16 pipes for R420 too?

I don't see anything unusual.

 

I must be blind.

 

Hmm.. maybe its just me. Time to take those pills again.

Seriously I was referring to those blue boxes. Yesterday there where only one, today there's two of them.

 

Are you talking about the pictures above and below the UT2004 link to nzone? There could have been 2 games there before, UT2004 and Battlefield Vietnam (both are on the software download page. So who knows.

 

Yes precisely. Perhaps all three pictures will be "wavy ocean blue" tomorrow and on tuesday the big GF 6800 picture will arrive with the same blue background.

 

*jumps on my tippy toes happy with joy* YAY!!!!!!! I won I won!

 

Come on, DC, you know this is BS....EVERYTHING you post about is alway put into the perspective of Poor ol' nVidia ....... If something is said in praise of any product other than nVidia you are there to protest. It''s your attitude - and you just don't get it..... Just look at how many times you have been called for your taking someone elses post in a very self centered, nVidia centric way.......

 

Wrong. I’m treating part with different market shares differently. And, by the time widespread Grantsdale option has taken the low end discrete solutions will be very different.

Do you think there are many developers that do treat it as Dx9? Do you think all of MS are overwhelmed with its performance? Have you been talking to a cross section of the industry?

Nope. Another one you imagined DC, I’ve not said anything of the sort. I’d contest that 5200 will probably have better DX9 performance. However it is not enough, alone, to push the industry up from DX7 to DX9. You need the mass market leader to do that – yes?

Again, no. They will not treat Grantsdale because of performance, but because of compatibility – how many times have you managed to overlook that? One it becomes widespread that will be the shift developers look for to up to base target from DX7/8 to DX9. 5200 doesn’t have the presence to make that feasible, so there is no point in fixing it to DX9 and having poor application performance (which, yes, you will do on Grantsdale anyway, just as you do on any of Intels current solutions, but as I;ve said numerous times the application will work).

 

Who said it will run the same code paths? I'm saying that withing a few years it'll likely be the catalyst for DX9 being the baseline because it will run it.

 

Dave Baumann wrote:

I think you are being way too generous (like other industry analysts) in your estimation of intels market share as it relates to gamers.

You need to remove laptop, business, internet only and educational use only PC's from Intels raw market share before you have a relevant number for gamers. Even then you need to consider that perhaps 15 - 20% of that figure is in all probability using some sort of add in card and not the integrated solution.

 

you should look at the most popular games on the market currently

 

I sell quite a lot of internet/soho boxes, usually to women with youngish families. trust me, I know their gaming tastes are different to ours.

However, that does not alter my original point, most PC's actually making use of integrated Intel video will never have a game played on them, or perhaps, very, very rarely.

 

How do you know what people will do with their comptuer once they get it? They may buy it for education/internet, but I'd be surprised if the average person would never try playing a game on it. And even if .5% of the average person does play a game or two, the game companies could lose out on quite a few sales by not providing basic support for integrated graphics.

 

First of all I stay in contact with my clients. I'm not a sell it and forget it sort of businessman, unlike a lot of people who try to sell PCs (hint for those who want to sell PCs for a living, this is the most important part of your business, period.)

Secondly, and speaking only for myself here, I've never supplied a PC with less than a TNT vanta powering the graphics, so I'm confident that if a client wanted to try a game they could easily do so.

 

Why is it BS? First of all, I have never "flipped" my argument on features vs performance. I have been consistently pro features since the beginning, an advocate of early adoption, ever since the 3dfx days on this BBS. What I do is show how other people flipped by turning around their own arguments on them.

Secondly, a defense of a bogus criticism of Nvidia does not constitute a blanket amnesty for everything they done, or an uncritical and blind following and agreement of them.

If I end up defending Nvidia more, perhaps it's because this board is full of more bogus criticisms of NVidia. Except for Chalnoth, how many people on this bbs continually attack ATI? And I will point out that when mistruths have been spoken about ATI, I have responded and corrected people.

 

Just because the card might not be listed there doesn't mean it won't work with it.

 

You hinted at it by suggesting the "dial down" approach. If developers were supposed to write separate code paths for the Grantsdale, you should have said it. What you said gave the impression that users would simply drop resolution, textures, and other details, but the shaders would run unchanged. The more likely scenario is that any DX9 shaders that run on Grantsdale are going to be more like PS1.4 shaders in length and features used.

 

"Dialling down" does not negate the possibility that different codepaths will be used - of course there will be different code paths becuase in many titles we'll be seeing Shader 3.0 and Shader 4.0 paths by that point, and that doesn't negate the possibility of separate high and low paths. However, what you'll also note is that I say this will probably be the catalyst for a full HLSL programming approach - again, that does not negate different paths, it likely makes it easier to do such.

 

So you never sell a computer with integrated graphics only.

Well, I'd bet there are plenty of people who buy games that only have integrated graphics.

I think it'd be a horrible business deicision for a game developer to only target gamers, and totally ignore the average person/grandma/etc. There are a lot more average computer users than gamers, even casual gamers included. They make the game for the widest possible audience and that means there's a good chance that person will have integrated graphics. If they don't provide some kind of basic support for integrated graphics, their potential customer base will shrink substantially.

Do you only sell to people who have a certain level of computer knowledge? Or if all someone wants to do is send email, would you provide something that will allow them to do so? Of course not. If you only sold to people who could touch-type or add/remove hardware themselves, you'd be cutting out the majority of your customer base.

I think it's a similar thing with games and integrated graphics.

 

Is MS satisified with Grantsdale's performance? Are devs satisified with it? Why are you singling out the 5200?

I asked you if you had any evidence that Grantsdale performed better, and I also asked you if 5200 users could use DX9 by "dialing down" as you suggested for the Grantsdale, you agree with both.

And NVidia's mass market chips have no effect on the market? Steam's hardware survey shows otherwise. It shows the GF4MX dominating, and 5200 as the most widespread DX9 card currently. If Nvidia did not have that much of an effect, people wouldn't have been so pissed off at the GF4MX being a DX7 card.

I understand what you're trying to say, I just don't agree with it. It's not enough to simply be API compatible, you've got to have performance. Without performance, devs will have to treat Grantsdale and 5200 as "defacto" DX8 cards, writing far less complicated shaders for them that do not take advantage of the real benefits of DX9. Sure, they'll be calling D3D9 functions, and compiling to "PS2.0" profile, but they will likely be using a build system that generates two kinds of PS2.0 shaders. PS2 in name only, but effectively a PS1.1/1.4 shader in complexity.

Yes, once Intel has Grantsdale for a few years, devs won't have to worry about old DX8 cards as much more and can stick with a single API/HLSL. But their concerns will still be split. They'll just have to worry instead about a mass of shitty performing DX9 cards, and a smaller population of mid-range and high-end DX9 cards.

It's not really advancing the industry forward as much as you claim.