Lets face it, you either got'em or you don't.
Voodoo5 Vs. GeForce2
Let's see who's got'em and who don't.
First let me make sure you understand the context of our little article here. This is NOT a review of either the ELSA GeForce2 or the Voodoo5 5500. It is rather a look to see which chipset has got the strength to win your hard earned green. I am talking frame rates and I am talking Quake 3 Arena. Yep, what has basically become "the defacto standard" in benchmarking 3D Vid Cards.
We are not covering FSAA, Pixel Shading, GPUs, TnLs, T-Buffers or any of that other marketing BS. While FSAA is a VERY real part of the equation with both cards, we will leave that till a later date. Consider this to be part one in the ongoing battle between 3dfx and nVidia. If we are lucky, we will have ATI join the Battle Royal later this summer, but I am not holding my breath.
What are the facts? 3dfx is about a late SOB. This product should have been out months ago. I know it, you know it, 3dfx knows it. It has been a concern of mine that 3dfx was moving themselves to the back burner in the 3D market. The thought of this saddens me, but if you can't run with the big dogs, stay the hell on the porch.
nVidia has been churning out technology like it is going out of style. While I used the term "marketing BS" above, they have brought some very real features as well as blinding speed to 3D graphics in the last year. nVidia has always been synonymous with image quality but it seems as if they have become more obsessed with speed than their competitors. Not only in Frames Per Second (FPS), but also in making sure new and quality products are in the market place. nVidia deserves big credit for driving the market place forward and not allowing it to stagnate.
With that said, many people think that the GeForce2 GTS is simply going to blow 3dfx out of the water with the latest line up. Is nVidia a whole "cycle" ahead of 3dfx? Let's see if it is true.
We are lucky enough to have in our possession a "production" 3dfx Voodoo 5 5500 AGP. Why did he put quotes around the word "production"? This card is NOT a production card but rather an Alpha 2 state board with a couple of reworks done to it. 3dfx has assured me that this board functionally represents what will be on the store shelves soon. We stopped by 3dfx sales offices here in Dallas, Texas yesterday and picked up the GOLD drivers to be shipped with the board. So what we have here is basically a retail package. I think this is solid enough to put into a comparison and have it represent what will be purchased by the end user.
On the nVidia front, we are lucky enough to have purchased an ELSA GLADIAC GeForce2 GTS. So we KNOW the ELSA is a retail product. We had problems with the ELSA drivers that came with the card and have reverted to using the 5.16 Beta drivers that have been floating around the Net. These are the drivers we understand that the Hercules GeForce2 card will ship with so they seem acceptable to use in our comparison.
As we said this is not a review of either card but a grass roots comparison. Of course from the pics (all of which are clickable) you will notice the size difference. The Voodoo5 is HUGE. More on issues associated with this later in the actual review.
Here are a couple of screen shots of some of the card configurations reached with the drivers. We used H-Oda's WCPUID program to get this information.
[IMGhttp://www.hardocp.com/reviews/vidcards/shootout/v5-gf2/id-gf2.gif[/IMG]
Notice here that 2X AGP is all that is allowed in our 5.16 drivers along with Side Band Addressing being disabled. Of course the BX chipset ONLY allows the 2X AGP physically, but the drivers limited the card on a 4X AGP also.
[IMGhttp://www.hardocp.com/reviews/vidcards/shootout/v5-gf2/cpuid-v5.gif[/IMG]
Same thing here except that Side Band Addressing is used with the Voodoo 5 in this driver release.
UPDATE! We have a correction to make in the screenshots that are posted below. We have incorrectly shown Vsync to have been enabled when in fact it was not (my bad). Chris Gordon at PlanetGeForce questioned us on our testing. If in fact Vsync would have been used, the tests would not be worth the bandwidth the data was transferred on. We realize that Vsync MUST be disabled to comparison test fairly. To MAKE SURE, I just got finished running the benchmarks AGAIN. We have verified theses benchmarks to be correct. Our second set of numbers, did not match exactly, but they WERE NOT MORE THAN 1 to 3 FPS difference from what is posted. Which I think that is to be expected considering the test bench has been used and configged for other tests since then. I am still very comfortable with our results and stand by them.
TEST 1
All of these bench marks are run on Quake 3 Arena, a very OpenGL sort of game. No advantages of Glide or other nonsense to tip the scales one way or another. This test 1 is based on 640 x 480 Rez with 16 bit color. I have included screen shots of each setup we have done here so you can duplicate it EXACTLY at your own casa. All tests were run three times back to back and then the HIGHEST FPS number is used.
We see the GeForce 2 SMOKE the Voodoo 5 by a total of 27 Frames per Second. Not a small number by any stretch of the imagination. I have been playing deathmatch with both of these cards now for a few days. I have used them at low (640x480) resolutions with a lot of the eye candy turned off in Quake 3. The Voodoo 5 will deliver solid framerates between 120 to 170 frames per second during gameplay. This no doubt gives you a hefty advantage over your competitors. The GeForce2 will punch out frames in the 140 to 200 FPS range. It is almost scary it moves SO fast and SO smooth.
TEST 2
This time round we went for more of what these cards have really been designed for in my opinion. Delivery of framerate while also serving up the eye candy. Rez is 1024x768 with 32 bit color turned up. Here again are the exact settings
Check the second pic for sure, we have turned on all the goodies, sky, blood, shell casings, etc....
After running the low resolution benchmarks these quite surprised me. While the graph shows a big gap, the actual frame difference in ONLY 3 frames. Hell, you guys know as well as I do that 3 frames at this point in the game means really nothing. For bragging rights nVidia kicks ass in our round two of testing but not by the landslide we saw at low resolution.
Test 3
While we were meeting with nVidia at WinHEC they showed us some 1600x1200 resolution numbers on a little chart much like ours. It showed the GeForce2 running at 59 FPS. We asked what the exact settings were that they got those results and we were assured it was with "all the good stuff and 32 bit color". Maybe I paraphrase a bit there, but that was the gist of it. Well here are OUR 1600x1200 resolution scores with all the good stuff and 32 bit color. Funny that ours does not look to be in the 60FPS range.
Again we have our beauty of a graph showing a pretty big difference. One day I will learn to make proper graphs and end this nonsense.
ONLY a 0.7 frame per second difference between the Voodoo5 and GeForce2. And while the Voodoo5 did NO BETTER than the nVidia neither came close to the 60FPS mark.
Conclusions & Delusions:
As we always do, we are gonna shoot straight with you. We are 3dfx fans, as well as nVidia fans, and I saw 3dfx becoming the serious underdog in the 3D wars. I already had it in my mind that the GeForce2 would summarily hand 3dfx their own ass till the next round of silicon hit the shelves in 6 to 12 months from now.
3dfx has blown me for a loop in that they have a card in the Voodoo 5 that will keep up with the GeForce2. My hat is off to BOTH companies here in the fact that both of them have brought some really nice products with muscle to market.
I think where we saw the 3dfx Voodoo5 get skunked in the lower resolution benchmarks is a driver issue and not a horsepower issue. I would guess that we would see some more driver sets out of 3dfx before the summer is up that bring the performance level up 15 to 25%. I have even heard of third parties developing drivers for the Voodoo family of cards. So there is no telling what we might see.
When we did the Voodoo 5 PREview here at the Hard|OCP we had the damn community about take our heads off over not showing resolutions above 1024x768. I assume from that reaction that MANY folks are utilizing the higher resolutions when gaming. If framerate at high resolutions are what you want it now looks as if you have a couple of cards to choose from.
I think we can venture to say that there was virtually no performance differences in the cards at medium to high resolutions with 32 bit color. One thing that is clear is that the Voodoo cards have lost that "Voodoo look". If you have ever owned a Voodoo card you will know what I am talking about here. 3dfx has certainly stepped up the image quality with this card. It is on the level with nVidia for sure.
We will be looking into the tons of features that both cards are packing. I do have to say that I think 3dfx is going to walk away with the quality prize when you bring FSAA into the mix. Also it is something we can USE IN GAMES TODAY.
We will surely see more offerings of cards with more memory from nVidia and their OEMs, but now I am wondering where the Voodoo 5 6000 is going to fall. I had originally thought that the 6000 would be needed to keep up with the GeForce2, now it seems it might just blow it out of the water. That will be fun to see for sure.
Now the argument of which is better is going to be a LONG RUNNING series for sure, but one thing we do know now is that the GeForce2 is the fastest, but not by much!
Now if we can just see the Voodoo 5s on the shelf and MORE of the GeForce2s also.
.................................................
PowerD By ChukE (
http://www.cultdeadcow.com )
:strip_icc():strip_exif()/u/7593/bennie.jpg?f=community)
:strip_icc():strip_exif()/u/8818/digidna.jpg?f=community)
:strip_icc():strip_exif()/u/9104/kennysig.jpg?f=community)
:strip_icc():strip_exif()/u/11115/crop5ddd078a6380b_cropped.jpeg?f=community)
