Regarding to ATI vs Nvidia. There is S3. There is Trident (although not a sane choice) and Sis for the low-budget people. The world simply don't end with ATI or Nvidia. Besides that, S3 has given and it will still give nice features and offer good quality for the money you pay.

ATI and Nvidia. Actually none is too good. On one side might be ATI on another one is Nvidia. What you have to admit is that ATI offers slightly better quality at the moment and a bit more speed for the money you pay. Drivers are quite buggy for both of them actually. Anyway, the Detonators thing is by far the most stupid driver set. They release drivers almost 2 times a month, and you never know if it is better or worst than the last ones. You don't know also where is the good part. If I have to be more critic, I would say that Nvidia has far more potential than ATI, but it does less because it cares less. It fully seeks profit all the way, and marketing. I am not saying that ATI is an angel, but clearly none of them are respecting the costumers.

And yes, "Nvidia, the way it's meant to be played!" is a crappy example. It offered some better compatibility with the mentioned games, but some boxes from 1999 were labeled "runs optimally on a Winfast (GeForce) card" when you have 3dfx mode. And the last thing: anyone Glide, S3Metal? These direct interfaces, if used properly can smoke the best direct3d performance with 20 fps in some cases, not including the compatibility and quality. But they don't want to develop such direct interfaces, and go happy MS Direct3D (sometimes I have some doubts on how direct it really is)...

I know what the latest and greatest things on the 3d gaming and accelerators market look like. I saw Unreal 2 and Unreal Tournament 2003. I saw the latest Radeons 9800 XT and the latest GeForce 5900. The games are impressive, even very impressive in certain aspects, and this is a thing to take note, as not many games look so good. However there is something that is not so good in terms of quality, again.

I looked at the green picture. On the floor you can see a metal plate. The "glitch" is the bilinear filter, am I right? We are in 2004, and this thing simply shouldn't happen.

I admit that the faces look excelent, and that the weapons have very high details. But having details as polygons is not the full job. The sky also looks realistically. But I know these games, and I know that they are not very Impressive if you ever saw Virtual Reality simulations, and chainlinked rendering in action. I don't compare these situations to a PC. But we could ask more from the GPU and game developers, as it is not impossible to push the envelope.

The explosions. Well, explosions, glares, glows are hard to make. First of all the efect is very dependent on if you use a particle system a 3D object, or a mapping of any kind. But these explosins are not the best I have seen (or at least not the best they can be). But any explosion consumes tons of processing power (right!) and it could drop the framerate quite a lot. This is one of the reasons for which the high detail textures were disabled in the first Unreal Tournament 2002.

Glide, S3 Metal. Most of the current needs of graphical devices and games are based on the assumption that Direct3D would deliver nearly the maximum thorughput of a video card. This is completely wrong. Some developers discuss in private that Direct3D might already hit it's limits because of the complexity and the high level of abstraction layers it has, having very few options to handle directly how a request is being solved. Not mentioning the known fluctuation of performance tipical for d3d. Also, d3d is slightly different when you read MS books, and when you actually implement it. ATI and NVvidia would have the money to invest in a direct video interface. But they don't do it because it simply works well enough for their expectations, they get their profits and are not willing to constantly invest in this direct interface.

Most people know, but they won't recongize that doing a D3d app cost more than an OpenGL one. But they choose D3d because it is simply the most used interface, and many videocard manufacturers improve their d3d performance over time, as opposed to OpenGL. Less to blame the developers is because of the documentation. OpenGL has scarce documentation and has no implementations on anything but C (as far as I know). Also, when you get the dx sdk you get tons of examples, documented instructions. This is one sane and acceptable reason for developing on d3d. Others are not so much.

Whoa Guys! 

Going a bit off-topic here, but allow me to respond to your comments! 

In aspects, I think you are both right.

Yes, we have advanced in terms of quality and performance in our graphics, this is true.  Some of the games I see today running on the high-end Radeon cards still amazes me!  We have really come a long way in the past few years.

Although, I can understand parts of what Andrew Boiu is saying.  Yes, we do have all these new improved features as super fast trilinear filtering we could have only dreamed of a few years back and the better support we now have for running games in FSAA.  The use of Pixel and Vertex Shaders.

But the fact remains, that over the past few years that although the technology has progressed, it has not progressed as fast as it was looking to do in late 2000.

For the past year (or more so even) the Radeon and GeForce series have only made updates to give better support for aspects such as DX9.

What we usually see a lot of is, increased GPU clock speeds and more memory added to the cards.  You must admit that is the main key to all the "improvements" in the cards.

Now we are presented with too many cards to choose from on the shelves, from each brand, ranging from "watered down" versions to full heavy loaded high-end models.

This shows they may be running out of ideas in the R&D department. 

The PCI Express range is coming, true.  But what are the companies really gaining from this?
At this point it only presents more bandwidth to use the current technologies we have now supported in games, only to run them in higher resolutions.
It's also another money making racket for the big dogs to cash in on, as it adds more variety to their series of models to spend your money on.

In the late 90's, 3dfx were way ahead of the game in terms of R&D (Research & Development, for the numbnuts).
They started designing the Rampage technoology back in the day when they were putting ideas together for the Banshee! The only trouble was, was that 3dfx didn't realise they were onto a good thing, and the ideas got scrapped.
If that kind of effort and know-how was around today in the companies that still remain, who know what kind of cards they could have produced by now. 

Oh dear, nVidia are at it AGAIN!

Go check it out...

http://www.forum-3dcenter.org/vbulletin/showthread.php?s=&postid=1845416#post184...

And YES, it has been confirmed that the pictures ARE genuine!

Indeed. Pure marketing ploy. Corporate rule.  Get your claws into the opposition before they do to you.

I see a lot of words from nVidia, and the fact that they dis the x800 technology.

I don't see ATI mouthing off at nVidia's faux pas.

I think ATI will wait for the two cards to be pitted together in competition and let them speak for themselves.

We'll see who's laughing then, and I assure you it will grow very quiet in the nVidia corner!

My first ATi card was a Hercules 3D Prophet 9700 Pro 128MB 256Bit DDR. well gotta say never took nVidia ever again!!!

The FSAA and AF is way sharper than that of the GeForceFX series that time and so was the performance while using FSAA & AF in nearly everygame.

My last nVidia was the Leadtek Winfast A250 ULTRA TD that's a GeForce4 Ti-4600 , still useing it now. That was to me one of the best GeForce cards ever.

Now I use a Connect3D 9800 Pro 128MB 256Bit DDR. in FSAA & AF it's sometimes 70% faster than my older R300 aka Radeon 9700 Pro. Normally I Play C&C Generals and FS 2004 and that is where I tested both cards. the R350 has actually some slight improvements with performance and  HyperZIII+.