I was able to get an Nvidia Quadro DCC (basically a Geforce 3) running smoothly on my FIC VA-503+ (MVP3). However, there's no reason to do this. My Geforce 2 MX performed just as well in my testing, and I suppose that if I had an original Geforce, the result would be close to the same. I might test with a TNT 2 Ultra. For now, the GF2 MX works fine, and it is quiet because there is no fan or heatsink.

I keep the GF2 MX in the AGP slot and a Voodoo 3 3000 in the adjacent PCI slot. Good combo for this setup.

boxpressed wrote:

I was able to get an Nvidia Quadro DCC (basically a Geforce 3) running smoothly on my FIC VA-503+ (MVP3). However, there's no reason to do this. My Geforce 2 MX performed just as well in my testing, and I suppose that if I had an original Geforce, the result would be close to the same. I might test with a TNT 2 Ultra. For now, the GF2 MX works fine, and it is quiet because there is no fan or heatsink.

I keep the GF2 MX in the AGP slot and a Voodoo 3 3000 in the adjacent PCI slot. Good combo for this setup.

Isn't that too much on the power requirements of both cards? Considering a K6-3 for example the wattage on the mainboard seems to be quiet high isn't it?
On the 2D quality I find the Savage4 image quiet good compared to the V3 in desktop, but it's probably difficult to say. And the G2 MX I remember that shouldn't be better either but is certainly a much more advanced and complete solution.

From my own testing with k6-2 and k6-3+ on an asus p5a (ali V chipset) i found that g400, banshee and tnt2 ultra performed best on this platform (in that order). I tested all nvidia's from tnt1 to ti4200 including every mx version and all voodoo's from voodoo Rush to v5 5500. v4 and v5 also performed well.

Tested with 3dmark99 and a in-game benchmark from n.i.c.e 2 (racing game).

On faster systems the other cards ofcourse pulled away easy including the voodoo3's but on a k6 above mentioned cards performed best. Banshee for example was 15-20% above every v3 (pci, agp 2000,3000 and 3500). I don't know why, but it was.

Don't know if the same holds true on a mvp3 board but i do think the g400 will be fastest on that as well. When testing the same 20+ cards on a p2-450 the v5 5500 was fastest together with the g400. Only on a faster p3 system the g400 couldn't keep up anymore and the faster voodoo's and nvidia's started to shine.

G400 is a great card.

The Voodoos are the way to go because you'll want to use Glide if it's available. Less CPU overhead among other things for those old games. I wonder how many times this has been asked now.... What do you think F2bnp? 😉

I'll have to test again with my G400 Max. Last time I tried to use it, I got a BSOD. However, I had not installed the VIA AGP driver. Interesting about the Banshee too. I have one of those to test too.

G400 is likely to be trouble on VIA MVP3.

swaaye wrote:

G400 is likely to be trouble on VIA MVP3.

That's what I figured, something with the chipset. Is there a specific reason for this incompatibility?

The last time I tried the G400 it gave me beep-code errors and no boot. The first Voodoo with my config has problems too. The Voodoo2 run correctly but no SLI is possible for lenght cards problems.
Actually testing a Mystique 1164SG PCI 4MB. Great 2D, but in 3D could it run anything at all? 3Dmark99 info gave me only few 3d features capaibilities and actually blank shaded polygons at benchmark (but smooth framerate anyway 😁).

swaaye wrote:

The Voodoos are the way to go because you'll want to use Glide if it's available. Less CPU overhead among other things for those old games. I wonder how many times this has been asked now.... What do you think F2bnp? 😉

Ah yes, my favorite thread! Let me tell you about my K6-III+ 550... 🤣

Jokes aside, repetition isn't a bad thing as new and different people arrive here everyday and may have different opinions on matters. Having our views challenged by such people is beneficial for the community in the long run! That said, Voodoo 3 is the way to go, Voodoo 5 if you'd like some free 2x FSAA, but somewhat worse compatibility. Voodoo 2 SLI is also nice, but I've spoiled myself so bad with the Voodoo 3's fantastic picture quality that I really can't look back. 2x FSAA is free on the Voodoo 5 since the CPU is so slow that the card can spend cycles doing more stuff and still won't reach 100%.

However, and this ties in with the "new people challenging our views" bit at the start, it has recently come to my attention that ALi Aladdin V boards seem to develop something of a temperament. It seems that for whatever reason the system performs not nearly as fast as it should, independent of API and even resolution, which would betray that this is a CPU bottleneck of sorts. Well here is where it gets weird. I was testing Forsaken on my K6-III 550 + Voodoo5 last summer and I was scoring around 160FPS at 640x480 (tip: Tualatin 1.4 score about 400-450FPS with the same card 🤣 ). After trying out a few other games as well, I reran Forsaken and noticed that the framerate wouldn't go over 80FPS and it wasn't a Vsync issue either. Fellow member Skyscraper also had the same issue and despite multiple formats and re-installations to figure out what is wrong, I haven't come to a conclusion. By the way, this affects all 3Dfx cards I've tested (V2 SLI, V3 AGP/PCI, Voodoo5). I should really make a new thread about this, but I've spent a lot of time trying to figure out just what the hell is going on that I'm really tired. I don't have any MVP3 boards to see how these behave.

If anyone has any clue on what's wrong here, feel free to chime in!

Interesting details about the weird slowdown. I have a P5A but don't use it much. Most recently I was experimenting with K6-3+ DVD playback and was pulling hair over getting a Rage 128 to work without BSODs.

I think we should start a new thread about this and the few (or many? 😜) of us that own K6-2+/K6-III+ and V3/V5 cards post results with specific games. That way we can figure out any discrepancies. Maybe you remember, but I used to have the same problem with a Gigabyte GA-5AX years ago and it literally disappeared once I swapped my PCI Voodoo3 for an AGP one. Turns out I wasn't that lucky this time!

In my "4 in 1 retro gaming PC" project, I used the GA-5AX and AGP V3 and also noticed that it was slower than using another motherboard. I thought that it's to do with that specific motherboard, but maybe there is more to it.

I've got some data from my Voodoo 2 SLI scaling project. I used a DFI K6XV3+/66 with VIA chipset: http://www.philscomputerlab.com/uploads/3/7/2 … ing_project.pdf

So you reckon VIA chipset is the way to go?

Attached the Socket 7 and Super Socket 7 results:

I don't know 🙁. I have a few benchmarks on record that seem to be fine for Voodoo3, perhaps you could set up an MVP3 system with a Voodoo3 and K6-III+ 550 and post some results?

Unreal, Quake 2, Incoming and Forsaken are pretty good picks for benchmarking. If I had to add a few more, I'd go with Quake 3 (patch 1.16n), Descent 3, Shogo, MDK2, maybe even Dethkarz and Turok 2? I believe all of these offer internal benchmarks. I'll open up a new thread about this with more details!

Yes, Descent 3, MDK2 and Turok 2 are easy to benchmark. Descent 3 needs a bit of work with the command line options (you enter them inside the SETUP window).

Might be a project worth checking out.

I tested both platforms with a lot of cards (3dfx range and nvidia range and g400) and i found the ali V a tiny bit faster overall compared to the mvp3. However, I did the tests on a k6-2 450 and not a ''+'' cpu.
I only tested 3dmark99 and a game with build-in bench called n.i.c.e 2.

But how does this work exactly? Forsaken only runs fast ONCE and the second run and later ones are much slower? or does it become slow over a longer period of using the game etc. ?
Do you think it is the ali V or the ''plus'' cpu or just this combination?

I can compare p5a with ax59pro. Mvp3 system has normal k6-3 450 and p5a has k6-3+. I can clock them both at 450mhz and run them with the same cards. I also have the dfi mvp3 board if that makes things easier.

Are you serious that a Banshee is faster than a V3 3000 AGP in a k6-3 450 ? I was about to get one to replace the Banshee (monster fusion)

It just happens on every run. I haven't managed to get the system working properly for months now (July or so), it's just slower in games. I can't get past 30fps in MDK2 for example, no matter what. I reinstalled Windows 98SE this afternoon, installed everything I'll need and will create a new thread about this whole deal, probably tomorrow or the day after that. I'll compare with my older benchmarks and then we can all run specific games and 3DMark 99/2000 to compare and hopefully get to the bottom of this issue.

How many versions of the ALI AGP GART driver have you tried? I noticed that there are a ton of revisions, like with VIA and SIS hardware. Like everyone except Intel. 🤣 Probably a sign of endless attempts to work around errata.

There's also an official ALI AGP utility that lets you configure a bunch of AGP parameters in a GUI instead of messing with registry tweaking.

For Quake 2, the fastest result I got from my MVP3-based FIC VA-503+ with my K6-3+ 550 and a Voodoo 3 3500 (183 MHz) is 86.9 FPS at 1024x768. This is using the 3DNow! enhanced version of Q2.

For Unreal Gold, I got 49.4 FPS at 1024x768 with the same setup.