Nv40 16 full pipelines- The Inq.

Is that what I said? You certainly do take liberties reading things into my words.

That's nVidia's PR tag line.

 

What liberties? You atrributed some of the NV40 rumors to Nvidia, have you not?

Oh yes, ATi doesn't have such policy. My bad...

Although the Mpact was canceled, an ATI Technologies spokesman said that "we've said we're going down the integrated path, but we don't comment on unannounced products."

MMC: John Swinimer says we can't comment on unannounced products, but we never stop working on new products.

An ATI representative did not respond to requests for comment, but the company typically does not discuss unannounced products.

"While I can't talk about unannounced products, I can tell you with all certainty that ATI will not be announcing any new version of the ALL-IN-WONDER on Monday, September 16. "

Stephen Orr: Certainly it’s true that I cannot talk about unannounced products (unreleased is a different matter and we certainly have not announced anything in the area of PC DTV. As to the TV Wonder, I already indicated that MMC Web postings will support TV Wonder going forward so we clearly support the TV Wonder. The rest I think I answered in the first interview question.

Next time, hit google instead of plumping down idiotic rolleys smiles.

 

Just a thought for the nV doubters:

nVidia have produced 4 killer, market leading products:
Geforce1
Geforce2
Geforce3
Geforce4

All these products were clear performance leaders at launch. nV has produced one dud family, the FX line as we all know.

ATI, on the other have produced a single killer product, R300 and it's derivatives (i wouldn;t count R350 or R360 a significant enough revisions). Other than that, it's a story of largely worthy, but ultimately second place products eg R100 and R200.

Now I'm no nV fan boy, the FX line smells and what nV have been doing with drivers of late stinks. But desperate times call for desperate measures, and nV pretty much owed it their shareholders to do whatever it took to close the gap and not admit their products were second best.

The main point I'm making is this: nV have a strong track record for leadership and a single messed up product line does not erase that. I don't know whether NV40 will rule, but it seems pretty unlikely that nV will allow themselves to trip up a second time. The sensible money says NV40 will be at least as good as the best ATI can come up with.

 

So you count GF3-> GF4 as a significant revision but not r300--> r360?

Thanks for making it clear where your personal bias lies.

 

I knew someone would pick me up on the GF3/4 thing. Firstly, if am am wrong that doesn't necessarily make me biased, and I have to say i am not at all biased. My ideal scenario is near parity between nV and ATI, constantly pushing each other to produce better products for all of us. This is more or less what has been happenening for the past few years. Don't get to hung up on my use of the term killer - the delta has actually been pretty small in practice most of the time, but my point is that nV have a well established track record of leadership that many here seem to have overlooked thanks to the failure of the FX line.

Furthermore, the GF4 at least adds a vertes shader to the GeForce3's tally, although I'll admit it's a little marginal as to how 'new' the GF4 was. Anyway you look at it, it's still 3 to 1 at worst in the age of modern, T&L capable hardware. In fact, rather than me being a nV fanboy for points these rather obvious facts out, you'd have to be a bit of an ATI fanboy to strongly believe that ATI was likely to have an advantage for the next gen GPUs. A neutral observer would have no choice but to choose nV.

Oh, and wasn't Geforce3's pixel shader 1.1 spec vs 1.3 for GF4, could be imagining this so pls correct me if wrong...

 

Hmmmmmmm lets take a look at the products NV has released.

NV1
Riva 128 family
Riva TNT family
Riva TNT2 family
GeForce 1 family
GeForce 2 family
GeForce 3 family
GeForce 4 family
GeForce FX family

The NV1 was a unique architecture, and a market failure because Nvidia misread the market. The Riva 128 was a reasonable product, but by no means the market leader.

Then with the TNT based cards Nvidia started making a concerted effort towards market domination. TNT and TNT2 were direct competitors to the Voodoo2 and Voodoo3 families and managed to end up defeating them-- if not in direct retail, with OEM inroads.

GeForce was Nvidia's next salvo, and the first one that really took the total undisputed lead. GeForce2 was more of a revision than a whole new architecture, and served to capitalize on that lead. GeForce3 did it again, and once again they built on their lead with a revision, in this case the Gf4 Ti series.

Then ATI smashed them with R300, and GeForce FX was outclassed.

To me that looks like Nvidia having domination for two generations, plus refreshes, and ATI for one.

Neither company is out of the woods yet.

 

Or else ~3 years vs ~1 year.

 

I don't really want to get in a big argument about this but:

Geforce 1
0.22u
22m trannies
4x1 pipeline

to

Geforce 2
0.18u
25m trannies
4x2 pipeline

Geforce3 to 4 we have covered - same basic pixel pipeline set up, but new PS version and extra vertex shader plus a few other minor revisions. Same manufacturing process.

and

R300, R350 and R360 are all:
0.15u
110m trannies
8x1 pipeline

Geforce 2 wan't an all new architecture, but it was a much more significant revision on all counts than R350/60 was over R300 noteably using a new manufacturing process and adding a TMU to each pipeline. R300 uses the same basic pixel pipeline setup, the same shader definitions and the same manufacturing process, and virtually identical transistor counts.

Fact: GeForce 1 to 2 and 3 to 4 were more significant revision than R300 to R350/60. You have to accept that even if you don't think they deserved to be considered 'new' products as such.

EDIT: oh, and if we must argue, let's keep to the subject and have no mentions of gf4mx's or radeon 7500's confuse things, let's stick to flagship parts.

 

I think it's a bit of a push to say that the shaders were much different between GF3 and GF4. PS 1.3 isn't exactly light years ahead of PS 1.1 - definitely a refresh IMO perhaps similar to the inclusion of F-Buffer in the R3x0 refreshes? In fact, how much software ever specifically supported PS 1.3 at all?

I remember back when GF2 was released and everyone was very surprised that they could go from 4x1 to 4x2 whilst increasing the clock a fair amount. It rather took the wind out of 3Dfx's sales of the tardy Voodoo5. I would consider GF1 to GF2 a 'proper' generational step in the same way as Voodoo->Voodoo2.

 

Funny, they've never ever said that to me before....

 

I guess I am late to that party.

 

I'd left out the Gf4MX and similar cards for a reason. Not only weren't they flagship cards, but they also weren't part of the same pattern of advance.

I'll agree the Gf1-2 transition was a bigger one than R300/R350, as was Gf3-Gf4ti. In fact the only ATI transition I can think of that was similar to either was the R200 to RV250 transition and that's largely the other way.

However, regardless of the amount of change, both the Gf2 and the Gf4Ti were very much evolutionary developments of their predecessors, increasing functional units rather than featureset.

 

I'm just glad you made it at all, yours is a rare and dying breed and finding such a fine specimen as yourself just fills me with a warm and fuzzy tingle of anticipation.

This is going to be FUN!

 

Following their previous pattern-o-development then, wouldn't the NV40 be an evolutionary development of the NV30 then?

 

Agreed, and r300 to r350 was an increase in clockspeed and very little else. Anyway, this has reminded me that nV has been stuck on 4x2 since GF2. A jump to 16x1 would be pretty extraordinary in that context, eh?

 

Indeed I am one of rarest and finest specimen, even if I do say so myself. There is no doubt in my mind that I am the greatest human being who ever walked the Earth. So profound is my grandness that I am begging to seriously doubt my own mortality.

 

Wow, you even have nVidia's hubris....you're the real deal!

 

So great that you can't spell beginning correctly

 

If the numbers are an indication, then no, it wouldn't.
NV10 -> NV15
NV20 -> NV25
NV30 -> NV35
NV40

 

Doh! Thanks Xmas, I must have had a brainfart going.