more like Virge VX vs. GeForce 4 and to say that the GF is faster than a Virge VX at DOS graphics is an understatement.

Does DOOM (or any of the the other older DOS WAD based games that run in 320x200 8-bit) actually benefit from some of the more modern video cards (especially when ran on higher-end machines)?

Just a-wondering; as I've always thought that even with a cheaper video card, on an extremely fast CPU the difference would be nominal.

BTW, if you (or anyone else) has DOOM II v1.9, attached is a handy little utility that I threw together eons ago. It completely automates the framrate testing, and sets up DOOM II the same way each time the benchmark is run 😀

Yes, I do have DOOM II. I'll use the utility.

@swaaye : I've yet to use the ViRGE DX 4MB which is lying around. I'll post it's results soon. This will be interesting, going against the VX, that is.

ViRGE DX lags slightly behind the VX. I thought the DX would have performed a bit better.

I don't know what kind of limiter is at work here, but these results don't seem to reflect real hardware performance.

For DOOM 1.9, timedemo 3 on my K6-III, ATI Rage 128 AGP I get:
3863 / 835 * 35 = 161.92 FPS 😕

Update:
Voodoo5 AGP: 3863 / 834 * 35 = 162.12 FPS
Trio64V+ PCI: 3863 / 1432 * 35 = 94.42 FPS

I'm not sure what we're benching here...

Does anyone have both AGP/PCI versions of the same chipset?
It would be interesting to see 😀

Also may need Fastvid to get the most VGA speed out of that P6.

swaaye wrote:

Also may need Fastvid to get the most VGA speed out of that P6.

True, but AFAIK DOOM operates in the "standard" VGA address range, so the BIOS should already have taken care of that.

I tried with and without setting the MTRRs for the K6-III, made no difference.

Folks,

Attached is a ZIP containing an HTM that you might find interesting. The HTM contains the results of benchmarking using DOOM II on a lot of the popular CPUs and video cards back in the day.

P.S. BTW, the guy who compiled these results is who I mentioned in my readme in that old QBasic utility that I uploaded previously. After uploading the utility, I remembered that he had compiled a file with various results, and so I dug it up from my file archives (and figured I'd upload it here).

P.P.S Just a few interesting benchmarks using DOOM II and the QBasic utility (all ran on the exact same PC):

1Clean bare-bones (no CONFIG.SYS or AUTOEXEC.BAT) Windows 98 SE DOS bootup
2Gametics= 2001 Realtics= 562 
3124.6174 Frames Per Second
4
5Windows XP SP2 COMMAND.COM
6Gametics= 2001 Realtics= 672 
7104.2188 Frames Per Second
8
9Windows XP SP2 CMD.EXE
10Gametics= 2001 Realtics= 665 
11105.3158 Frames Per Second
12
13DOSBox v0.73
14Gametics= 2001 Realtics= 1290 
1554.2907 Frames Per Second
16
17DOSBox v0.74
18Gametics= 2001 Realtics= 1265 
1955.36364 Frames Per Second

(edit) Forgot to mention, specs of the above PC:

Dell Dimension 4600
Intel Pentium 4 CPU 2.66GHz
512MB RAM
Onboard Intel 82865G Graphics Controller (96MB shared system RAM)
Windows XP Professional SP2 (I used a bootable flash drive to boot to W98SE DOS, mentioned above)

I'm not sure what's the problem here. I'll post the bios pages soon.

http://www.complang.tuwien.ac.at/misc/doombench.html
btw, one of my results here is 487 (153.3 FPS).

Here are some pics of the test system :

Using 3dBench2, I got 197.4 fps. Just with the same configuration I used to test DOOM.

EDIT: That was for the S3 Virge VX.

For the Geforce4 Ti4800, I got 214.7fps.

I remember 3DBench being inaccurate on fast 486s and above.

I'm not sure if the 3DBench2 is showing true results. I wonder what is causing the limitation in the fps count in DOOM. And I can't judge by looking at the screen alone because all the cards I've used show similarly smooth movements.

I forgot to mention this system has 2 Voodoo2s running in SLI. The PCI cards always pass through the voodoo2.

In addition (forgotten to mention this too), I have the Gravis Eliminator Game Card in one of the ISA slots. Maybe I should remove this and test again...

New results posted in the first post above.

No other settings were altered.

No changes noted after loading uvconfig from univbe67 (the version that supports TNT2).

This info is rather valuable for me. I really appreciate it. It helped me a lot:)