By: GTX_SlotCar

    Most video card reviews concentrate on how fast a card performs in Quake 3. That's fine if First Person Shooters are your main pastime, but it leaves out those of us who are hard core simulation racers or arcade drivers. I recently read a review where the GeForce2 MX showed almost the same frame rates in Quake 3 as the Voodoo 5 5500 in some resolutions and color depths. In other resolutions, the V5500 was much faster. I'm sure that this is not a surprise to most of you who know that video cards can perform quite differently under varying stresses. Also, I have for some time believed that nVidia tunes their drivers for optimum performance in Q3.

    What makes this a fair comparison now is the extremely reduced  pricing on the V5500 card. If you're looking at buying a new video card now for under $200, there are a few choices to consider. The V5500, while it's still available, one of the GeForce2 MX entries, and the Radeon32 DDR are probably the top choices. If your budget isn't limited, then the GeForce2 Ultra for about $400 is probably the performance king, and in the $250 to $300 dollar range the Radeon64 DDR and one of the GeForce2 GTS (or Pro) cards is probably right for you.

Here are the specs on my system:

Abit BE6 II mother board Celeron 566E overclocked to 935 Mhz (8.5 X 110) 256 MB of PC133 cas 2 memory (Micron) Sound Blaster Live XGamer Creative Labs CD-RW Creative Labs PC-DVD Encore Dxr3 WD 27GB ATA66 hard drive Generic 3 1/3" floppy drive SMC network card ViewSonic 21" monitor Windows 98 LiteStep replacing the Explorer shell

    For the tests, I did a fresh installation of Win98 and installed only the video card I was testing. No other cards (sound, network, etc.) were installed. I also turned the overclocking down to 914 Mhz to ensure that no graphical artifacts could be attributed to oc'ing. Although both cards were quite capable of being overclocked, there wasn't much performance gain so, believe it or not, I tested them at their stock clock rates. This is probably the only time you'll see me do a test this way.

The GeForce card:

The card tested is a 32 MB reference design with TV out. It has several nice features. The dual head "Twin View" design lets you connect your regular monitor along with a flat panel screen or TV / TV-S. You can have your second viewing device set to clone your monitor's image, or set to extend your desktop. The nice thing about this design is that you can view both screens at the same time. Many video cards with TV-out will blank out your monitor if you turn on the TV-out. The overclocking controls were easy to use, there were several screen adjustment options and I particularly like the Digital Vibrance Control that is used to enhance the colors. The board and design are high quality. The mounting plate has a rolled edge that adds to it's structural integrity and makes the plate almost unbendable. The supplied drivers installed easily, though I quickly upgraded to the latest 631 version.

     So, now that 3dfx is out of business, is the Voodoo 5 5500 something to avoid? Let's find out.  The first thing I did with each card was to get a benchmark from Wintune98 online. All the tests that follow were done with the .631 GeForce drivers and the .666 3dfx drivers. 

GeForce2 MX 32 MB	3dfx Voodoo5 5500

    As you can see, neither card is very fast in OpenGL. On one computer at work I have an old TNT 16MB card that scored 107.1 MPixels/s on a 464 Mhz  computer. Otherwise, the cards were quite similar with a slight edge going to the 5500. Now let's see how meaningful these scores are in the 'real' world of racing simulations. All game resolutions were set at 1280 x 1024 except the Nascar4 demo which would only allow 1280 x 960, and color depth was set at 16 bits. All the card setups were left the way the setup program installed them with the exception of AA on the GeForce2 MX. It was set to about 30% and I turned it down to zero. Running any FSAA on the MX killed the frame rates.  Many settings for the V5500 were on 'automatic' and turning them down to 'normal' would have increased frame rates. Still, I left them in their default state.  The games tested were Grand Prix Legends (GPL), Colin McRae Rally2 demo (CMR2), Midtown Madness 2 (MM2), Nascar4 (N4), Nascar Heat (NH), and Need for Speed: Porsche Unleashed (NFS:PU). 

Let the games begin !

    First up is GPL. There weren't really any surprises here as both cards pegged the frame rate most of the time in D3D with every graphic enhancement turn up full. Running FRAPS to get the frame rates didn't do anything to slow the game  down when compared to GPL's onboard frame counter. In other games I get about a 5 fps loss running FRAPS. I also ran the V5500 in Glide with the same results. It's interesting to note that the graphics quality is still quite a bit better in Glide than in D3D. This was mostly evident with the gauges on the dash board. They were sharp and clear in Glide and a bit blurred in D3D.  Running hi-res add-on dashes with the MX helped, but they still weren't as good as the Glide version.  As a bit of a side note, I generally run the V5500 at 1600x1200 in GPL. It pegs the frame  counter with the exception of starting a race in 20th position where the rate drops to 22fps for about 2 seconds and then quickly climbs back to the top. The MX didn't fair as good as the V5500 at this resolution, but I don't really think that's an issue. Overall, both cards performed very well and the D3D patch for GPL is very well done.

    Next up, Nascar Heat. The MX2 did very good here, too. NH gives poly and fill rates that I don't consider important in the actual frame rates, and they seem to change each time you start the game. However, I'm sure that some of you will be curious about them so they're included. The fps scores come directly from the game's internal benchmark program. The frame rates didn't change much from one test to another, but to get the poly and fill rates, I took a high average. The tests were done with 'high detail cars' set to on, and all graphic options full except the "Effects Level" which was set to off. In other words, no skid marks or smoke.

(The Radeon32 DDR had a fill rate of over 30,000 and a poly rate in the 3000's, but the frame rate was 42.)

The graphics detail in both cards was excellent. Sponsor decals on the cars were crisp. This is important to note because NH lowers graphic details automatically when needed to keep up the frame rate. There will be more about this at the end of this article in the Bonus section. All in all, I'd consider either card very good for NH. Of course, with the V5500, you can get nice graphics at 800x600 resolution with 4xFSAA and the frame rate is the same as 1280x1024 without FSAA. As in all the games I tested, the MX slowed down quite a bit with FSAA so I wouldn't consider it an option.

    On to Nascar 4 (demo). And since it is a demo (the sim hasn't been released yet), things could change by the time the full version comes out. Again, I used the game's internal frame counter. These measurements are a bit subjective, but that's what this site is all about. I put the lowest and highest frame rates observed, and then used the frame rate I saw most often as I looked at the counter while driving the track at Michigan.

I must admit that I was a bit disappointed in the MX's performance here. The one shining observation is that there wasn't much difference between the Most Observed and High fps counts. Again, this is a demo and the graphics speed may increase for the MX when the game is released. The frame rates also increase when playing the game at 1024x768 resolution, and that still gives a nice experience. Setting the display driver to T&L showed no increase in speed for the MX, but once again, this may change with the final release.

    Ok, a little fun with Midtown Madness 2. Here I drove the Audi-TT in Big Kitty paint on the 'Parks But No Parking' course with a full field of cars. I'll keep this one simple and just give an average frame rate. I don' know what it is with the Midtown Madness series and nVidia drivers, but I'm sure the MX could give better frame rates with older video drivers. I used FRAPS to get the frame rates. As I've said previously, this knocks about 5 fps off the actual count as compared by running FRAPS in games that also have an internal counter. The MX2 averaged 24 fps and the V5500 averaged 36 fps. While running the game with the MX and FRAPS, the motion wasn't smooth. When I disabled FRAPS, the motion was fluent enough and the game played well.

    Now on to Colin McRae Rally 2 (demo). The track is Stage 5 at Australia. All graphics enhancements on full. Because there were places in the course where frame rates really dropped with both video cards, I'll give two low frame rates, a high (generally on the tarmac), and an average or Most Observed.

Again, take into consideration that this is a demo and the frame rates were measured with FRAPS. CMR2 demo only let's you set the maximum game resolution one setting below your screen resolution, which seems a bit backwards to me. I'm not sure if this has changed in the full release. This means that I had to change my desktop setting to 1280x1024 32bit color to get 1280x1024 16bit color in the game. If you check the Utilities page on this site, you'll see a program called GameBar that will do this automatically when you launch a game and then return your default settings when you stop playing. Another side note, CMR2 does a very nice job modeling Front Wheel Drive physics. I really enjoyed (and excelled) at the snow covered course. I'm sitting here in Maine, in the country,  with 15 inches of newly fallen snow on the ground. Hmmm, I'll be right back.........

    Brrrrr. Well, that was fun. And I must say that Officer Nash was a good sport and very understanding. 

    Okay, the last one. Need for Speed: Porsche Unleashed. If you're new to sim racing and you don't have this game yet, get it. Not just for the fine graphics, but for the driving school. You could learn a lot.

    Again, I had all graphics options turned up to maximum and the resolution set to 1280x1024. I drove the stock 914 4/2.0 with 7 opponents at Normandie.  Frame rates are given in Low, High and Most Observed and were counted using FRAPS.

I think if I was going to race NFS:PU with the MX2 card, I'd go with a 1024x768 resolution. However, the game play was quite a bit smoother when I wasn't running FRAPS.

    Well, there you have it. The GeForce2 MX card I used is a very nice design and I like a lot of it's features. For the price, I'd say it performed quite well. In the tested resolution, I may have preferred it's graphics interpretation on the car lighting textures slightly more in NH than the 5500's, but only there and both cards were excellent.  As is typical with nVidia drivers, I probably could have had higher frame rates in some games by changing to older drivers, but I was pleasantly surprised to see that all the games were stable with the 631 MX drivers. Still, if you're looking for an inexpensive nVidia card and don't care about Twin View and some of the other features, I just saw an 'old' GeForce 256 DDR on pricewatch.com for $100, and it's quite a bit faster than the GF2MX. As for the Voodoo5 5500, consider that many of your older games really come to life with FSAA (Full Screen Anti Aliasing).  Sports Car GT is one example, but perhaps one of the most dramatic is Monster Truck Madness 2. This game has nice physics (a bit slippery) and great graphics as well as 1000's of tracks and vehicles and tremendous on-line play. The problem is that it only plays in 640x480 resolution. But with 4xFSAA it's beautiful and you still get up to 200 fps!

Bonus!

    I said I had a bonus for you. This may be pertinent if you're considering upgrading from your old Voodoo3. Out of curiosity, I dusted off my Voodoo3 3000 and ran it through a few of the games. I used the already installed 5500 drivers, version 666. Here, briefly, are the results:

 NH - Wouldn't let me choose 1280x1024, so I had to use 1024x768. Otherwise the settings were the same. 39.6 fps, 171 Fill and 4698 Poly counts. But the game adjusted the graphics quality to get  those frame rates and the sponsor decals were just plain ugly. It would be better to use 800x600 and/or turn off 'high detail cars'.

N4 - Surprise. Low 22, High 31, Most Observed 27. Graphics quality VG.

MM2 - Average 28 fps. Graphics quality VG.

NFS:PU - Another surprise. Low 29, High 48, Most Observed 40. Graphics quality VG.

    That's about it. If you're looking for real big gains in frame rates, I think the only card that will deliver is the nVidia GeForce2 Ultra, but it'll cost you plenty.  I tried GPL with the 5500 in 1600x1200 (with forced 32bit color, for what it's worth) and 4xFSAA. Of  course, I only got about 10 fps, but the picture was amazing. Absolutely no jaggies at all. It's a real shame that we will never see the Voodoo5 6000. Then again, maybe nVidia will do something with the technology. There must be some reason why they bought it from 3dfx.

GTX_SlotCar  (Gary DeRoy)

January 3, 2001