Doom Clone Shows What An Alternate-Reality Amiga Could’ve Had

Can you run Doom on the Amiga? No, not really, and arguably that was one of the causes for the computer’s demise in the mid-90s as it failed to catch up on the FPS craze of the PC world. [Krzysztof Kluczek] of the Altair demogroup has managed not exactly to remedy that status with the original article, but to show us how a potential contender could’ve been designed for the unexpanded Amiga hardware back in the day.

Many developers tried to emulate the thrill and ambiance of the id Software shooter, but they all required high-end Amigas with faster processors and expanded memory, limiting their player base on an already diminished demographic. Not only that, but even with fancier hardware, none of them quite managed to match how well Doom ran on your run-of-the-mill 486 at the time. [Krzysztof] isn’t trying to port Doom itself, but instead creating an engine custom-designed to take advantage of, and minding the limitations of the OCS Amiga as it existed in 1987. The result is Dread, a 2.5D engine that resembles the SNES port of Doom and uses assets from the Freedoom project in order to remain copyright-abiding.

It might not be Doom, but it’s a good peek at what the 33-year old hardware could’ve done in the right hands back then. Technically it already surpasses what the Wolfenstein 3D engine could do, so there’s an idea if someone ever aims to make a straight up port instead of their own game. If you like seeing Doom run on machines it wasn’t meant to, boy do we have some posts for you. Otherwise, stick around after the break for two videos of Dread’s engine being demonstrated.

[Thanks Janusz Magrian for the tip!]

26 thoughts on “Doom Clone Shows What An Alternate-Reality Amiga Could’ve Had

  1. This is incredibly impressive and I like to learn more on how it’s done at the bit plane level. Wolfenstein/Doom shattered the Amiga’s dream, as it didn’t have the addressable pixels that made the game possible on PC. In the 1990s everybody and their mother cracked their head on how to workaround the Amiga bit plane graphics.

      1. I think it would be even more interesting if you also took as much time to see how much better Doom could have been on the PC by creating a new engine on the PC with an alternative highly optimised strategy too.

        It’s 2020. Doom was 1993. What could a PC Doom variant have been like if somebody had spent 27 years optimising it?

    1. It’s amazing how people kept saying “the Amiga is superior because of its distributed parallel architecture” or something to that extent well into the 2000’s – when it’s exactly this approach of specialized chips that made it impossible to upgrade and difficult to program effectively. Sure, some of the concepts were later adopted back on the PC side, out of necessity, but still you don’t split the function of a GPU into three different chips where each draws different parts of the graphics while also dealing with DMA and disk IO.

      The entire point of the PC was that it had raw generic processing power in a single chip, instead of hacks and tricks that would only work in special cases and combinations.

      1. Also, remember, the OCS was supposed to be part of a third generation game console, where you plug in cartridge roms with a simple CPU that programs these chips to play games similar to the 8-bit NES. With the video game market crash, they put in a Motorola 68000 instead to make it into a “home computer” and it became this half-and-half frankenstein of two different system paradigms.

        It was such a weirdo that the marketing department couldn’t figure out what it was, so they called it a “multimedia computer”. After all, it could do multiple things almost as well as machines designed to do those things from the start, but it was never really best in any of it. It was simply the best you could afford to buy at the time, so people had the illusion that it’s a superior platform.

        In reality it was a compromise in all of its qualities, and the hacks you had to pull off to make it do something was like punching your way out of a mailbag: when you extend your arm out in one direction, the bag closes in on you from all the other sides.

        1. For example, Amigas were often praised for their music/audio capabilities, but few people remember that it didn’t actually have MIDI support – it was hacked in using the serial ports, and it suffered from bugs with the hardware timers, which meant that it would sometimes drop notes or crash the machine. That’s why it could not really be used for serious music production, although you could dick with it in your home studio just fine.

          The tracker scene emerged of course, but since the machine didn’t actually have the memory or processing power to handle high resolution samples and effects, it too became mostly just people dicking around making horrible techno and trying hard to emulate proper music. Now it seems like a legit thing because we’re having chiptunes nostalgia, but back in the day it just didn’t work.

      2. Interesting theory… but the PC didn’t initially HAVE raw generic processing power, 68000 series kicked x86s ass at MIPs per MHz. The first generation Amigas had practically double the MIPs of IBMs top of the line 8Mhz 286 AT at launch… The A3000 in 1990 launched with more MIPs in it’s base model 16Mhz ‘030 than a 33Mhz 386DX, the A4000 in 1992 with the MIPs of approx a 486DX2-66. The A1200 also launched in ’92 with the MIPs of a high end 386, low end 486 for the price of a used 286. Bolt on accelerator upgrades could take practically any of these machines up to the performance of the fastest. However, the IPC of Intel CPUs improved, and their frequency scaling was better than Motorolas, so from 93 with the release of the P5 Pentium they actually began to move ahead. Arguably the 68060 was competitive, but never available in enough quantity, and Commodore was by now falling apart, and did not get one out the door, Escom did in ’96, but by that time 100Mhz plus pentiums made it look slow. During the same timespan, high end professional graphic workstations and Macs were using 680×0 for the same reason, x86 did not have raw generic processing power.

        1. Interesting theory? What you have written is a popular Amiga fan fiction.

          First of all, MIPS is known to be a pretty useless speed measurement because you can speed things up a lot with instructions which don’t don’t do anything but do it quickly. A Commodore 64 will develop about .5 of a MIP if you don’t do anything and just chain together a massive load of No Operation instructions. What it generates in normal workloads is much smaller. Also, if one were to write a Floating Point maths library for the Commodore the MIPS would still be about average, but this would of course set aside the fact that the FLOPS generated by that library would be extremely low so it gives really no idea at all of workload.

          More accurate benchmarks have been written which provide various CPUs with a specific take, such as integer or floating point operations, and try to optimise them similarly to make them fair.

          In fair realistic figures from actual real benchmarks there was never a time, not ever, when the Amiga competed with the PC.

          Let’s take a look at actual CPU performance results which are not just a fiction made-up by Amiga fans:

          If we try to make the contest as fair as possible by running the same benchmark generated by the same compiler on both platforms the nearest we can get is this test which runs the same Lattice C compiler series, but on the PC the version, Lattice C 2.15, is a generation BEHIND the Lattice C version 3.02 running on the Amiga:

          The Amiga/68000 generates about 51.81 Dhrystones per Mhz.
          The 8088 generates about 95.17 Dhrystones per Mhz.
          The 8086 generates about 102.72 Dhrystones per Mhz.
          The 80286 generates about 223.125 Dhrystones per Mhz.

          So when the Amiga firs appeared, ie, at the effective Zenith of comparisons between the Amiga and the PC, the PC was already more than four times as fast. Taking Moore’s law of speed doubling every 18 months as the measure of a generation at the time, the PC standard PC AT CPU performance was almost 3 generation ahead of the Amiga even before the day the Amiga was released.

          On that same compiler benchmark there was only one PC ever to be slower than an Amiga and it was an HP portable running a cut down low voltage 8086 designed for portables running an even earlier 2.14 edition of the Lattice C compiler

          Amiga fans are the most dishonest, unreliable, naive, arrogant fantasist commentators on computer performance and literally care nothing about just making things up to eulogise the Amiga.

          The only thing the Amiga ever really had an edge on was being able to copy and paste rectangular regions of RAM from one place to another. Great for sprite games and moving windows around, completely useless for general purpose processing.

  2. There were some Amiga Doom ports when the source code was first released, although they basically require either an upgraded machine or an A4000 060.
    In 1995 there was a Doom clone called Gloom, although it didn’t have multiple heights, so in some ways it’s closer to the original Wolfenstien.
    Alien Breed 3D was closest to being a proper Doom clone, but it was released at the tail end of the Amiga’s reign, and again, it required a beefy machine to run at anything above postage-stamp-sized screens.

    1. As to my knowledge in 1995 Polish company Arra-Kis made Cytadela which minimal requirements was A500 and 1MB RAM. Later on there was PC version released as open source project The Citadel.

      1. Yes, and I mention them in my channel intro as one of the games that could run on A500. Sadly, screen size had to be reduced to a poststamp to talk about any remote playability. But it was still fun and made my days when I was a teenager.

  3. “Can you run Doom on the Amiga? No, not really, and arguably that was one of the causes for the computer’s demise in the mid-90s as it failed to catch up on the FPS craze of the PC world.”

    Mostly incorrect. There were plenty of Doom ports, and Doom really only needs a fast m68030 or an m68040 to run decently.

    Also, the failure of the Amiga had much more to do with the mismanagement of Commodore than the Amiga itself. Commodore went out of business when demand was very high for the then-contemporary Amigas (the 600, 1200 and 4000).

    That said, the Amiga has always been an interesting example of how to do things simply – less bloat, less junk, more doing. It still is :)

    1. Some VGA cards were dog slow. They didn’t have much work to do other than write pixels to the screen, but if they were slow, it made the game a slideshow. It could also happen to reasonably good chipsets paired with slower (cheaper) memory, to keep BOM cost low. Decent VGA card should have allowed you to play bearably with one level of border on that. Sometimes caching the VGA BIOS helped, sometimes it didn’t, sometimes it used up just enough extra RAM that DOOM would refuse to launch with a bare 4MB (Depending on what else the motherboard cached)

        1. Details have got a bit foggy with not messing with it for a few years, but it was something like it managed on 4096kB with up to 128k used for BIOS shadows, but if your board was a prick and shadowed the whole 384k UMB, or shadowed a network or other option ROM too, it wasn’t enough. Also it needed quite a bit of conventional memory to launch, so you had to get all your drivers and TSRs optimised. 580kB is sticking in my head for something, but that might have been another game.

          1. Oh, some games that said they needed 8MB actually worked on 6, we got one working on 5 once, but not remembering what that one was. For laughs, ran Duke Nukem 3D, on a 386sx40 under windows 3.11, for the virtual memory, because it only had 2MB, of course it was agonisingly slow, think it took an hour to get into the game, then to play the intro piece took another 5 mins.

    2. Why would you play it on a 386SX-25? It’s a cut-down 386DX. The 386DX was released in October 1985. It predates Doom by eight years. When Doom was released in December 1993 the 486 was common. The 486DX-2 66 had been released the previous year. The Pentium 66 had been released in March 1993.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.