NVIDIA's offical response to Futuremark

[Zenikase]

Originally posted by ImaginaryFiend
So PS 1.X is still for register combiner hardware, but MS made game programmers' lives easier by defining a much nicer way to "program" them.

But wouldn't that also classify PS2.0/3.0-level hardware as using register combiners? It may not be as rigid and one-track as previous generations but they're still using different specialized registers together in the same fragment program.

[solofly]

Originally posted by Hellbinder
It more than a little irritates me that you would print something like that while you hang out here and apparently have no problem with the Nvidia Zealotry and such that goes on.

It's very simple. nVidida users don't bash ATI users like ATI user bash nVidia users. Rage3d is a good example of what i just said. Not only that, they insult one another on top of it. I know of ATI users (including me) that are disgusted by rage3d attitude. I hope NV40 can come back and blow ATI away in every way so we can go back to peace and quiet as it was even a year ago.

[digitalwanderer]

Originally posted by solofly
It's very simple. nVidida users don't bash ATI users like ATI user bash nVidia users. Rage3d is a good example of what i just said. Not only that, they insult one another on top of it. I know of ATI users (including me) that are disgusted by rage3d attitude. I hope NV40 can come back and blow ATI away in every way so we can go back to peace and quiet as it was even a year ago.

That isn't an ATi/nVidia thing, that's an overwhelming input of newbies thing!

It's almost seasonal over at Rage3D, everytime ATi puts out a new card there is a HUGE influx of newbie fanATics and they tend to be a bit over-zealous as they get all comfy with ATi.

I strongly do disagree with your sentiment, I find the fringe element crazy crowd to be too much on either side.

[R.Carter]

Originally posted by ImaginaryFiend
Regarding the F-Buffer, the paper was by Bill Mark, et al., at Stanford.

Why don't you link to the publication?

The F-Buffer: A Raterization-Order FIFO Buffer for Multi-Pass Rendering by William R. Mark and Kekoa Proudfoot.

[Hellbinder]

ImaginaryFiend

I really don't understand why so many of you are determined to think Nvidia is evil. They're not "the man." Remember, they're not much more than a startup. They're less than one one-hundredth the size of Intel. They even have fewer employees than ATI.

True Statement, ATi currently has about 2x the Employees of Nvidia.

Also, think about how much they've done for you. I guarantee that our game playing would not be anywhere near the quality it is without Nvidia,

yes, you could say this. However you could also say this about 3dfx, PowerVR, Rendition, and ATi. Currently it is true mainly from the insane ammount of Competative driven Technology. inovation.

whether or not the card currently in your machine was made by them. They invented register combiners. They invented vertex programs. They invented cube maps in hardware.

Nvidia were the first to use register combiners. They did not invent Vertex Programs. They were the first to impliment Cube maps in hardware. On the other hand 3dfx was the first to impliment Multi texture, Matrox was the First to impliment bump mapping, , and Displacement mapping. ATi was the First to impliment Z compression, and Heirerachical Z. They were also the First to impliment Pixel and Vertex shader 2.0, they were the First to impliment N-Patches in hardware, They were also the First to offer high performance 32bit color, S3 created S3TC, Matrox was the First to offer a 256bit bus, Ati was the first to offer an 8 pipeline Architecture, Nvidia was the First to offer a 4 pipeline architecture..

I mean the list goes on and on.. Its a Stream of Progression accross many companies, and many peoples ideas.

They are the only company that offers fragment programs good enough to do "cinematic shading". These are huge contributions.

Are you saying you have not looked at the serious issues Nvidia is having delivering Dx9 Rendering??? Isn't this what this entire thread is really all about??? Or does *Cinematic shading* in your book use a lot of old style Fx12 support and FP16.

http://www.hardware.fr/articles/468/page4.html

http://www.beyond3d.com/forum/viewto...ht=3dmark03+iq

Is just a small example, and it coincides with the evidence on why Nvidia is Cheating in 3dmark and trying to discredit them. Nv3x cannot seem to handle the Percision they are claiming makes them better. On top of the fact that they Do not adequately support several Key Dx9 Features. Such as addressed in this thread at B3D

http://www.beyond3d.com/forum/viewtopic.php?t=5451

No Floating point Textures, No Floating point Cube Maps, No Multiple Render Targets. All correctly supported by the Radeon 9700pro for almost a year now. Are you basing your statement soley on Theorectical instruction support?? If so.. then again, As i stated already the F-Buffer clearly pushes far beyond the limmited abilities of the GFFX architecture.

As for ATI, they are just as big a company. They play all the same moves as Nvidia.

Really, Ati sent out documents like the now Infamous Kyro PDF???

http://www.theregister.co.uk/content/3/19426.html

It was Discussed in detail at B3D by many people in the industry, unfortunately that Thread is no longer available becuase B3D Changed hands and got a new Server/Owner a while back.

Lets not forget what Nvidia did With Quak

http://www4.tomshardware.com/blurb/20020825/index.html

Where Nvidia shopped around until they found a couple Sites that Ran the Story. Mainly [H]. Of course If lowering IQ for speed is Cheating and Evil then lets look at some older Reviews like this one.

http://www.hardocp.com/article.html?art=MTEzLDM=

Where we see just one example of how much lower the Texture details are in the entire Quake Game on the GF2. You can also see what they Did with S3TC which was nothing but a cheap driver hack for speed as well, that one dates all teh way back to the Origional Geforce.. Yet no one Screamed cheating.

What about this...

http://www.3dvelocity.com/reviews/gffx5800u/gffx_5.htm

Where we can see the detail levels lowered yet agian for the sake of speed. Yet its not cheating??

They haven't contributed as much to the industry, and over the long term and for the near future, their products aren't as good (though they were ahead lately).

i already posted a short list to some of what all the various companies contributed to the industry. I think i have already proven that their products are currently Flat out SUPERIOR. I see no reason why this pattern wont Continue. ATi has had superior products on the Market since the Introduction of the origional Radeon. It was simply a matter of their management choosing only to release a new product once a year at most. The Radeon is clearly more advanced than the GF/GF2, the 8500 is more technically advanced than the GF3/4, the 9700 is STILL more advanced than anything Nvidia has on the market, and the 9800pro is a step beyond that.

If you only consider Raw speed instead of features then You can make a case for Nvidia... at least up until last summer.

Remember that ATI cheated on 3dmark, too. ATI cheats on Quake3. ATI cheats on all trilinear MIPmapping and on aniso filtering.

No ATi *optomized* Nvidia Cheated. Well actually Officially now nothing anyone did was cheating. However Perhaps you should read Tim Sweeneys and John Carmacks take on the matter. Here is a sample.

http://www.beyond3d.com/forum/viewto...c=6041&forum=9

You know perfectly well (as you are likely an Nvidia employee) that There is a big difference between reordering some shader instructions and inserting clip planes, Lowering Percision to below Dx9 standards in a Dx9 Benchmark as seen here.

http://www.beyond3d.com/forum/viewto...light=3dmark03

ATi Cheats with Trilinear??? Or do you simply mean that their AF implimentation on the 8500 was limited to 16 sample Bilinear filtered Anisotropic filtering...

ATi Cheats in AF?? Or do you mean You dont like their Hardware adaptive method. How do you defend that statement when according to your own rules Nvidias nearly entire Current AF offering is *cheating*. As they Clamp AF to Bilinear for several modes as well. Which affects ALL angles not just one or Two odd angles. There are several Reviews such as at [H] that show in detail what Nvidia is doing IQ wise to get their AF speed up.

But is all that really cheating???

As for PR fiascos, remember that ATI got in deep trouble with Id for leaking a Doom 3 prerelease last year.

No, Ati did not leak doom 3, i was told this point black by Terry Makedon ATi's Software Product manager in a private conversation. There is also info about this at Shacknews.. and a couple other places i believe. where Tod Hollenshead says point blank that They have no information that it was Ati who leaked it.

They got in big trouble with Apple for leaking specs of Apple's new machines before MacWorld.

Yes, you are right they screwd that one up big time. But what does that have to do with the kinds of things we are talking about here?

They got themselves in the Quake/Quack fiasco.

Already covered, A Clear example of why Nvidia Corperate is underhanded.

They trashed on the Geforce 4 MX's name because it's a DirectX 7 part and stated that ATI would always name their products clearly, with the first digit being its DX version. But then they misled us with the Radeon 9000 and 9200, which are DX 8 parts.

Is this a complete and total Fabrication? First of all ATi trashed nothing about the GF4MX??? can you post any and i mean aything to back this up?? ATi has made no public statement whatsoever about the GF4 MX that I can find. When Did ATi state they would name their products clearly in regards to the GF4????

For the record...

Radeon 9xxx

The first TWo numbers stand for the hardware generation of the card, not the DX version it supports.

The second two numbers indicate the realative performance Within the product generation. (at least this is what I think i was told for the second Two numbers)

I really think that neither Nvidia nor ATI are evil. They're just trying to make and sell as many of the best products they can, as expected.

Nvidia has a history of Unethical behavior, some of which I showed examples of.

Futuremark isn't evil either, but Nvidia's right that their benchmark isn't written the way games are written, making it not very useful for anyone.

thats interesting perhaps you should read this..

http://www.beyond3d.com/forum/viewto...ark03&start=40

Then why does Nvidia's developer website have a demo and source code for some of the very same Techniques used in 3dmark prior to its release? Why did David Kirk give a presentation advocating that exact same shadowing method?

Further non of the other Beta partners nor Developers that have commented on the issue have had anything negative to say about the way 3dmark03 is Coded. In fact there are a few posts from developers at B3D that indicate that exactly the oposite is true. Could the TRuth simply be that the Nv3x architecture has to many problems to adequately handle Real Dx9 functionality or True *cinematic shading* as you put it?

It makes perfect sense to me that Futuremark and Nvidia would want to resolve this dispute; it's to both their benefit and it doesn't mean that either party caved or that either party strong-armed the other. Disputes are supposed to be resolved. That's progress!

There is every indication that Nvidia Didnt get what it wanted and resorted to strong arm tactics. If you read the statements released by Patric Oja he clearly states that he as to use careful wording due to *leagal* reasons. here is a Quote from the man

First I must admit that there is very little I can comment about the joint statement between Futuremark and Nvidia due to legal aspects. What I can do is answer some frequently asked questions about this and quote some parts of the statement.

I am out of Time for now, But I will re address the rest of your comments later..

And I officially Agree with several of you that I was out of line origionally with the waY i was handling the Situation.

[Hellbinder]

It's very simple. nVidida users don't bash ATI users like ATI user bash nVidia users. Rage3d is a good example of what i just said. Not only that, they insult one another on top of it. I know of ATI users (including me) that are disgusted by rage3d attitude. I hope NV40 can come back and blow ATI away in every way so we can go back to peace and quiet as it was even a year ago.

im sorry, that may be what it currenly looks like. But Historically Nvidiots got their name for a reason. You have a very one sided view of what is really going on. It is really unfair that you think this, totally blowing off the long history of Nvidia Fan Sites, Nvidia Biased Review sites hammering ATi for about 3 years straight.

It is only recently changing.

Also consider that Rage3d has about 300,000 active posters. Nvnews has... maybe 1,000????

[Hellbinder]

This is a good question because it's been kind of confusing to everyone ever since DX 8 came out.

Register combiners can be considered either as fixed function or as programmable. Think of them as a bunch of black boxes that do math - dot products, component multiplies, subtracts, muxes, etc. Each box is a fixed function. However, you configure (program) how these boxes are wired together. By wiring them in different ways you can make them compute different mathematical expressions to get different shading effects in the rendered result.

Register combiners are really nothing more than an Extention of the origional Concepts put forth in Bump Maping. What he is talking about here is setting various Hardware States.

In OpenGL, each wire connecting one box to another was described by a function call that ended up being about four lines long. This was very painful. I'm not sure how register combiners were programmed in DirectX 7 - something similar, I think.

Hmmm, each IHV uses different OpenGL Extensions to accomplish what you are talking about. It really is not much different than what DirectX does. The main difference is that DX is a set standard with a universal compilers.

However, for DX 8, Microsoft came up with a different way of expressing how the wires are hooked together. It's a much shorter syntax that looks like a very simple assembly language. They called it "Pixel Shaders 1.0". Then they modified it for the Geforce 4 and called it "PS 1.3" and they modified it for the Radeon 8500 and called it "PS 1.4".

Actually its Pixel shaders 1.0, 1.1, 1,2, 1.3, 1.4 and 2.0

Of which Pixel shader 1.0 while suppotedwas never really openly adopted. GF3 introduced Pixel Shader 1.1 which was the Dx8 standard. Pixel shaders 1.2 and 1.3 simply added some Color constants and did not alter Functionality or performance in any way. Ps 1.4 is a Much bigger step forward, especially with its ability to support Greater ammounts of Fragment processing. Of course PS 2.0 is where we are at today.

So PS 1.X is still for register combiner hardware, but MS made game programmers' lives easier by defining a much nicer way to "program" them.

True becuase DX is a set universal standard. However even now Nvidia with the Nv30/35 do not expose all the features that are supposed to be present in DX9, Or PS/VS 2.0. They do seem to include more support from their own OpenGL extensions. Which is likely becuase they have to Lower percisions to below DX9 Spec to get adequate performance. There is also the issue of Hardware Bugs in the Nv30 that can only be worked around with Custom OpenGL Extensions.

[R.Carter]

Originally posted by Hellbinder
Nvidia were the first to use register combiners. They did not invent Vertex Programs. They were the first to impliment Cube maps in hardware. On the other hand 3dfx was the first to impliment Multi texture, Matrox was the First to impliment bump mapping, and Displacement mapping. ATi was the First to impliment Z compression, and Heirerachical Z. They were also the First to impliment Pixel and Vertex shader 2.0, they were the First to impliment N-Patches in hardware, They were also the First to offer high performance 32bit color, S3 created S3TC, Matrox was the First to offer a 256bit bus, Ati was the first to offer an 8 pipeline Architecture, Nvidia was the First to offer a 4 pipeline architecture..

Well, Matrox was the first with dual-head technology in their G400 card and both the G200 / G400 supported 32 bpp.

Was the Nvidia TNT2 out before the Matrox Millennium G200? I think the TNT2 was the first 32 bpp from Nvidia wasn't it, or did the TNT also do 32 bpp? Course at release the G200 wasn't all that fast.

Dunno... think TNT2 showed up mid 1999 or so while the G200 Millennium was about a year earlier August 1998 wasn't it.

Oh well...

[Hellbinder]

Yeah dude.. the G200 was mid 98 if i remember and it offered 32 bit color.

Just to make it clear I was not saying that Ati was the First to offer 32bit color, but the first to offer a really low Performance Hit for it.

If you remeber back in the day 3dfx made some pretty brash statements about how Nvidia was Hacking their way to 32bit and not doing it correctly.

Who knows though... i was only really trying to point out that this whole thing is one big group effort. And even Nvidias Patented Register Combiners are only an extension of Matrox's Bump mapping hardware.

[Zenikase]

Hellbinder, could you elaborate on the changes from PS1.3 to 1.4? I know that nVidia got some flak (especially from certain big-name developers) for not implementing support for this in their NV2x line, not even in NV25 (after the R200 was released).

Originally posted by solofly
It's very simple. nVidida users don't bash ATI users like ATI user bash nVidia users. Rage3d is a good example of what i just said. Not only that, they insult one another on top of it. I know of ATI users (including me) that are disgusted by rage3d attitude. I hope NV40 can come back and blow ATI away in every way so we can go back to peace and quiet as it was even a year ago.

Because nvidia is no longer top dog. You weren't around these forums when the r8500 was released with its horrible drivers. I think the r9700 shut up most of the fanboys because it dominated the gf4, and then we had the terrible NV30 launch.

There were a lot of nvidiots taking potshots at ATI all the time here in the past. Hell, I was one of them Of course back then I was ignorant and didn't know jack about anything technical.

[digitalwanderer]

Originally posted by Hellbinder
Also consider that Rage3d has about 300,000 active posters. Nvnews has... maybe 1,000????

No FREAKING way! R3D has that many posters?!?!

Geeze, I thought it was a quiet little place.