@Karlos

MIDI is unimportant, i.e. Atari ST's 2 million install base says Hi.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

You appear not to understand sarcasm.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Karlos wrote: @pixie Quote: pixie wrote: @Karlos If it was to be taken advantage for other games, a combo with a graffiti and fast ram could perhaps do help bringing 3d games to Amiga Well, if we are talking a dedicated 3D accelerator on the sidecar, additional RAM and an RGB port expansion for displaying pseudochunky displays

What did CD32 FMV module contain?


CD32's FMV module has the following:

1. 24-bit DAC (STM's STV8438CV) for 16.7 million colors display.

2. MPEG-1 decoder from C-Cube CL450, 352 x 240 pixels @ 30hz, 352 x 288 pixels at 25 Hz, pixel interpolation and frame duplication to produce output formats of 704 x 240 pixels at 60 Hz or 704 x 288 pixels at 50 Hz.

https://websrv.cecs.uci.edu/~papers/mpr/MPR/ARTICLES/060803.PDF

CL450 has about 398K transistors with up to 40 MHz. CL450 includes a licensed MIPS-X RISC processor with semi-custom extensions. In quantities of 100K or more per year, the price is less than $50 in 1992.

CL450's MIPS-X RISC processor still has the usual RISC instruction set. CL450's MIPS-X RISC processor's compute power exceeds 68LC040 and the Amiga couldn't use this CPU.

CL450's MIPS-X is a three-operand capable RISC CPU.

3. LSI l64111qc (Digital Audio Decoder, 16-bit DAC),

4. 512 KB local RAM, NEC 423260 DRAM 4Mbit (512 KB) with 80 ns.

5. Lattice ispLSI 1024-60LJ CPLD.

Last edited by Hammer on 15-Jul-2024 at 01:55 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

The FMV module was not created to run a single movie. That's the parallel being made with a custom cartridge for doom.

_________________
Doing stupid things for fun...

@Karlos

SNES's SuperFX and game content ROM share a single slot while A500/CD32's math accelerator and game content media are separated.

https://eab.abime.net/showpost.php?p=1164371&postcount=48

The transistor count for AGA's two chips = 200,000. Alice has 80,000 transistors.

ARM60 has 35,530 transistors.

With enough skill, a 100,000-transistor RISC math chip could be fabricated like CSG's Alice.

For example, ARM 9TDMI (32-bit, no cache) has 111,000 transistors.

Commodore has the fabs with misguided direction. In terms of transistor count, CSG's AGA-era fab capability is about ARM60 to ARM 9TDMI level. Commodore is closer to ARM's "cheap RISC" instead of "big iron" RISC.

$40 Amiga Hombre has a 1 million transistor budget.

Last edited by Hammer on 15-Jul-2024 at 02:20 PM.
Last edited by Hammer on 15-Jul-2024 at 02:17 PM.
Last edited by Hammer on 15-Jul-2024 at 02:13 PM.
Last edited by Hammer on 15-Jul-2024 at 02:09 PM.
Last edited by Hammer on 15-Jul-2024 at 02:08 PM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

You are just waffling utter gash.

Tell us, based on your extensive library of early 1990's hardware prices, to build a completely custom sidecar expansion for the A500 with a dedicated accelerator intended to run DOOM and ideally display it in something more than 32 colours, what retail price are we looking at in 1993?

Remember, you have to pay a bunch of hardware engineers to design and debug it, build a similar prototype application to prove it's possible (which is what happened with the SNES/SuperFX2 port), approach ID for a license, factor in the cost of ROM big enough to hold all the assets, RAM to run it, whatever else you need. Then you have to actually port the game, make modifications to fit your particular limitations and get final approval. And you have to pay for all that before it can actually go on sale.

It sure as hell isn't going to be the typical asking price, or take a sensible amount of time. Commodore isn't going to stump up anything given it's already dying on it's arse by this time.

Could such a device be built? Probably. Would it be remotely practical given what we know of the state of the Amiga in late 1993? No chance.

The utterly nonsensical whatifery going on is silly, even by the standards of the typical "what if doom" threads.

_________________
Doing stupid things for fun...

@Karlos

Well. If the price was right... it wouldn't be that stopping people back in 1992, it doesn't stop them now with all the kludges. Fast Ram gives you about 2x speed on a A1200 at a low cost, Graffitti.. well people had all sorts of modulators back then, design wasn't what it is now, let's say. ;) I did not talk about no 3D accelerator, though
Not even that big
Graffitti

Last edited by pixie on 15-Jul-2024 at 07:17 PM.
Last edited by pixie on 15-Jul-2024 at 07:07 PM.

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

Karlos Quote:

You seem to be living under the delusion that your suggestion was feasible. Sadly, I don't think it was. First of all, there was no "SuperFX" processor for the Amiga. The best you could hope for was an off the shelf DSP. Now you have to display what is being rendered. Is that going to pass back into the 32-colour OCS chipset? How? Or are you going to have to add some sort of minimum framebuffer implementation as well and a suitable display output? That hardware is starting to sound more and moreo cumbersome by the second.

Actually, the SNES and Amiga were headed in a similar DSP direction with similar technology. The SuperFX chips are DSPs based on the NEC μPD7720 using the famous Bell Labs (AT&T) DSP1.

https://en.wikipedia.org/wiki/List_of_Super_NES_enhancement_chips#DSP
https://en.wikipedia.org/wiki/NEC_μPD77C25
https://en.wikipedia.org/wiki/AT&T_DSP1

The DSP1 was released in 1979 like the 68000 and licensed by NEC and Nintendo.

SNES SuperFX DSPs
16-bit AT&T DSP1 with fixed point integer only (up to 20-bit int datatype)
custom microcode

Amiga and Mac DSP
32-bit AT&T DSP3210 with single precision floating point (up to 40-bit fp datatype)
customizable AT&T Application Library/codec software

The SNES DSP tech was primitive compared to the Amiga DSP tech but it was cheap enough to put in a cartridge while the Amiga DSP was too expensive for CBM to make standard. Software programmability requires more memory and CBM was against increasing the Amiga memory and more chips with the "read my lips, no new chips" mandate.

The AT&T DSP3210 was a very nice DSP but would the Amiga have been better off with alternative enhancements instead? I tend to believe VRAM or fast mem and more integrated chipset upgrades would have been preferable. Specifically, upgrading the memory bandwidth and adding chunky support needed to be priorities. A DSP has to share the system memory bandwidth or have its own memory but the former is a problem as Amiga systems already had too low of memory bandwidth and the latter increases the cost without increasing the more general purpose and useful system memory. It's not that I'm against a standard Amiga DSP but I believe a DSP would have been better for high end Amigas, similar to the Mac Quadra AV models, to offload the 68k CPU (a possible exception is a DSP integrated into the Amiga chipset). Someone is probably going to Hammer the number crunching superiority of a DSP over a 68k CPU but DSPs are difficult to extract performance for general purpose use.

http://ftp.vim.org/pub/ftp/ftp/infomac/info/hdwr/av-dsp-faq-101.txt Quote:

You will need MPW to use these tools. The tools include the AT&T DSP3210 assembler, linker/loader, simulator, Macsbug D Commands, Apple Snoopy browser/debugger, and Apple BugLite graphical DSP module installer. It does not include the C Compiler but as you may have already read, the compiler is not very good. Fortunately, DSP3210 assembly programming (unlike other DSPs) is very C-like, so getting up to speed is not too difficult. The main part is getting used to the latencies involved with the pipelined architecture. Additionally, the AT&T Application Library includes many (about 100?) useful DSP algorithms (FFTs, FIR and IIR filters) with complete source code. It should sell for about $100. (Walter Horat) replies: Apple got the 3210 tools from AT&T, which were written to use VCOS (AT&Ts 3210 realtime O/S). Apple modified the tools somewhat for use on the Mac, and began shipping them to developers. When Apple realized that it might cost money to actually *support* the tools, they foisted the tools off on a company called Spectral Innovations (actually not a bad choice since Spectral has been doing programming for AT&T DSPs for over five years). BTW, the 'C' compiler is a complete piece of shit. It produces some of the worst code I have ever seen (trying to do a matrix multiply in 3210 'C' ran 5 times slower than the host 68k on a Quadra 700 - rewriting the same in assembler ran 7-8 times *faster*). If you do any serious 3210 programming, you will need to learn 3210 assembler.

Part of the problem with DSPs is not just pipeline latencies but memory latencies which the programmer has to optimize for in assembler as compilers have trouble. I'd rather have more system memory bandwidth, more memory and an upgraded 68k CPU before a DSP. Today, SIMD units have replaced DSPs in all but low end CPUs and they are difficult to program and a challenge for compilers too.

Karlos Quote:

You don't get any support from id in 1993 because the considered opinion of their cheif technical guy is that it's basically impossible. For the SNES port, the SuperFX already existed and had been tried and tested in other 3D tasks so it looks more achievable immediately. You have a large installed platform base so it's worth considering.

The ID guys were Apple/Jobs/Mac fans and clearly biased against the Amiga. An Amiga port was technically possible and easier than a SNES port but there were not enough high spec Amigas and CBM was dead.

@Karlos

Quote:

Karlos wrote: @Hammer You are just waffling utter gash.

Bullshit. Read the fukcing "Commodore The Final Years by Brian" for CSG component BOM prices.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Karlos

Quote:

Karlos wrote: @Hammer You appear not to understand sarcasm.

You've too high expectations from a bot...

@matthey

Quote:

matthey wrote: Software programmability requires more memory and CBM was against increasing the Amiga memory and more chips with the "read my lips, no new chips" mandate.

This sentence was reported by Haynie, but we know that it wasn't true: Commodore created new chips. Even completely (and expensive) new ones (Amber + dedicated memory).

This was said to justify the inability to do proper enhancement to the Amiga chipset, as I've proved on a couple of recent articles on the topic.
Quote:

The AT&T DSP3210 was a very nice DSP but would the Amiga have been better off with alternative enhancements instead?

Absolutely! And I'll write another article to further prove it.
Quote:

I tend to believe VRAM or fast mem and more integrated chipset upgrades would have been preferable. Specifically, upgrading the memory bandwidth

You don't even need more bandwidth to make sensible and important changes to the chipset.

What you (read: Commodore engineers which supplanted the original team) need to know is... how the chipset worked. And I'll prove it (but the above two articles gave already some good points about it).
Quote:

and adding chunky support needed to be priorities.

And it was possible with very minimal efforts, as I've written on another article about Akiko.

With the added bonus that it perfectly fits on how the Amiga chipset worked. IF "someone" knew it...
Quote:

A DSP has to share the system memory bandwidth or have its own memory but the former is a problem as Amiga systems already had too low of memory bandwidth and the latter increases the cost without increasing the more general purpose and useful system memory. It's not that I'm against a standard Amiga DSP but I believe a DSP would have been better for high end Amigas, similar to the Mac Quadra AV models, to offload the 68k CPU (a possible exception is a DSP integrated into the Amiga chipset). Someone is probably going to Hammer the number crunching superiority of a DSP over a 68k CPU but DSPs are difficult to extract performance for general purpose use.

Indeed. And... you don't really need to add another alien component to the Amiga ecosystem to improve the functionalities of the machine.

One thing that the people like Hammer completely ignore is the context of the time: how the Amiga worked, which kind of software/games where developed for it, what was its customer base.

The dumb mantra is: add the the DSP. Only because some engineers played with this toy. And completely ignoring the Amiga identity and market. Only Commodore engineers made it possible...

Last edited by cdimauro on 16-Jul-2024 at 05:26 AM.

@matthey

Quote:

@matthey Actually, the SNES and Amiga were headed in a similar DSP direction with similar technology. The SuperFX chips are DSPs based on the NEC μPD7720 using the famous Bell Labs (AT&T) DSP1.

NEC uPD7720 has a VLIW-like instruction format.

https://en.wikipedia.org/wiki/Super_FX
Quote:

The Super FX chip design team included engineers Ben Cheese, Rob Macaulay, and James Hakewill. While in development, the Super FX chip was codenamed "Super Mario FX"[3] and "MARIO". "MARIO", a backronym for "Mathematical, Argonaut, Rotation, & Input/Output" .... According to Argonaut Games founder Jez San, Argonaut had initially intended to develop the Super FX chip for the Nintendo Entertainment System.

Super FX is a custom 16-bit RISC CPU designed for 3D games.

SuperFX was turned into its commercially-licensable IP core i.e. the "Argonaut RISC Core", or "ARC", which became its own business, similar to ARM Holdings: https://en.wikipedia.org/wiki/Synopsys#ARC_International

Intel used an ARC core in its Management Engine. Not only can your PC motherboard run rootkits, it can also run StarFox.

https://en.wikipedia.org/wiki/ARC_(processor)
Quote:

ARC processors employ the 16-/32-bit ARCompact compressed instruction set instruction set architecture (ISA) that provides good performance and code density for embedded and host SoC applications. ... They have been licensed by more than 200 organizations and are shipped in more than 1.5 billion products per year.

CSG/MOS 65xx CPU family has influenced the RISC families of ARM and ARC.

ARC (Argonaut RISC Core) has evolved into a 64-bit RISC CPU design via ARCv3 ISA.

RISC-V ISA was attached to ARC-V.

ARC processors became part of the Synopsys DesignWare series.

Quote:

@matthey The SNES DSP tech was primitive compared to the Amiga DSP tech but it was cheap enough to put in a cartridge while the Amiga DSP was too expensive for CBM to make standard. Software programmability requires more memory and CBM was against increasing the Amiga memory and more chips with the "read my lips, no new chips" mandate.

FYI, SuperFX is a custom 16-bit RISC CPU.

Quote:

@matthey The AT&T DSP3210 was a very nice DSP but would the Amiga have been better off with alternative enhancements instead?

AT&T DSP3210 is designed for IEEE-754 FP32 3D math with DSP audio being secondary.

AT&T has DSP16 for DSP audio.

Should I repost AT&T's marketing materials for their various DSP products?

Quote:

@matthey I tend to believe VRAM or fast mem and more integrated chipset upgrades would have been preferable.

VRAM wouldn't solve the 3D math problem.

Quote:

@matthey Specifically, upgrading the memory bandwidth and adding chunky support needed to be priorities. A DSP has to share the system memory bandwidth or have its own memory but the former is a problem as Amiga systems already had too low of memory bandwidth and the latter increases the cost without increasing the more general purpose and useful system memory.

DSP3210 has 8 KB on-chip local memory i.e. think of CELL SPU's 256KB on-chip local memory.


Quote:

@matthey It's not that I'm against a standard Amiga DSP but I believe a DSP would have been better for high end Amigas, similar to the Mac Quadra AV models, to offload the 68k CPU (a possible exception is a DSP integrated into the Amiga chipset). Someone is probably going to Hammer the number crunching superiority of a DSP over a 68k CPU but DSPs are difficult to extract performance for general purpose use.

32-bit 140 ns read/write FP DRAM that is fitted in A1200 has 7.1Mhz effective, hence 28MB/s bandwidth.

Good luck with 68030 @ 50 Mhz pushing 28 MIPS MUL instructions with 28MB/s bandwidth.

68030 does NOT have the compute intensity of RISC-based CPU/DSP.

68030 needs to be modified for intense MUL instruction processing i.e. 68030M concept.

68EC040 is brain-dead for DMA-equipped desktop computers and game consoles.

68K road map is dead for a reason.

Last edited by Hammer on 16-Jul-2024 at 04:17 AM.
Last edited by Hammer on 16-Jul-2024 at 04:16 AM.
Last edited by Hammer on 16-Jul-2024 at 04:13 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

One thing that the people like Hammer completely ignore is the context of the time: how the Amiga worked, which kind of software/games where developed for it, what was its customer base.

That's fucking bullshit.

Last edited by Hammer on 16-Jul-2024 at 04:19 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Karlos

Quote:

Karlos wrote: ... to build a completely custom sidecar expansion for the A500 with a dedicated accelerator intended to run DOOM and ideally display it in something more than 32 colours, what retail price are we looking at in 1993?

I dare say somewhere in the area of $399 - $499 USD.


Quote:

It sure as hell isn't going to be the typical asking price, or take a sensible amount of time. Commodore isn't going to stump up anything given it's already dying on it's arse by this time.

At this point, the only thing that could've helped C= climb out of their debt pit without spending more money, is if they became a mule for heroin shipments out of the Philippines.

Quote:

Could such a device be built? Probably. Would it be remotely practical given what we know of the state of the Amiga in late 1993? No chance.

Even with all that heroin mule money, no chance.

_________________
All the way, with 68k

@Karlos

Quote:

First of all, there was no "SuperFX" processor for the Amiga. The best you could hope for was an off the shelf DSP. Now you have to display what is being rendered. Is that going to pass back into the 32-colour OCS chipset? How? Or are you going to have to add some sort of minimum framebuffer implementation as well and a suitable display output? That hardware is starting to sound more and moreo cumbersome by the second.

It also wasn't exactly appropriate for Doom or suitable. I mean it's optimised for rendering polygons. Doom wasn't a hardware 3d game. It was an "old fashioned" software renderer designed for PC computers using commodity graphic cards. It was optimised for VGA tricks where multiple spans could be written in parallel based on the framebuffer planes.So even SuperFX wasn't exactly suitable for Doom either. It could transform 2d bitmaps so all the 2d warping could be converted. But, it's like doing X to produce a result of Y and creating Z to get it all working, if that makes sense at all.

Quote:

Well, if we are talking a dedicated 3D accelerator on the sidecar, additional RAM and an RGB port expansion for displaying pseudochunky displays, why not throw in the requirement for a serial port MIDI rompler for the music?

At that point it's another external box and pointless. Unless the MIDI and a 32 channel MIDI hardware chip can be included inside one box it's no good. One box to fix it all.

@Hammer

Quote:

MIDI is unimportant, i.e. Atari ST's 2 million install base says Hi.

MIDI put the Atari ST on the map. The ST had Cubase. The Amiga didn't so end of story there.

On the Amiga you have people sharing stories of ProTracker. ProTracker is fun but real musicians don't use ProTracker. The Amiga is good for an instrument but it's not good using to play all your audio. That's ridiculous. Where are the vocal tracks going to be? Real musicians used MIDI and a MIDI box on an Amiga. Samples and sampling was a bonus.

ProTracker and other trackers also changed Amiga game music. It wouldn't be the same without it. But in some ways it was also worse. For example, in the beginning you had these recorded sound tracks, what the Amiga was made for. A good example, in a bad game, was Thunderboy. This is one game that even features a guitar. Now, with memory limitations, you're forced to create a loop so the music can only be short. A tracker uses smaller samples sequenced together so it can produce a longer song. But, imagine, if instead of a natural guitar recording, there is a guitar sample played at different notes to simulate a solo. It's just not the same as a real recording. But we did see some fun examples, like Zool and Metal Machine from Lotus III.

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: One thing that the people like Hammer completely ignore is the context of the time: how the Amiga worked, which kind of software/games where developed for it, what was its customer base. That's fucking bullshit.

"You can't handle the truth".

But don't worry: as I've already said, I'll write another article that further proves it (since you were not able to understand the other three).

@Hammer

Quote:

Hammer wrote: @Karlos Quote: Karlos wrote: @Hammer You are just waffling utter gash. Bullshit. Read the fukcing "Commodore The Final Years by Brian" for CSG component BOM prices.

I don't need to. You don't get it. Your list of BOM prices and random SKU transistor counts does not magically yield a solution to the "sidecar A500 doom expansion" and certainly for nothing remotely like the retail price of the game! It doesn't cover the R&D costs, licensing costs, development costs, projected sales return etc. It shouldn't be remotely necessary to explain this.

SNES got a doom port because it already had the SuperFX2 expansion and id software were shown a technical demo of a comparable renderer, to prove viability.

Amiga had literally nothing by this time. Perhaps if the CD32 had a better Akiko solution, fast ram and a beefier processor, a similar port could've been developed. But that's the CD32, sales of which are a small fraction of the number of basic OCS/ECS machines still in use in 1993.

_________________
Doing stupid things for fun...

@matthey

Quote:

he ID guys were Apple/Jobs/Mac fans and clearly biased against the Amiga.Â

Not true. They never refused a license to port it, they were never approached for one. You can't expect them to decide to do it themselves, they had enough work on the next big thing. Carmack saying the Amiga (as the installed user base of 1993 goes) couldn't run it was probably a key fact there but he wasn't incorrect in context.

If they were truly biased against the Amiga, they would've turned down Clickboom's licence bid. Quake, like Doom before it was prestige IP to the company.

The 68K Amiga port of Quake was basically the second in line after Mac and before any consoles.

_________________
Doing stupid things for fun...

matthey Quote:

the ID guys were Apple/Jobs/Mac fans and clearly biased against the Amiga.

Karlos Quote:

Not true. They never refused a license to port it, they were never approached for one. You can't expect them to decide to do it themselves, they had enough work on the next big thing. Carmack saying the Amiga (as the installed user base of 1993 goes) couldn't run it was probably a key fact there but he wasn't incorrect in context.

The competition between the 68k Amiga, Atari and Mac wasn't always friendly. Amiga users didn't feel threatened as much by the Mac or Atari PCs as we had the superior hardware, at least until we realized CBM was the bigger threat than the competition. Imagine the horror of a Mac fan in 1985 with their black and white display, mono audio and what they thought was a fancy mouse driven GUI and OS when the Amiga came out with hardware accelerated 4096 colors, 4 voice stereo sound and a mouse driven GUI and OS with preemptive multitasking. The first thought had to be that CBM was going to put Apple out of business and some Apple fans reacted by attacking the threat.

It is obvious that id software developers were big Apple fans from posts I made in this thread.

http://rome.ro/2006/12/apple-next-merger-birthday.html Quote:

Apple-NeXT Merger Birthday! It's been 10 years since Apple merged with NeXT and then later brought us the awesome OS-X. I still remember seeing OS-X in action for the first time and how the Finder displayed directories in NeXTSTEP format and the rotating CD for the new Wait cursor.....YES! I was totally convinced at that point that some version of Mach was at the core of OS-X. Jobs had brought The Power back home. At that point NeXTSTEP had morphed into Cocoa - the development environment that Mac coders live in. Check out the celebration at NeXT Computers.org and browse the forums to find out more about this legendary company, their hardware and their unsurpassed software. Why do I care so much about NeXT computers? Because we at id Software developed the groundbreaking titles DOOM and Quake on the NeXTSTEP 3.3 OS running on a variety of hardware for about 4 years. I still remember the wonderful time I had coding DoomEd and QuakeEd in Objective-C; there was nothing like it before and there still is no environment quite like it even today. When id Software was stationed in Madison, Wisconsin during the winter of 1991, most of us were gone for the Christmas holiday - except John Carmack. John's present, which he bought with $11,000 of his own money, procured by walking through the snow and ice to remove from the bank, arrived during the holiday and he spent the whole time learning as much as he could about the computer and started working on vector quantization algorithms for compressing graphics. His test graphic was a 256-color screen from King's Quest 5. After his research was done it was agreed that the entire company needed to develop our next game on NeXTSTEP. ... As I was leaving id Software in August 1996 the move to the Windows 32 platform was underway. John Carmack was porting our QuakeEd editor over to Win32 and preparing for a NeXT-less future. Several short months later NeXT made their fateful move over to Apple and a new era was begun as Steve Jobs set about changing the future. Again. Up to that point I had spent 15 years of my life working on computers that Steve Jobs was involved in bringing to the world. First the Apple II+, then the IIe, the IIgs and finally NeXT. Maybe someday I'll get one of those kickass iMacs.

There is the famous e-mail response from Carmack about the Amiga too.

https://groups.google.com/g/comp.sys.amiga.games/c/MZb9cC0FMhw Quote:

[original letter to Id Software] Hi. I would appreciate an answer to this letter from you or someone who is able to do so... I noticed that you have recently released a version of Doom for SGI, meaning that porting isn't a difficult task. Perhaps a port of Doom for Amiga would also be a good idea ? I realize that the Amiga lacks a chunky graphics mode and it is a relatively limited market compared to IBM clones, but the game market for Amiga is quite large. The lack of the chunky graphics mode has been solved via fast conversion routines (which can be found in ftp.wustl.edu:/pub/aminet or any other Aminet mirror) or as in the case for Amiga CD32, the conversion routine is provided in hardware. There are numerous texture mapping demos available showing that a Doom-type game is possible on the Amiga. The market for Doom on Amiga is also fairly large. The Amiga CD32, which is basically a games console with a CD drive built in, together with the SX-1 expansion unit could provide sufficient memory and speed requirements. Many owners of Amiga 1200 have upgraded their systems with high speed accelerators, bringing their machines performance similar to the A4000, which has more than enough horse power to handle Doom. These machines mentioned are the ones equipped with the AGA chipset, and by including the old generation, (but still fast) A3000, the market is large enough for a port of Doom to take place. If you have any further specific questions, I would be happy to answer them. George Sanderson ---- [response from Id software] From jo...@idcube.idsoftware.com Sun Sep 4 02:52 EST 1994 From: John Carmack Date: Sat, 3 Sep 94 11:50:23 -0600 To: G.San...@ais.gu.edu.au Subject: amiga doom The amiga is not powerful enough to run DOOM. It takes the full speed of a 68040 to play the game properly even if you have a chunky pixel mode in hardware. Having to convert to bit planes would kill it even on the fastest amiga hardware, not to mention the effect it would have on the majority of the amiga base. John Carmack

Carmack's statement was not true as even a high clocked 68030 with fast memory is adequate for Doom but it seemed to be effective at discouraging Amiga interest. It may be true that nobody tried to license Doom for the Amiga but Carmack statements likely discouraged it from happening. There never was an official Amiga port either but it was eventually released as freeware and was finally ported to the Amiga.

https://archive.org/details/cuamiga-magazine-097/page/n17/mode/2up Quote:

Back when the Doom erase was at its peak. Amiga users campaigned for its release for our platform. Doom originated on the PC and has been ported to the Mac, various Unix platforms, as well as most modem consoles with varying degrees of success. The developers, ID Software, were unwilling to undertake such a port themselves, feeling that the Amiga gaming market did not have sufficient horsepower overall to support Doom. They may have had something of a point, as many users in 1992-1993 were using base machines, and graphics cards were still out of the reach of most users. They were unconvinced that the fairly common 030 machines were sufficient to play Doom - they would later be proven wrong, but that's getting ahead of the story. So, id was unwilling to port the game themselves and nobody stepped up to pursue a license. The years passed. Doom was ported to the various platforms, and id moved on to Ouake, which has its own storied history for the Amiga. Not long ago, id realized that Doom's value for future sales had more or less run its course, and released the source code as freeware, encouraging not only Amiga users but hackers worldwide to fiddle to their heart's content with the heart of Doom.

While the size of the 68k Amiga user base shrunk after the death of CBM, I expect the average performance and spec increased with accelerators, memory upgrades and graphics cards. I expect this continues today although the user base may not shrink and start to expand again with affordable enough 68k hardware instead of throw away emulation toys. It's not like the A500 Mini has increased the 68k Amiga user base by hundreds of thousands of users like real 68k hardware with competitive value could.

Karlos Quote:

If they were truly biased against the Amiga, they would've turned down Clickboom's licence bid. Quake, like Doom before it was prestige IP to the company. The 68K Amiga port of Quake was basically the second in line after Mac and before any consoles.

At least the Sega Saturn port was before the Amiga release and the N64 port was the same year. The 1996 N64 CPU was unusual for a console as it was 64-bit, high clocked and had a FPU. (the successor GameCube went back to a 32-bit CPU but higher performance and a FPU soon became standard for consoles).

Quake ports
1996 Linux, SPARC Solaris
1997 Mac OS, Sega Saturn
1998 N64, Amiga

The problem for consoles was that most still used fixed point integers instead of floating point for 3D (usually no FPU). Software fp is easy but low performance while converting from fp to fixed point is a major undertaking and helps explain why PS1 and Jaguar ports were not released. Desktop PCs had standard FPUs for awhile but older CPUs lacked the FPU performance to run Quake like the 68040. Amiga 68060 accelerators came after CBM went bankrupt as the remaining 68k Amiga spec increased and the 1994 68060 was adequate for Quake, despite the inconceivably low clock speed rating for an 8-stage pipeline. Even though the CBM successor Amiga Technologies sold 68060 Amigas and there were at least five 68060 accelerator designs, the 68060 Amiga market was still likely too small at that time (there are at least 5 new or planned 68060 accelerators today not including the old 5). PPC Amiga1 was a shared ClickBOOM Quake target that had standard FPUs until the A1222 but that market ended up being tiny. I expect there were more 68060 Amigas than PPC Amiga1 systems then and today. While the number of 68060 Amigas have grown today, they are just a fraction of the high end 68k Amiga market despite the lack of competitive hardware.

Last edited by matthey on 16-Jul-2024 at 11:20 PM.
Last edited by matthey on 16-Jul-2024 at 10:48 PM.