Was anyone else disappointed with the A1200?

[Gorf]

Originally Posted by TCD

The fact that they released the A570 after they stopped making the A500 and it being incompatible with the A600 that was available at the time still blows my mind.

C= had to fulfill the contract with Matsushita ... and since they had to lower the price of the CDTV below production cost to sell it, they would make more loss producing more of it.
What to do with all these 1x speed CDROM drives and controllers?
So the options were the CDTV-CR, which was probably still too expensive, or turning the A500 into almost a CDTV.
The expansion slot to the A500 made this possible ... the A600 of course not.

The A2000 would have been a better candidate in my view as it has the large drive bay, but too few units of it were in sold previously and they needed to dump a lot of drives...

[babsimov]

I would argue the time for such a device actually never came.

The PS2, both games console and media player in the living room. In many ways the CDI and the CDTV were the precursors of this.

The mistake was the forced release along with the preorder of hundreds of thousands of CD-ROM drives from Matsushita ....

I didn't know that. Thanx.

Which brings me back to what Steve Jobs said in the video above:
Management had no clue what makes a great product, they did not understand their customers and they would not even use such a device on their own.

If I remember correctly, it wasn't really the management who wanted the CDTV, but rather the engineers. I even believe that it was Carl Sassenrath who supervised the project.

P.S.:

I love the design of the CDTV and the initial thought was "that's hot!"
And CD-ROM with its 600MB was fascinating at the time.
But in a matter of days after I read the first articles about it, I realized I would not need or even use such a device ... at least not without keyboard and floppy and then the whole purpose of that device is in question.

A bit similar. At the time I found the design cool and the concept may be interesting. The CD medium was really a novelty whose potential we did not yet really see. But, as the CDTV was still quite expensive anyway and I already had an Amiga 500, there was no point in buying this equipment for me.

Quite quickly I told myself that in fact it was money wasted for nothing by Commodore, while the chipset had not really been improved yet and it was starting to become urgent.

But in the hands of the right artist, the CDTV could have shown its full potential.

[ Show youtube player ]
[ Show youtube player ]
[ Show youtube player ]

Originally Posted by Gorf

The insane lineup of CDTV products, that must have cost Commodore a fortune:

• CDTV istself
• CD12XX Remote Control
• CD1200 Trackball Controller
• CD1221 wireless Keyboard
• CD1252 wireless Mouse
• CD1253 wired black Mouse
• CDTV Brick (Prototype?)
• CD1300 Genlock Module
• CD1321 SCART Module
• CD1411 black Floppy Drive
• 1084S-D2 black Monitor
• 363713-02 PAL Video Module
• 363713-01 NTSC Video Module
• CDTV SCSI Modul
• A570 CD-ROM for the A500 (to get rid of the preordered drives...)

And yet we argue in this thread, if the use of a bunch of TTL chips on a motherboard was too expensive ...

Thank you for this list, I didn't remember that there had been so many things for the CDTV.

Originally Posted by Gorf

C= had to fulfill the contract with Matsushita ... and since they had to lower the price of the CDTV below production cost to sell it, they would make more loss producing more of it.
What to do with all these 1x speed CDROM drives and controllers?
So the options were the CDTV-CR, which was probably still too expensive, or turning the A500 into almost a CDTV.
The expansion slot to the A500 made this possible ... the A600 of course not.

The A2000 would have been a better candidate in my view as it has the large drive bay, but too few units of it were in sold previously and they needed to dump a lot of drives...

Thank you once again for this interesting information. I didn't know the real reason for creating the A570, now I know.

[babsimov]

Originally Posted by sandruzzo

Power is nothing without control Don't forget that professional developer don't like complexity con its hard to exploit. it require a lot of time, effort and above all Money.

A 020 28mhz + blitter and copper enhanced at 32bit 28mhz would have been great...

Please, remember PS3S' cell processor: on power a true beast on real life a Nightmare

Yes, with these improvements the 1200 would have been better (in fact it was more or less what was planned for the blitter and copper of the AA+). But that wouldn't have been enough.

They also had to improve the sound to 16 bit and at least 8 channels. And of course it would have needed a chunky 256 color display at least. The DSP would also have been a significant addon.

Already at the time programming OCS was more complex than that ST or PC and yet we saw that the chipset was exploited and even some made it do things that even the creators thought impossible.

So I'm confident that the Amiga community would have gotten the most out of AGA and DSP, as originally intended for AA3000+.

[hammer]

Originally Posted by Gorf

AGA has already not enough bandwidth an ChipRam for VideoDMA and CPU and is slowing down the CPU ... so adding a DSP there would be a terrible idea.

To completely rework the Chipset, go full 32bit or better 64bit and give Alice some DSP functionality and Paula 16 channels at 16bit ... sure ... but that is nothing that could have been done in 90-92 by Commodore

Adding a DSP would at least mitigate some of the obvious or perceived deficits of the A1200.

Game consoles like 3DO, Saturn, PS1, SNES and Mega Drive had separate video memory. The same for gaming ing PCs.

A500 with shared 16 bit Chip RAM is already behind Mega Drive's 16 bit system memory + 8 bit VRAM (the ram type that was rejected by a softdrinks CEO).

Fast RAM is needed for A1200's faster 68EC020. PCJr head has JR'ed A1200 despite a healthy profit margin.

Turrican 2 AGA with VGA's 256 color artwork port needs Fast RAM. What's on the PC VGA is applied for the Amiga.
7 bit planes is the compromise example for stock A1200.

Lion King AGA wasn't the 256 color SNES/PC VGA version.

[hammer]

Originally Posted by babsimov

As I said, in no way am I saying that the DSP3210 would not have been a big advantage for the Amiga and a more than welcome addition. Dave Haynie also explains well, in fact, that MPEG and JPEG decoding would have been one of the things that the DSP would have allowed. As well as a software modem (a 2400 bds would have been included as standard). As for sound, he explained that there was a 16-bit audio codec included in the AA3000 for at least 8 channels 16-bit.

It is precisely because I always regretted that the Amiga never had DSP that I wanted to find out what a DSP could provide. And the only real example we had at the time (besides the NeXT one) was in the Falcon. And already the Falcon's DSP 56001 allows for a lot of great things.

But, in fact, we'll really never know if the DSP3210 would have made any real difference to the one included in the Falcon. The example of the DSP3210 in the MAC is not really telling, because I don't have the impression that Apple has really done anything to promote its mass exploitation. All I've read is that Photoshop filters use it, but that's it. It seems like a big waste to me. I am convinced that if the DSP3210 had been included in the Amiga, the community would have used and exploited it in many areas and even in unexpected areas (demomakers are creative).

Apple's DTP, publishinh photo editing and WYSIWYG word processing target market is larger and it pays for somebody's employment.

DSP3210 was used for QuickDraw acceleration in place of Am29K RISC CPU in 1990 Mac II. Am29K wasn't a $20 to $30 part.

[hammer]

Originally Posted by Promilus

That's how all of 3D games for PC were done up to mid 90s. Up to the point someone actually managed to create hardware acceleration for 3D and solid API for that. Because the first solutions were hardly solid (but there were different solutions... like Rendition Veritee V1000 which was based on RISC processor doing 3D stuff on own local memory). So as you can see it is way different than what amiga did provide back then. And when it comes to consoles - all of those had their computing cores fairly close to graphics unit. Exactly because it's way more efficient that way.

Rendition Veritee V1000's MIPS-like core is only clocked at 25 Mhz similar to 3DO MADAM's 25 Mhz.

The host CPU still provides point geometry calculations.

3DO has separated 2MB system RAM and 1 MB video RAM.

For 3DO, the ex-Amiga engineers didn't repeat the same mistake as the original A1000's UMA!

For PS1,


CPU, Geometry Transformation Engine and system control co-processor has 2MB EDO DRAM.

GPU has 1 MB VRAM.

Audio has 512 KB DRAM.

Originally Posted by Promilus

And I say to you it's not something you can easily fix by slapping DSP beside 68k. At some point (if using 040 or 060) that DSP would've been a waste of resources anyway. On the other hand DSP embedded into chipset structure could've been used for many things even with existence of much more powerful main CPUs. But I would say AAA improved blitter&copper duo with very simple RISC processor was already damn good solution. Too bad they did not managed to implement it.

Mixing raster display with 3D compute wasn't a good idea.

Following Xbox One's 32 MB ESRAM, the current NVIDIA ADA and RDNA 3 PC GPUs have a large cache with delta color compression (DCC) to fit the entire 4K framebuffers.

Render target framebuffers needs small data storage with fast memory bandwidth. The rest of video RAM storage is for fast texture and geometry storage access.

AMD Strix Halo APU with 40 CU RDNA 3.5 IGP has 32 MB cache and 256-bit LPDDR5-8000 configuration. IGP 's 32 MB cache is acting like a hyper-fast VRAM that reduces the load on the system 256-bit LPDDR5-8533 (273 GB/s).

I'm against "plain jane" UMA.

[sandruzzo]

@babsimov

If you couple my specs with 28mhz 32/64bit chipram + 28mhz fastmem + 28mhz "trapdoor ram"(without bitplanes penalties) + 28mhz chip registers' bus(without bitplanes penalties), you should have a lot of room without ramping up AGA COST.

What would you have? 28/56MB * 4! Better than DSP I would have add a "true" processor like PPC...

[sandruzzo]

Remember, Amiga is an Actual Home Computer (PC) not a pure consolle!

[sokolovic]

Originally Posted by hammer

.

Lion King AGA wasn't the 256 color SNES/PC VGA version.

Lion King isn't 256 colors on the SNES because SNES doesn't use 256 Colors screen mode in game.
"Only" up to 121 max colors on screen for the vast majority of games.

[Promilus]

Mixing raster display with 3D compute wasn't a good idea.

You're (again) missing the point. Let's see what 3dfx had with voodoo1
hw texturing? Yes
alpha blending? Yes
texture filtering? Yes
transformation and lighting? Nope!
Ahh, damn... well maybe Riva TNT2 Ultra had some, let me just quick check ... Nope, no luck. I know, I know, GeForce 256 was the first one to introduce HW T&L ... yay! We're in Pentium III 600MHz speed territory now so... yup, it's so damn relevant to 1992 ...

Let me just quick check what kind of hardware options VDP1 had from Saturn (late 1994 in JPN). Texture mapping and Gouraud shading... hmmm...

Ok ... so the mentioned 3DO ...
What's GPU? Is "GPU" just framebuffer and RAMDAC? No. It's everything related to hardware processing of graphics. Which coincidentally in case of 3DO consist Cell engine and vector co-processor along with CLIO functionality regarding pixel operations and CLUT. Obviously VRAM. And actual display chip.

Now you said it's bad to have all 3D compute and raster data in same memory. But ... all that data is there in one memory. Pixel shaders operate on ... well, pixels. Which means that's the raw 2D data fed to framebuffer... vertex shaders and geometry shaders operate on ... well, geometry data, how 3D scene actually is calculated. And you have to store textures somewhere as well. You say "cache" but cache it's just another level of memory which have been implemented at some point in GPU chip itself and just buffers the data from local gpu memory. All the data you were so kind to describe actually resides in one unified memory on basically any GPU for 2 decades now. Or more.

But I digress...
Amiga.
Did it have h/w texturing? Aaaa, nope!
So it didn't have texture filtering either? Nope!
Maybe Gouraud shading then? Alpha blending? Mipmapping? Transparency, fog, dithering? No?
Ok... so it appears there was a need to do SOMETHING to make 020 free of such tasks... and well, it doesn't seem like OCS blitter and copper were up to the task. I can hardly imagine DSP3210 being on 68020 bus as something which would be up to the task. Sure - performance wise yes. But again it's damn inefficient to move the data back and forth and use AGA chipset basically as display chip only. If it was going to be like that... dump whole chipset all together and just hookup RAMDAC with VRAM for display of the precalculated data.

So let us assume they put 2MB chipram acting like local memory for ...graphics mostly (and some audio too, remember that modern GPUs DO support audio! and video decoding!), that chipram is connected to much faster controller and - by extension - is much faster itself. And there in the middle of the chipset there's DSP and big, improved blitter... It is all supported through the DMA engine which is made to fit actual crunching speed of DSP and it's need for memory bandwidth (which won't change over time). Now it does open up some pretty juicy options there right? Right... so ... after 3 years you replace that battered 020@14MHz to 060@50MHz ... is DSP inside the chipset useless just because you've got powerful CPU now? No! It still can supplement it without obstructing 060. You just delegate it to different set of calculations. Maybe audio positioning system. But it still has some use and doesn't interfere with main CPU operations due to being hooked up to different memory interface.

Now imagine DSP sits on 68k bus. Everything it works on has to be in FASTRAM and any copy over to chipram obviously blocks the 020 (and by extension - through the use of the same frakking interface) DSP as well. Now DSP is inherently deterministic so should be able to access memory at specific intervals as well kind of forcing it to be highest priority. Which wouldn't affect performance of 14MHz 020 that bad (due to much higher DSP clock, faster dram chips for FAST RAM and glue logic) but... 060/50MHz is another beast here. And can also burst transfer (which 030 can and 020 cannot). Would interfacing 060 with DS3210 still provide bigger peak performance? Sure. Would they interfere with each other on daily basis limiting actual gains? Sure!

I'd say that solution at some point would've bite ass of the users. The only redeeming factor is it would've been fairly easy and cheap to implement. But also would've become obsolete fairly fast as well. It's actually like h/w T&L from GeForce256 ... it was great at the time but once 800MHz or faster CPUs came out it was already SLOWER so what's the point of having such unit?

[hammer]

Originally Posted by sokolovic

Lion King isn't 256 colors on the SNES because SNES doesn't use 256 Colors screen mode in game.
"Only" up to 121 max colors on screen for the vast majority of games.

[ Show youtube player ]
Lion King, Mega Drive, SNES and Amiga AGA comparison.


[ Show youtube player ]
Simultaneous Colors of Ultimate Mortal Kombat 3 (Genesis vs SNES vs Sega Saturn) Color Comparison
Look at 2:19 example. The Amiga AGA version is closer to the Mega Drive version.

SNES exceeds 121 colors,

At 0:06 of 2:13. SNES reached 161 colors.
At 1:00 of 2:13. SNES reached 151 colors in gameplay.

[sokolovic]

Originally Posted by hammer

[ Show youtube player ]
Lion King, Mega Drive, SNES and Amiga AGA comparison.


[ Show youtube player ]
Simultaneous Colors of Ultimate Mortal Kombat 3 (Genesis vs SNES vs Sega Saturn) Color Comparison
Look at 2:19 example. The Amiga AGA version is closer to the Mega Drive version.

SNES exceeds 121 colors,

At 0:06 of 2:13. SNES reached 161 colors.
At 1:00 of 2:13. SNES reached 151 colors in gameplay.

https://nesdoug.com/2020/04/02/snes-...208%20palettes

Mode 1
2 layers of 16 colors per tile.

1 layer of 4 colors per tile.

This is the most used mode. Nearly every game uses mode 1 most of the time. Typically, the first 16 color layer for foreground and other for background. Then the 4 color layer is used for text boxes or HUD / score display.

(8 palettes, both 16 color layers share the same 8 palettes. 15 colors per palette x 8 palettes + 1 backdrop color = 121 BG colors. The 4 color layer has to share palette space with the first palette of the other layers.)

The excess Come probably from the sprites

[hammer]

Originally Posted by Promilus

You're (again) missing the point. Let's see what 3dfx had with voodoo1
hw texturing? Yes
alpha blending? Yes
texture filtering? Yes
transformation and lighting? Nope!
Ahh, damn... well maybe Riva TNT2 Ultra had some, let me just quick check ... Nope, no luck. I know, I know, GeForce 256 was the first one to introduce HW T&L ... yay! We're in Pentium III 600MHz speed territory now so... yup, it's so damn relevant to 1992 ...

That's not relevant when A1200's 68EC020 @ 14 Mhz CPU is damn weak!

My argument for DSP3210 on Fast RAM's side is to patch the weak 68EC020 CPU. 68EC020 @ 14 Mhz CPU is the 1st port of call to be patched.

https://segaretro.org/History_of_the...rn/Development
Saturn has ditched 68030 for with two SuperH2 @ 28.2MHz (50 MIPS)

EGM reported that this new Saturn project was likely to use a Motorola 68030 processor. It also became increasingly unlikely that this new project would not be compatible with Mega Drive or Mega-CD software[2].

(skip)

On September 21st, 1993, Sega announced a joint venture with Hitachi with the intention of producing a "32-bit" video game multimedia machine, the idea being that Hitachi would be responsible for producing the processor.

PC had the luxury via a higher price segment, Pentium and "workrelated" income tax offsets.

Originally Posted by Promilus

Now you said it's bad to have all 3D compute and raster data in same memory. But ... all that data is there in one memory. Pixel shaders operate on ... well, pixels. Which means that's the raw 2D data fed to framebuffer... vertex shaders and geometry shaders operate on ... well, geometry data, how 3D scene actually is calculated. And you have to store textures somewhere as well. You say "cache" but cache it's just another level of memory which have been implemented at some point in GPU chip itself and just buffers the data from local gpu memory. All the data you were so kind to describe actually resides in one unified memory on basically any GPU for 2 decades now. Or more.

In early gaming PC 3D acceleration, transform, clipping, and lighting are done on the CPU. This is on top of running game world logic.

PC CPU evolved by gaining SIMD support e.g. Direct3D 6.0's geometry pipeline supports 3DNow and SSE with x87 being the fallback path.

In 1999, CPU power still influence frame rate results despite using GeForce 256.

For Quake III demo1
https://www.anandtech.com/show/391/14
Athlon 700 - OpenGL,
32bit render, 107.9 fps
16bit render, 126.3

https://www.anandtech.com/show/391/10
Pentium III 600B - OpenGL
32bit render, 98.1 fps
16bit render, 104.3 fps

A1200 would need two different DSPs i.e. geometry DSP for the CPU side and DSP for the visual side. For near console price targets, Motorola is out.

Track 68040 vs 486DX prices into 1994 and Motorola wasn't price competitive.

Against 3DO, A1200 might get away with a single DSP3210 on fast RAM.

Against PS1, can't avoid Amiga Hombre level specs.

Commodore has two paths and they are
1. Target game console and near game console price points.

2. Target mid-range gaming PC with a good bang per buck. A500's $699 in 1987 would be $863.20 in 1992 and $937 in 1994.

In 1989, I purchased A500 (Rev6A) with slight premium cost and slight premium hardware over Atari STe.

In US dollar terms in 1989, https://archive.org/details/amiga-wo...e/n75/mode/2up
Amiga World May 1989, Page 76 of 108
Amiga 500 has $529.
Amiga 500 with 1084S monitor has $799.
Amiga 500 and 1 MB RAM has $729.

$729 in 1989 would be $825.15 in 1992.

Reference
US Inflation Calculator
https://www.minneapolisfed.org/about...ion-calculator

My 1989 AUD budget in US dollar terms would be about $1000 USD or 500 UKP.

[Tigerskunk]

Originally Posted by grelbfarlk

What if the A1200 had Blast Processing and cartridges, and a lot of fans on the bottom so the A1200 could hover over the desktop and blow crumbs and small bugs away.

With BLAST PROCESSING reaching a solid 20 MegaTWERPS the Amiga would have wiped the floor with the PC scum.

All depending on the amount of NANO-FLONAKs, though.

Without that, it's open again.

[Amigajay]

Originally Posted by Gorf

C= had to fulfill the contract with Matsushita ... and since they had to lower the price of the CDTV below production cost to sell it, they would make more loss producing more of it.
What to do with all these 1x speed CDROM drives and controllers?
So the options were the CDTV-CR, which was probably still too expensive, or turning the A500 into almost a CDTV.
The expansion slot to the A500 made this possible ... the A600 of course not.

The A2000 would have been a better candidate in my view as it has the large drive bay, but too few units of it were in sold previously and they needed to dump a lot of drives...

The A570 drive was planned even before the CDTV was released and Commodore were talking about it and ‘not letting A500 owners down’ with it not coming out, and was shown as the A690 drive at CES 1991 so this theory is not true about the drives.

What is true is the delay of them to shift more CDTV units first, hence the A570 was delayed a few times to autumn 1992 a year after planned. (Exactly the same thing Commodore did with the A1200 CD drive, except this time there was no extra year to wait!).

But i agree with one of your earlier posts about the money Commodore lost on the CDTV project, not just money but reputation and trust with retailers which harmed them more than in the long run.

Also don’t forget another CDTV canned project, the add-on card the ‘AVM’!

[TCD]

Originally Posted by Gorf

The expansion slot to the A500 made this possible ... the A600 of course not.

"One Step Forward, Two Steps Back"

[Bruce Abbott]

Originally Posted by TCD

But they never replaced the C64 with an 'updated' model.

The C64C was an 'updated' model that replaced the original C64. The C65 was going to be the replacement for that, but it was canned because by the tine it was ready for production because the C64 market had already collapsed.

Releasing a replacement for the A500 didn't make a lot of sense,

You can blame Jeff Porter and George Robbins for that.

From the book 'Commodore the Final Years', by Brian Bagnall...

Amiga 500 Plus

Back in July 1990, just as the A3000 was going into production, Jeff Porter submitted his proposal for the next generation A500, which he called the A500 Plus. This new system would update the original A500 with the ECS chipset and the new AmigaOS 2.0...

Porter assigned the project to George Robbins [who improved it by] adding a full megabyte of memory [and] real-time clock and battery.

However Robbins barely advanced the project throughout 1990. By February 1991, Bill Sydnes wanted to know the timeline for the new computer... As it turned out the project had been delayed waiting for the new R5 Agnus... it looked like Robbins might have the A500 Plus ready to launch by September.

Putting 1MB onboard made sense because games were requiring it, and it was cheaper to put it on the motherboard than on a bundled trapdoor RAM board.

The real-time clock and AmigaOS 2.0 weren't needed for games, but the engineers wanted to deprecate 1.3 and make 2.0 the standard. In their minds 2.0 brought a lot of good stuff to the table, while 1.3 was buggy and limited. But most users didn't care about that - they just wanted their games to run properly. As usual the engineers were pushing what they wanted, not what the market wanted.

Originally Posted by TCD

but releasing two of them in such a short amount of time makes it very clear that Commodore didn't have a clear strategy for the Amiga. It almost seems like a lot of decisions were second guessed the moment they made it to the market.

The strategy was clear enough - it just didn't match their abilities. The 'second guessing' was a result of being reined in by the constraints of reality.

Like many companies, Commodore wanted to have a finger in every pie. Take Microsoft for example. Were they in the business of making an OS, or applications? Which OS would they push - DOS, or Xenix, or Windows? And which Windows - the one that ran on top of DOS, or the one that didn't (Windows NT)? And what hardware would they support - just the IBM PC, or one they designed themselves (Windows CE), or something completely different (smart phones)? And what about games - should they just supply a general purpose OS which supported them along with business apps etc., or should they make a dedicated games console to compete with the likes of Sony and Nintendo? Being Microsoft they tried to do it all of course!

[hammer]

Originally Posted by sokolovic

https://nesdoug.com/2020/04/02/snes-...208%20palettes



The excess Come probably from the sprites

Mode 3's 1 layer 256 color and 256x224 resolution.

https://nesdoug.com/2022/05/30/other-modes/

Mode 3
Mode 3 has 1 layer 256 color and 1 layer 16 color per tile.

NBA All-Star Challenge
It’s a one-on-one basketball game. Nearly every screen in this game uses mode 3.

Mode 4
Another Offset-Per-Tile mode. You get 1 layer of 256 color and 1 layer with 4 color per tile.

Mode 7
 You get one layer with 256 colors that can be stretched, skewed, and rotated. Any time you see something zoom in or out, that’s mode 7.

https://www.gamedeveloper.com/progra...super-nintendo

From a technical standpoint, the SNES excelled at 2D. Its 16-bit 65C816 3.58 MHz CPU had 128 KiB RAM available. 
It piloted a PPU (Picture Processing Unit) with 64 KiB of RAM to manipulate large sprites, 
using up to 256 colors at a resolution of 256x240. 
...
Despite its impressive 2D sprite engine and especially its "Mode 7" capability, the machine struggled 
with computationally-intensive operations such as 3D calculations.
 Nintendo was vividly aware that 3D would be the next big thing in gaming 
...
As fate would have it, a small UK firm would hold the solution to the problem.

...
Out of the native 256x224 resolution, only 216x176 was actually drawn and only 216x144 for the 3D canvas (32 rows for the status bar). 
With vertical lines duplicated, the Reality Engine was actually rendering at 108x144. 

Even at this low resolution, the average framerate was around 10 FPS which was a remarkable achievement.

The "low" framerate was not enough to discourage players from enjoying DOOM. 
According to Randy Linden the game sold very well.

An Amiga was used to develop SNES's Doom port.

“DOOM was a truly ground-breaking title and I wanted to make it possible for gamers without a PC to play the game, too. 
DOOM on the SNES was another one of those programming challenges that I knew could be accomplished.

I started the project independently and demo’d it to Sculptured Software when I had a fully operational prototype running. 
A bunch of people at Sculptured helped complete the game so it could be released in time for the holidays.

The development was challenging for a few reasons, notably there were no development systems for the SuperFX chip at the time. 
I wrote a complete set of tools - assembler, linker and debugger - before I could even start on the game itself.

The development hardware was a hacked-up Star Fox  cartridge (because it included the SuperFX chip) and 
a modified pair of game controllers that were plugged into both SNES ports and connected to the Amiga's parllel port. 
A serial protocol was used to communicate between the two for downloading code, setting breakpoints, inspecting memory, etc.

I wish there could have been more levels but unfortunately the game used the largest capacity ROM available and 
filled it almost completely. I vaguely recall there were roughly 16 bytes free, so there wasn't any more space 
available anyway! However, I did manage to include support for the Super Scope, mouse, and XBand modem! 
Yes, you could actually play against someone online!"

- Randy Linden, in conversation with Gaming Reinvented

The Amiga (perhaps AGA) was used to develop the SNES Doom port and the Amiga itself didn't have Doom until open source in 1997. This hurts...

PC Doom was reverse engineered on the Amiga based on "Unofficial Doom Specs" by Matthew Fell which explained the .wad lump layout in detail.
The sprites, textures, music, sound effects and maps were extracted from DOOM.WAD.

Nintendo backed the developer for the bundled SuperFX2 accelerator.

[Bruce Abbott]

Originally Posted by Promilus

I'd say that solution at some point would've bite ass of the users. The only redeeming factor is it would've been fairly easy and cheap to implement. But also would've become obsolete fairly fast as well. It's actually like h/w T&L from GeForce256 ... it was great at the time but once 800MHz or faster CPUs came out it was already SLOWER so what's the point of having such unit?

Exactly. More complicated dedicated hardware that was only present on that one platform - and a subset of it to boot. Meanwhile in the real world (Doom) how was it being done? Purely in software which could run on any platform with enough compute power.

An A1200 with 14MHz 020 and DSP chip could achieve similar results to putting a 50MHz 030 in it - except the 030 would accelerate everything, not just the few things suited to the DSP. And the same stuff would run on any other Amiga with a fast CPU (even better if its CPU was faster). Furthermore - perhaps more importantly - you could port games from that other platform (the PC) with minimal effort.

[dreadnought]

Thank god somebody mentioned Doom again, I was getting worried.