Well, since the V4 outperforms a GTS MX in UT,
Ah, the classic zombie argument. Start "benchmarking" a game based around Glide and proudly show off the benchmarks.
It does have trilinear mip-mapping but only when single-texturing. EG almost never, and it does not have anistropic.
Right. And neither does the V5 5500.
You are comparing a previous generation performance card to a current generation entry-level card.
Previous
two generations performance card. Anyway nVidia's MX = GF1 SDR while your "entry level" V4 = TNT2 Ultra and probably costs more than the MX.
So nVidia's entry level card has more features and equals the performance of the previous high end board, while your board has less features and equals the
previous previous speed king.
And since the performance of the cards is so similar, what you are actually saying here is that the MX=TNT2 Ultra. Not only is the comparison invalid, it is also damaging to your argument for the MX.
I love it how you PR types twist words. I never said the MX = TNT2 Ultra. I said the V4 = TNT Ultra and the GF2 MX > either of those two. The only thing I am damaging is your "entry level" theory because nVidia's definition of entry level is higher than 3dfx's.
You bet, especially when compared with a Voodoo4.
Don't make me laugh. I have seen benchmarks of the V4's lackluster FSAA scores. You can barely play modern games at all with it nevermind with using FSAA. (See later for the benchmarks).
These also are non-issues. AGP texturing and 4X (or any X for that matter) yields virtually no performance advantage
I have seen benchmarks which disagree with this.
T&L has to be present in the application to be taken advantage of (unlike FSAA)
So does 32 bit colour. Why did you bother including it then? What about OpenGL? Games have to take advantage of it too. Why did you even bother with it? Oh wait: the V3 didn't have
any of these things. Now I understand 3dfx's philosophy.
A quote from Carmack about Doom 3: 3DFX Voodoo4/5, S3 Savage4/2000, Matrox G400/450, ATI Rage128, Nvidia TNT[2]: Much of the visual lushness will be missing due to the lack of bump mapping, but the game won't have any gaping holes. Most of these except the V5 probably won't have enough fill-rate to be very enjoyable.
In other words you will need the V5 5500 costing twice as much to play the game and it will still be inferior to an MX. The V4 will be already be useless and it has only just been released. What does that tell you?
Not only that, but the MX with all its "not needed" features is able to perform better and faster than the V5 5500, which costs twice as much and has twice the RAM and CPUs. Boy you're right, nVidia sure did include a lot of "useless" features on their boards.
Read the whole thing here.
What about the Voodoo4 has LOD adjustments which can be used to enhance image quality and the
Of course sharpening the image in this way results in perfomance hits.
MX only has better performance in some areas while the V4 has it in others.
V4 = Unreal/UT and MX = everything else. In fact now UT is better on the MX because of S3TX not available on the Voodoo. So I guess that just leaves Unreal.
Yes, and any comparison you look up (or any benching you do yourself) will tell you the well established fact that the 3dfx implementation is faster than nVidia's. Most will also tell you it looks better.
I have seen FSAA benchmarks showing nVidia's boards outperforming the Voodoo simply because it doesn't matter what you do differently, the fact of the matter is you are still rendering more pixels like the others. Since FSAA is almost entirely dependent on raw fillrate, I think we know who comes out on top with these benchmarks.
The image quality is also very subjective and a lot of sites think that high resolutions are far better than FSAA.
A quote from planet hardware: And while the anti-aliasing routines of the V5-5500 are absolutely the best in the group of cards tested here, most of our staff feels that they'd rather not give up playing games at a resolution of 1024x768x32bpp with all graphic features on besides AA, than have to play at a modest 640x480x16bpp with AA enabled.
Oh, and lets not forget the V5 finished
last out of all of the tested boards.
Planet hardware round-up.
That is a fascinating conclusion since it is documented fact that the 3dfx implementation is faster.
Perhaps it's documented but where are the benchmarks to prove this "documentation"? I have some
here.
The V4 in HQ Q3 is scoring 36 fps with FSAA * 2. How on earth can someone recommend a V4 over a GF2 MX based on FSAA? Wait don't tell me: the human eye can't see a difference over 30 fps right? Or is it 24 fps, just like a film?
Yes, which at the ultimate high setting will get your image looking as good as 3dfx's in 2X while dropping the performance way below what the Voodoo4 would be doing.
I don't think so.
Here we have some FSAA x 2 screenshots on both boards. The difference between the two was so low that the V4 comes slightly better after downloading both still images and looking at them for a while. I don't know about you but I don't know anyone who plays Quake 3 to look at still images. In contrast raising the resolution results in an
immediate and obvious image quality gain.
They really don't stand up to even a small measure of scrutiny and as you can see from the responses in this thread, they make people question your motivation.
And what makes you think they don't question your motivation? I don't work for any graphics card company nor do I have shares in one. I am completely independant consumer and I simply pick what option is the best.
You on the other hand are a 3dfx PR guy. Are you suggesting you have no bias toward 3dfx? If you don't, I'm not sure why they gave you such a position.
Your statement is the most extreme case of "pot-kettle-black" I have ever seen.