• Please review our updated Terms and Rules here

Video: Is Doom playable on a 386?

Now that my curiosity is piqued I'm running a series of benchmarks on my machine to see what the performance differences are in regards to doom. As soon as I'm done I will post my results here in a rtf file, if that is acceptable.
 
High Quality Benchmark

High Quality Benchmark

Okay, I've finished benchmarking Doom in high quality with all the different number of mixed sounds. Unsurprisingly, The Audio has an almost negligible impact on the game. Difference between 8 channels at full screen versus 1 channel at full screen is just shy of 2 percent. Dropping the screen size is more beneficial. However I noticed that when mixing 7 channels versus 8 channels there was almost always a negative impact in performance. so I would run the demo three or four times and it still showed worse performance than 8 channels. Very strange. Here is the total benchmark. I apologize for no graphics or bar charts, but I just don't have the time currently. Once I finish all the testing and have the time, maybe I can put everything into a spreadsheet and bar chart it. Benchmark - View attachment doom_high_qual_test.txt
 
philscomputerlab said:
Excellent work! So now we know that the voice mixing setting has little impact. Good to know.
Click to expand...

Technically what we can say is that voice mixing has little impact on a 486. I regrettably don't have a 386 to test on. It's on my list to get. Need to reduce certain sections of my collection first though...
 
fallemarg said:
Once I finish all the testing and have the time, maybe I can put everything into a spreadsheet and bar chart it. Benchmark - View attachment 25363
Click to expand...

Here's a horrible chart I made using your results:

pC5t1aD.png


-1, -2, -3, etc. indicates how many size downs there are. The higher the number, the smaller the render window is.
 
I must say, what surprises me the most about the results is how slow Doom actually runs. I'm almost finished with the Low Quality tests and the results are higher, but probably not as much as some might expect. I should also preface that this is benchmarked utilizing a 486DX-33 with 8MB of RAM, no cache, with a Sound Blaster Pro 2.0 and MPU-IPC-T outputting midi to a Sound Canvas SC-55 MKII. I don't know if using the Sound Canvas introduces any additional slowdown or not, but I am guessing that others mileage will vary. I'm thinking I might try to get my hands on a 486DX4-100 and see what that's like for performance on this system. I've encountered the dreaded graphics corruption and trails sadly making Descent pretty much unplayable because of the pixel issue. I seem to recall that there was a fix for this for the HT216 chip and some of the Video7 chipsets, but I cannot seem to find any information anywhere on the internet. I remember we had a utility that I ran on games where the graphics screwed up and they worked better. Not perfect but better. Have to see if I can dig that out of the old floppies I have. Yes, sadly I still have all of the floppies from when we got this computer new in 1993.. I can be such a packrat..

-Edit-

Thank you so much for making that graph chart GiGaBiTe! It looks very good. Pretty gentle curve until you hit the No sound test.
 
fallemarg said:
I must say, what surprises me the most about the results is how slow Doom actually runs. ..., no cache
Click to expand...

No cache probably has something to do with that I guess?
All I remember is that I had a 486DX2-66 back in the day, with 256KB cache and a zero-waitstate VLB CL5426 videocard. It ran smooth as butter on that system.
I still have that system actually, except it is now clocked at 80 MHz. Perhaps I can dig it out and see what it does. It has both an SB Pro 2.0 and a GUS MAX in there. I used to play with the GUS in MT-32/Sound Canvas mode, because I liked those sounds better than the default GUS instruments that came with the game.
 
Doom is likely quite cache sensitive. A DX2-66 is a very good match for Doom. Some levels can cause a bit of slowdowns, but you get that on a IntelDX4 as well. Pentium should be smooth throughout. I believe the game uses V-sync at 35 fps / 70 Hz with the monitor, so any extra performance is a nice buffer for these demanding levels. Or for Doom 2 :)
 
It took me longer than I had expected to finish this but I got it finished. Here are the results for Doom on a 486DX-33 no l2 cache 8MB RAM on a Headland Technologies HT216 VLB chipset with 512KB RAM with audio through the Sound Blaster Pro 2.0 and Midi provided through a MPU-IPC-T connected to a Roland Sound Canvas SC-55 MKII. One thing of note, is even at the lowest quality with the smallest view for the game, the difference between 8 channels and 1 channel of audio is less than 10 percent. I assumed with a greater reduction of requirement on the CPU that the difference in the sound mixing would increase and that does appear evident but still it does not have as large an impact as reducing the view size.

Interestingly, going through the utilities provided with the packard bell there is a tool called altparm which allows you to modify screen modes. I wonder if changing 13h will cure the pixel crawl and corruption issue? I think it's fairly safe to try because it does have a reset to default settings option. I may have to fiddle around see what happens. So attached here is the benchmark text file. Thanks for sticking with me while I finished this --View attachment Doom_low_qual_test.txt

-Edit-
I just checked the cache chips I ordered and the video ram and both indicate that they are out for delivery to me today, so I may test 8 channel and 1 channel again in all screen sizes and both qualities to see if there is a marked difference in regards to l2 cache. That is, if the cache chips I purchased actually work..
 
Last edited:
Updated chart with the lowres modes:

cKbIR1W.png


fallemarg said:
I'm thinking I might try to get my hands on a 486DX4-100 and see what that's like for performance on this system.
Click to expand...

I don't have a DX4/100, but I have an Intel DX2/66, AMD DX/2 80 and an AMD DX/2 120 if you want to pay shipping for them. I can't guarantee they work though because they've been sitting in a box for 15 years, and the pins are slightly bent so you'd have to spend a bit of time with a flat edge on them. They're not so bad that they're at right angles or anything.
 
Sadly, the packard bell doesn't support 40Mhz FSB so I don't know if those will work at all. Thank you for the offer. I have another board that will take them, but it is in storage right now with no idea on when I will pull it out again. Granted storage is a watertight sealed container in the basement on the storage shelves, but it will be difficult to get to.
 
I know the AMD DX4/120 can take a 33 MHz FSB and run at 100 MHz, but I'm not sure if it works on all motherboards. The motherboard I used it in was one of those really cheap chinese knockoff boards with fake cache chips and it seemed to run with pretty much any setting for FSB and multiplier.

I would test it again, but that system is long gone. I only have a dinky AST Advatage! machine now and it only takes up to a DX2/66 or a P24T without using one of those special upgrade chips.
 
The main issue for testing a DX4 chip is that they all run at 3 or 3.3V and the information I have doesn't indicate if there is a jumper to change the voltage supplied to the CPU and I definitely don't want to fry the CPU.
 
fallemarg said:
The main issue for testing a DX4 chip is that they all run at 3 or 3.3V and the information I have doesn't indicate if there is a jumper to change the voltage supplied to the CPU and I definitely don't want to fry the CPU.
Click to expand...

If you run a 3 volt rated DX4 at 3.3 volts it's within specs. Not the same when attempting to run a DX4 rated at 5 volts on a 3.3 volt setup.
 
Yes, but the Packard Bell Legend 115 plus I am using appears to be only a Socket 2 board, and as such does not support any voltage other than 5V. At least, that's the information that I have found. But it is a Blue Socket that indicates it supports Pentium Overdrive chips. I would guess that means it's only Socket 2, but I can always be wrong. You know what they say, failure is always an option.
 
fallemarg said:
Yes, but the Packard Bell Legend 115 plus I am using appears to be only a Socket 2 board, and as such does not support any voltage other than 5V. At least, that's the information that I have found. But it is a Blue Socket that indicates it supports Pentium Overdrive chips. I would guess that means it's only Socket 2, but I can always be wrong. You know what they say, failure is always an option.
Click to expand...

The POD, whether it be a 66 MHz (25 FSB), or 83 MHz (33 FSB) adjusts for whatever voltage, i.e; 3.0v, 3.3v, or 5v. Personally, I was not overwhelmed by the so called performance gain. My POD is in its original box, sitting on the shelf. BTW, you can always find a CPU voltage interposer floating around on Ebay.
 
Just my two cent here:

Doom (II) was more or less unplayable on an 386SX/25 using a OAKtree chipset OTI-67 / -77. Some indoor Levels have been quite smooth but outdoor ones like in DII I think Level 13 or so was not really playable with smallest Windows and low Details.
cheers
 
You must log in or register to reply here.
Back
Top