far as I know, DOOM on the SNES does not run on the doom engine.

[Roebloz]

22 hours ago, Kyle07 said:

Is this a port of an existing wad or is this a recreation? As far as I know, DOOM on the SNES does not run on the doom engine.

This is a WAD I made myself, and inserted it into SNES Doom with Mopoz's custom and (sadly) in closed-beta. SNES Doom is indeed not the Doom engine, but it is way closer to it than let's say, Doom 2 GBA.

[Dexiaz]

On 3/9/2022 at 6:44 PM, Kyle07 said:

As far as I know, DOOM on the SNES does not run on the doom engine.

But it reads WAD content. It's not an entirely different engine.

[Redneckerz]

On 3/9/2022 at 1:50 AM, Dark Pulse said:

Admittedly what makes this tough is the ROM limit. The SuperFX 2 can only address 16 MBit of ROM. Some sort of "extended" version of it would be needed to begin to approach that sort of stuff.

 

While you could technically increase the ROM size (The SNES can address up to 128 Mbit/16 MB, but only 117.5 MBit of that would be accessible for technical reasons), it'd be at a relatively slower speed. A more "normal" mapping could do ~95 MBit (12 MB) but with 48 Mbit (6 MB) at what's called FastROM speed.

 

But the SuperFX can still only address 16 MBit. You'd need to bank data in and out of that 2 MB window, or "extend" the SuperFX 2 to address more ROM.

Would a mixture of FastDoom, MiniWAD and this be sufficient on a stock system? It will likely take some know how.

[Dark Pulse]

2 hours ago, Redneckerz said:

Would a mixture of FastDoom, MiniWAD and this be sufficient on a stock system? It will likely take some know how.

Maybe, but at that point you might as well play Faceball 2000.

 

Obviously if you reduce the data down to under 2 MB/16 MBit (including code, of course), and can make it run within the limited resources the SuperFX had access to (64 KB/512 KBit of work RAM), then it should be theoretically possible.

 

But if you expand the resources the SuperFX can address, you wouldn't need to cut anything down, either. Then you can fit more data it can access.

 

Also, again, the cartridge can fit more ROM to hold the data. The "hard" limit is that the SuperFX can only address 2 MB/16 MBit of that data at a time. So an "extension" could be as simple as implementing a bankswitching scheme within that space, though obviously an address extension will also solve that problem in a less flukey way. Even adding a single extra address line should shoot that limit up to 4 MB/32 MBit, two gets 8 MB, and so on.

 

It's ROM, so paging and loading data would be relatively fast.

Edited March 10, 2022 by Dark Pulse

[Roebloz]

4 hours ago, Dexiaz said:

But it reads WAD content. It's not an entirely different engine.

Indeed. (The WAD does have to be converted though.) Essentially, Randy built it off the Unofficial Doom Specs, so it's still closer to Doom than Doom 2 GBA.)

[Nikku4211]

On 3/8/2022 at 7:50 PM, Dark Pulse said:

Admittedly what makes this tough is the ROM limit. The SuperFX 2 can only address 16 MBit of ROM. Some sort of "extended" version of it would be needed to begin to approach that sort of stuff.

 

While you could technically increase the ROM size (The SNES can address up to 128 Mbit/16 MB, but only 117.5 MBit of that would be accessible for technical reasons), it'd be at a relatively slower speed. A more "normal" mapping could do ~95 MBit (12 MB) but with 48 Mbit (6 MB) at what's called FastROM speed.

 

But the SuperFX can still only address 16 MBit. You'd need to bank data in and out of that 2 MB window, or "extend" the SuperFX 2 to address more ROM.

 

The SuperFX limits really make me wonder what SNES Doom would be like had it used the SA-1 instead of the Super FX 2.

 

SA-1 is a faster 65816, much faster than the 65816 used by the SNES itself. It can also do character data conversion, so you can write in a linear 8BPP format and convert it to a planar 8BPP format, if you feel like it.

 

Now if only SNES Doom supported using the SNES mouse and a SNES controller simultaneously. I've heard this along with the Super FX draws too much power, does it also draw too much power when an SA-1 is used instead?

 

Of course, SA-1s aren't produced anymore, but how hard would it be to produce another 65816 containing changes similar enough to Nintendo's SA-1?

[Dark Pulse]

23 minutes ago, Nikku4211 said:

The SuperFX limits really make me wonder what SNES Doom would be like had it used the SA-1 instead of the Super FX 2.

Poorly. The SA-1 was not really designed to render polygons, and that's the main thing the SuperFX is doing. The only other thing that even came close was the Cx4, and that was limited to wireframes.

 

The SA-1 is good for speeding up 2D games, for sure. 3D is an entirely different beast.

 

26 minutes ago, Nikku4211 said:

Now if only SNES Doom supported using the SNES mouse and a SNES controller simultaneously. I've heard this along with the Super FX draws too much power, does it also draw too much power when an SA-1 is used instead?

Not sure where you heard that. Fundamentally it's just polling devices.

 

28 minutes ago, Nikku4211 said:

Of course, SA-1s aren't produced anymore, but how hard would it be to produce another 65816 containing changes similar enough to Nintendo's SA-1?

It should be possible to program it as a FPGA or something like that, but I don't think any of the chips on the market currently would have all the features needed.

[Nikku4211]

1 hour ago, Dark Pulse said:

Poorly. The SA-1 was not really designed to render polygons, and that's the main thing the SuperFX is doing. The only other thing that even came close was the Cx4, and that was limited to wireframes.

Sure.

 

Though the SuperFX is really just a fast CPU with its own RAM that has to be manually programmed to draw polygons by the software itself, rather than using a specialised hardware polygon rasteriser graphics pipeline. Though it does have a PLOT opcode in its instruction set.

1 hour ago, Dark Pulse said:

Not sure where you heard that. Fundamentally it's just polling devices.

I'm in communities that specialise in learning how the SNES works. I don't remember who exactly told me this, but it's been said time and time again.

 

They are polling devices, but they need electrical power to work, so when they are instructed to do something, they take power from the console. The extra SuperFX hardware also takes power from the console.

 

That's presumably the logic, but I don't actually know if anyone has proven it to be true. If someone finds out it's true, they can end up having a console that no longer works anymore, which isn't good considering the console hasn't been produced for around 20 years.

[Dark Pulse]

5 minutes ago, Nikku4211 said:

Though the SuperFX is really just a fast CPU with its own RAM that has to be manually programmed to draw polygons by the software itself, rather than using a specialised hardware polygon rasteriser graphics pipeline. Though it does have a PLOT opcode in its instruction set.

Point still stands it is much more tuned for it than the SA-1 could be. The SuperFX is also double the clock speed of the SA-1, although that's an apples-to-oranges comparison.

 

If memory serves right, the game code would also have to be rewritten - the SuperFX is essentially a flavor of RISC, while the SA-1 (and the 65C816) are basically based off of the 6502.

 

15 minutes ago, Nikku4211 said:

I'm in communities that specialise in learning how the SNES works. I don't remember who exactly told me this, but it's been said time and time again.

 

They are polling devices, but they need electrical power to work, so when they are instructed to do something, they take power from the console. The extra SuperFX hardware also takes power from the console.

 

That's presumably the logic, but I don't actually know if anyone has proven it to be true. If someone finds out it's true, they can end up having a console that no longer works anymore, which isn't good considering the console hasn't been produced for around 20 years.

I'm really not sure why. There's a limited amount of power that can go through the controller ports (your standard +5 volts). Nothing can really consume more than that.

 

Admittedly, though, I don't think there ever was a simultaneous two-player SuperFX game, but I'd have to look into the specifics of it all.

[Nikku4211]

15 hours ago, Dark Pulse said:

If memory serves right, the game code would also have to be rewritten - the SuperFX is essentially a flavor of RISC, while the SA-1 (and the 65C816) are basically based off of the 6502.

A good amount of it, yeah.

 

SNES homebrew programmers (and ROM hackers) today know 6502 assembly more than they know SuperFX assembly, so it'd be a lot of work to learn the SuperFX flavour of RISC, though it may pay off in the long run, and even after working on this one game, could result in more homebrew games/demos that use the SuperFX 2.

 

That being said, SuperFXes aren't being made anymore, so you'd have to either emulate it with an FPGA(some consider it not emulation, but I do, and I don't consider it real expansion chip hardware anyway) or take a chip from the donor cart and increase scarcity.

15 hours ago, Dark Pulse said:

Admittedly, though, I don't think there ever was a simultaneous two-player SuperFX game, but I'd have to look into the specifics of it all.

Dirt Trax FX supports 2 player split-screen.

 

Here's a 2-player Dirt Trax FX game.

 

The game seems to support 8 players (which would require 2 multitaps) according to its menu, but I couldn't find any footage of this in action. I can't imagine how a multiplayer racing game where you take turns would be like, though.

 

Stunt Race FX also has 2-player split screen.

 

You're excluding XBand, right?

[Dark Pulse]

3 hours ago, Nikku4211 said:

You're excluding XBand, right?

XBand was intentionally excluded because the point you posed is a local machine problem - that you can't do two players with one using a controller and the other using a mouse along with a SuperFX because it draws too much power. The other two are valid though.

 

The SNES Power Supply is 10 Volts, 850 mA, and the extra pins that a SuperFX cart uses are mostly address pins off a separate address bus from the main address bus, your usual clock lines, read/write control, WRAM access, expansion detection, refresh lines, a couple audio channel lines, and that's about it. Obviously an active coprocessor will consume a bit more power, but none of the extra pins are devoted to any sort of extra power delivery, or even a ground. The two VCCs and two Ground lines are part of the main row of pins (5-27, 36-58) that all cartridges use, expansion chips or not.

 

I mean, I'm no technical expert on the hardware (though fairly versed myself, as you can tell), so if the experts are saying it's not possible, I'm generally gonna defer to them, but I'm struggling on my own to see why this is not possible. Anything plugged into the controller port gets the same 5V, the SNES mouse is an oldschool ball mouse so it's just reading changed values (so if anything it has no more to read than a regular controller, possibly even less due to having less buttons), and the expansion pins on the cartridge slot give no extra power - nor demand it. Logically speaking we'd have to say either the mouse somehow drew more power, which is silly, or that the expansion chips drew more power - but if that was the case, there would be no simultaneous two-player modes available for the chip, and you just proved that via Dirt Trax FX and Stunt Race FX.

 

So I'm very puzzled.

[Roebloz]

Funny how I posted a little experiment, and now we are having a thesis on SNES Doom. 

[Dark Pulse]

21 minutes ago, Roebloz said:

Funny how I posted a little experiment, and now we are having a thesis on SNES Doom. 

 

[Nikku4211]

5 hours ago, Dark Pulse said:

the SNES mouse is an oldschool ball mouse so it's just reading changed values (so if anything it has no more to read than a regular controller, possibly even less due to having less buttons)

 

From what I've heard, it uses a full microcontroller instead of a shift register. Not sure if that matters, though.

 

5 hours ago, Dark Pulse said:

...you monster...

Monster?

How should I feel?

Creatures lay here...

looking through the window.