7.27 Detonators to be released Christmas Day?

[BFG10K]

BenSkywalker:

When did you first say this? I went around with MrNSX over this quite some time ago, prior to the Det3 6.31s if memory servers.

A while back I used to ocasionally post in a another hardware forum (it's closed down now) and there was an ATi engineer who posted there regularly. When the GF2 Ultra was released and the Detonator 3 drivers first came out, myself, the ATi guy and a few other video card techies started discussing on how the GTS was suddenly able to beat the Radeon with the new drivers.

Anyway two of those guys had GF/GF2 boards and after a lengthy discussion coupled with some tests, we came to the conclusion that nVidia are using a forced 16 bit Z-buffer. Those guys also posted up screenshots of Quake 3 and pointed out the new visual artifacts at long distances (ie Z errors) which had not been present with the Det 2 drivers.

I'm wondering if he is just taking something you had made a comment on and trying to spread it as truth that he has particular technical knowledge about.

Absolutely not. MrNSX is a very smart guy who has a lot of industry insider information. It's a complete coincidence that I just happend to start a thread about the issue and he joined in. The fact that he not only agreed with me but also explained the issue even further (ie 20 bits instead of 16 bits) gives me good reason to believe it's true.

 

[BenSkywalker]

BFG-

"MrNSX is a very smart guy who has a lot of industry insider information"

He is an Apple industry insider, he doesn't even know what is going on with the ATi PC drivers(and he works writing drivers for ATi). I have told him exacting circumstances in which you will see rendering errors using ATi's PC drivers, running freely downloadable software in many cases, and he replies by saying that he doesn't know what the PC driver guys are doing??? WTF? He knows exactly what nVidia is doing by looking at source code for drivers for a different platform, different OS, that hasn't been released yet anywhere and he doesn't know what the drivers are doing for hardware he works with every day?

He is wrong, it is easily provable, I have told him how to do so. Clipping errors in the distance are far from being an immediate indicator of Z-Buffer issues, particularly when combined with the other factors of slower performance at lower resolutions and the TNT2 is slower. We all saw how much the GF2MX benefited from the Det3 drivers, moreso then the GTS in many aspects. If the Det3s were dropping to 16bitZ then the TNT2 would also be seeing a nice performance boost.

That ignores the fact that there already are industry standard tests for such things, and they all display that nVidia boards are in fact running 32bit Zbuffer.

Look at how he acquired these supposed facts. He claims he can prove this because he has looked at the code for pre-release Mac drivers for the GF2MX. WTF does that have to do, in any way imaginable, with what level of Z precission PC drivers are using? Nothing, nothing at all.

Looking at that further, how exactly did he get the source code for this? He claims to have pre-release source for non announced drivers from nVidia, and we should just assume that this is the truth?

This subject was asked over at B3B by one of the Ars reader(linkage), with the anti nVidia sentiments held by many of the board frequenters there don't you think someone would have been able to find a single thing that backed this assertion?

Look at the facts- He works for ATi, nVidia's number one competitor. What's more, he works for the Mac segment of the company who is facing the very serious threat of nVidia taking over, or at least a big chunk of, their market. Why should anyone believe anything he says? When DaveB3D had something to say about a poor nVidia implementation, he could give very specific examples, MrNSX has absolutely nothing, not one single thing to back his claim. With all the various proffesional OpenGL applications that stress ZBuffer accuracy, he can't find one that backs his assertion.

Also, if you are running 16bit Zbuffer it effects things in close proximity also, unlike the issues that some have brought up with earlier Det3 revisions.

 

[StickHead]

Yeah, so where are the stupid things I checked NVnews.net, 3DGPU.com and nVidias site, no metion of them.

 

[BFG10K]

BenSkywalker:

He is an Apple industry insider, he doesn't even know what is going on with the ATi PC drivers(and he works writing drivers for ATi).

OK, I did NOT know that MrNSX writes Mac drivers for ATi. All this time he never actually said what he did. Now I finally understand why he would be attacking nVidia and defending ATi.

I have told him exacting circumstances in which you will see rendering errors using ATi's PC drivers, running freely downloadable software in many cases, and he replies by saying that he doesn't know what the PC driver guys are doing???

Yes, I know what you mean. He often posts a lot of highly controversial pro-Mac stuff (not just about video cards) and then hides behind NDAs when asked to prove it.

But I will admit he is a smart guy and does know a lot of stuff about computing.

We all saw how much the GF2MX benefited from the Det3 drivers, moreso then the GTS in many aspects. If the Det3s were dropping to 16bitZ then the TNT2 would also be seeing a nice performance boost.

OK. I'll go back to the orginal thread because I have a few more questions to ask you.

He claims he can prove this because he has looked at the code for pre-release Mac drivers for the GF2MX. WTF does that have to do, in any way imaginable, with what level of Z precission PC drivers are using? Nothing, nothing at all.

Not only that, but why would nVidia be handing out the source code to their competitors (ie ATi)? Most source code is intellectual porperty that is protected by a wide number of laws.

Looking at that further, how exactly did he get the source code for this? He claims to have pre-release source for non announced drivers from nVidia, and we should just assume that this is the truth?

I agree, some of his "evidence" is highly doubious and seems entirely fabricated.

Also, if you are running 16bit Zbuffer it effects things in close proximity also, unlike the issues that some have brought up with earlier Det3 revisions.

It does? I didn't know that. I guess you learn something new every day in this forum.

 

[Atom Ant]

Sooooooooo where's the drivers??