I had done this a while ago but now that i have my bench system setup i am going to start doing it again with the gammet of Video cards i've collected.
I've got most of the 3dfx cards minus some of the rare cards, almost all of the Nvidia cards upto the 6800, and quite a few of the ATI cards (Rage Fury Maxx!), i even have a Kyro2 i'll be adding the numbers for.  I'll update this as i Go.. if anyone has Q2 and wants to add some numbers.

The settings i am doing are

default settings
800x600
s_initsound 0
snd_restart
timedemo1
demo1.dm2

AMD3500+ Software	84.8
AMDX2 4400 Software	112
Voodoo2 8Mb @90Mhz	59
Voodoo2 8Mb @99Mhz	65
Voodoo2 8Mb SLI@90Mhz	103
Voodoo2 8Mb SLI @99Mhz	114
Voodoo2 12Mb @90Mhz	59
Voodoo2 12Mb @99Mhz	66
Voodoo2 12Mb SLI @90Mhz	100
Voodoo2 12Mb SLI @99Mhz	114
Geforce 256	178.1
Geforce Ti4200	461.2

Intersting Note:   the AMD3500+ is faster than a single Voodoo2 but the Voodoo2 SLI beats it..  Kindof goes to show how important the hardware rendering (not that we need convincing) but it wasn't until more modern cpus that they can finally process the 3d just as fast as some of this really old hardware

I did post some benchmarks for the 4500 agp and 5500 agp and pci . Topic was Weird benchmarks.

5500 getting better scores than the GeForce 256 :)

Did you get any rresults for your rage Fury maxx yet ?
Also interseted in Kyro results, only have one card myself but dont want to breeak the seal.

oldskool wrote on 26.01.12 at 22:12:44:

5500 getting better scores than the GeForce 256 :)

That's no surprise, V5 5500 was on par with GF2 MX400 (in lower resolution GF2 wins, in higher resolution V5 5500 wins). And GF256 is slower than GF2 MX400, so it's obvious that V5 5500 is better than GF256.

Some here too-

http://www.thedodgegarage.com/3dfx/bench.htm

http://www.thedodgegarage.com/3dfx/bench_v6k.htm

I've got a bunch of Quake 2 640x480 / Software rendering (demo1) benchmarks from a different perspective (CPU-wise) at thandor.net - benchmarks - Quake 2 640x480.

Interesting yet confusing :)

Why :)?

To much Data  from Gdonovan  and the rage fury Maxx results :)

i took a bit of a break and getting back into it..

i was digging up some of my old data and this was some fun comparison i had done on SLI vs AFR scaling comparing the V2, V5 and MAXX


the Voodoo2 @ 90mhz 
Single: 59 SLI: 103 / 75% Increase


the Voodoo5 5500
Single: 125.7 SLI: 220 / 75% Increase

ATI Rage Fury Maxx
Single: 90.1 AFR: 129.7 /43% increase

my 8800GT's in SLI back in the day also were seeing about a 43% increase using AFR.

What i found really interesting is at the time ATI had a patent pending on the AFR technology.. Makes me wonder if they ever got that Patent and if Nvidia is having to license that from them.

Very interesting.

Which setup are you getting these figures from?

i also done some benchmarking with several voodoos
but i prefer a other demo file how uses more Textures and more models
with this demo a Miro/ Canopus is much faster than a 4MB V1

http://www.abload.de/thumb/q2wibx0.png

Will have to test the crusher demo out next time I have my retro rig apart :)