• Please review our updated Terms and Rules here

Video: Is Doom playable on a 386?

Someone at school had Doom shareware and I was fortunate to have just gotten a new machine, an Ambra 386sx (25 or 33MHz I think) and other than needing to wipe off nearly everything to get room enough to install it, we were all stunned by it. It had to have a bit of a border around the screen, but perfectly playable and like nothing else we'd seen before on the PC platform.
 
doom on a 386-dx 40 is definitely do-able. Feed it 8 or 16mb ram, and give it 256kb cache and even with the generic isa vid cards, it will fly like a champ. Doom 2, not so much, especially when you hit levels 15,16, and 30. Level 15 is the WORST simply because of its size.

the next worst level on a 386-dx 40, is the level called Go-2-It, simply because of the amount of enemies, and the archviles constantly resurrecting everything.

Don't get me started about it on a 486-slc type chip.....
 
Feed it 8 or 16mb ram, and give it 256kb cache and even with the generic isa vid cards, it will fly like a champ.

I don't see this happening. The board I used has 256KB Cache and the graphics card is also quite a fast one. Maybe with a 486 upgrade processor you could make it run a bit better, but still. But I'm happy to be proven wrong. Eager to see other people's 386 beasts and some benchmarks, or videos of your machines running Doom :)

Personally I go straight with a Pentium. Even a 486DX2 or IntelDX4 will run into slow-downs in certain levels / areas.
 
That's true, I left the graphics at the default setting, which is 2 screen sizes down from full screen I believe. Looking forward to seeing some Doom action on a DOS PC :)
 
My 386/40 w/256K Tseng 4000AX:

Doom-1: Loads and runs - no problems
Doom-2: Loads and runs, but real slow w/tiled video at times
Wolf3D : Loads and runs - no problems
Duke3D: Won't load on my 386

So, it kind of obvious the ID software moved on with the times. There may be a hack floating around but I haven't seen it.

P.S. Just because Wolf won't run doesn't make me want to put my 386 back on the shelf.
 
I seem to recall playing doom on my 486SX-25 with 4MB of ram with a Sound Blaster 2.0 and it ran fine. In a lot of cases the 386DX-40 is faster, so I'm surprised to see this machine not working as well. Have you tried changing the number of mixed sound channels to see if the speed picks up? I'm curious as to how fast you can get it to run without removing audio completely and making the game screen the size of a postage stamp.
 
I seem to recall playing doom on my 486SX-25 with 4MB of ram with a Sound Blaster 2.0 and it ran fine. In a lot of cases the 386DX-40 is faster, so I'm surprised to see this machine not working as well. Have you tried changing the number of mixed sound channels to see if the speed picks up? I'm curious as to how fast you can get it to run without removing audio completely and making the game screen the size of a postage stamp.

- 486SX-25 runs Doom faster than a 386DX-40:


- I don't think any tweaking will get Doom to run well on a 386. Hey maybe the sound costs 10% performance, but when you frame rate is already 10 fps, that makes bugger all difference
 
I watched that video as well. I recall Doom running better than that on my 486SX-25. Since I have it back I will have to run the tests and see what I get. I'll post what I get using your benchmark suite on your website when I get a result.
 
I think as well our expectations were (significantly) lower in 1993.

Exactly! I think the gist of this thread was if Doom would run at all on a 386. Back in the day, if you had the 386 and the other guy had the 286, well who's going to win that round?
 
I watched that video as well. I recall Doom running better than that on my 486SX-25. Since I have it back I will have to run the tests and see what I get. I'll post what I get using your benchmark suite on your website when I get a result.

That would be awesome :) Do not however that I used an ISA card to better compare it with the 386DX-40 and give it a fighting chance :). So I would say this represents a fairly average 486DX-25, but it still beat the 386DX-40 by quite a bit. VLB or PCI systems would score even higher, but if you've got such a system, are you really going to bother with a SX-25?

On my VGA benchmark database project, with ~ 500 computer entries, you can see that it is with DX4 machines when the benchmark starts hitting 35 fps.

The fastest 386 entry gets around 7 fps. And that's with an overclocked ISA bus.

I think as well our expectations were (significantly) lower in 1993.

Exactly! I think the gist of this thread was if Doom would run at all on a 386. Back in the day, if you had the 386 and the other guy had the 286, well who's going to win that round?

Maybe the title could have been worded better. Today, would you play Doom on a 386? or something like that. I mean, of course people played it on a 386 back in the day, that's all we had. But doesn't change that the game is a lot more fun on a faster PC :)

But maybe people are into playing it like in the old days.
 
Last edited:
Hmm. Running the benchmark suite from your webpage I got 2134 gameticks in 5333 realtics which following your way of identifying the frame rate comes out to 14 frames or so. Your benchmark also is full screen with no health bar and the window is not sized down. How is it your 486 ran slower than mine? PCBench got a score of 6.0 at 320x240 (or 200 can't remember which) and a 3d Bench of 25.2 which seems kinda average. Wait - I've found the problem. Running cachechk indicates that the processor is running at 33.4MHz. RAM is really slow read access time is 210ns and write access time is 127ns with a speed of 19.9MB/s and 52.6ns/byte reading. Going to open the machine tomorrow and look more carefully at the chip. My sister had this after my dad and my stepmom got a new machine so there is no telling what was done to this computer. I think the 486 processor has been swapped out with something faster..
 
Well, once again you didn't benchmark it in low-res mode, which is how we all played it on our 386dx-40 systems when it came out. If you do that, does it then match the 12fps of the 486sx-25?

But I did. I tested all 18 combinations of details and screen size :) Check out the video and / or the chart of results on the first page.

To compare machines you test at the same settings of course. Don't you agree?

Also the 486SX-25 is very average, whereas the 386 is high end. One of the fastest 386 boards you can find, 256 KB Cache, fast ISA card. The 486SX-25 a VLB or PCI card, 512 KB Cache, faster RAM timings, the lead would even grow. I gave the 386 a good fighting chance and wanted to avoid people saying "But it's got a VLB card and double the cache and things like that".

I highly encourage you to do your own tests. There are really cool hybrid boards that take 386 and 486 processors. Perfect for comparing. Testing is fun and maybe you discover something cool and interesting!
 
Last edited:
Hmm. Running the benchmark suite from your webpage I got 2134 gameticks in 5333 realtics which following your way of identifying the frame rate comes out to 14 frames or so. Your benchmark also is full screen with no health bar and the window is not sized down. How is it your 486 ran slower than mine? PCBench got a score of 6.0 at 320x240 (or 200 can't remember which) and a 3d Bench of 25.2 which seems kinda average. Wait - I've found the problem. Running cachechk indicates that the processor is running at 33.4MHz. RAM is really slow read access time is 210ns and write access time is 127ns with a speed of 19.9MB/s and 52.6ns/byte reading. Going to open the machine tomorrow and look more carefully at the chip. My sister had this after my dad and my stepmom got a new machine so there is no telling what was done to this computer. I think the 486 processor has been swapped out with something faster..

Great effort for testing this. Going from 25 to 33 should give you a nice boost. The low bus speed of 25 MHz really hurts the SX-25.

The SX-25 seems to be ugly duckling of machines, but I really like it as a retro gaming PC.
 
Last edited:
To compare machines you test at the same settings of course. Don't you agree?

True, but the topic-title asks if Doom is playable on a 386. So I guess the answer to that question is which settings are the best compromise of speed and quality for a 386 in terms of playability.

I personally don't think VLB/PCI matters much in Doom by the way. It has a very efficient renderer, no overdraw. When I bought my 486, I recycled the Paradise ISA card from my 386SX-16 for a while, until I could afford a decent VLB card (which became the CL5426-powered Diamond Speedstar Pro). In Doom at least, the difference was very minimal. And I know a friend of mine who had an ISA ET4000 in an early 486-system which didn't have VLB yet, and that thing was faster in Doom than even some real VLB cards, such as a Trident. Its framerates were as near my Speedstar Pro VLB as makes no difference.

I guess that would be interesting tests indeed... Running with ISA vs VLB vs PCI.
 
Sure you can lower the detail and screen-size on the 386 to match the speed on the 486. But then you can do the same thing on the 486 and get even better performance. Which brings us back to square 1. Either way, it's "more" playable on the 486, be it faster speed, or better graphics, depending on what you prefer.

I've got an Acer board from an old Acer OEM 486, that one has ISA, PCI and VLB :) It also has a decent onboard card, ATI Mach 32. I believe what I found was that on a 25 MHz bus, the differences aren't that great, but as you go higher the difference becomes greater and eventually the ISA card becomes a graphics de-accelerator :)
 
Back
Top