ATI is the Summer King?

[Joe DeFuria]

As far as ATI goes, it's not trivial to do a die shrink. It will cost them more money than going nVidia's route...

Correct. 2 designs (0.15 and 0.13) instead of one.

and the end result will most likely be a lower-performing processor (given that it's going to be based on somewhat older technology) than the NV30/

Now that is a real stretch. First of all, for anyone to claim NV30 and R300 to be anything other than being based on similar technology is quite funny. From all estimates, both were planned to be released in the same time frame, it's just that ATI executed better than nVidia.

Second, having the design done first on 0.15, learning from it, and then tweaking it should enable ATI to be a better "shrink" to 0.13 than if they had just gone to 0.13 straight off. (Similar to GeForce to GeForce2 GTS "shrink".)

Finally, the fact that ATI gets their product out FIRST, means that they have at least a few months of higher profit margin on their part. That is the BENEFIT that ATI receives for spending the extra money to do two designs. NV30 will not be able to command the same type of profit margin upon NV30 release when it has to compete with the already existing R-300.

Honeslty, Chalnoth, sometimes your logic defies, well, logic.

 

[Ailuros]

a million per patent hmmm
i seem to remember microsoft gave them 200 milliion to speed up the research on the xbox gpu and that was how they spent the money...nice

Depends how you look at it. NV got the xbox deal because it was able to present a working prototype on time quite some time before the buyout. A deal which Gigapixel lost due to lack of execution.

 

[Oompa Loompa]

Now that is a real stretch. First of all, for anyone to claim NV30 and R300 to be anything other than being based on similar technology is quite funny. From all estimates, both were planned to be released in the same time frame, it's just that ATI executed better than nVidia.

Agreed that there is no justification for saying anything about the relative performance of the two designs at this point. I'd say that ATI and nVidia chose different strategies rather than crediting ATI for executing better, however.

Second, having the design done first on 0.15, learning from it, and then tweaking it should enable ATI to be a better "shrink" to 0.13 than if they had just gone to 0.13 straight off.

Assuming that the time frames being bandied about (0.15um R300 on shelves Sept. 1, 0.13um R300 on Dec. 1) are correct, I don't think you're right. 90 days is not enough time to make any design changes. By the time you've debugged 0.15um, 0.13um should have been long-since frozen. 0.13 will be a straightforward shrink unless they hold off a bit longer. Perhaps 0.15um yields will be good enough to permit that.

Finally, the fact that ATI gets their product out FIRST, means that they have at least a few months of higher profit margin on their part. That is the BENEFIT that ATI receives for spending the extra money to do two designs.

Indeed, that is the gamble. Can they recoup their costs by having the market "to themselves" for a quarter? Even if they can't (and given that the semi market is still in the toilet, it's no sure thing), it was probably a good gamble given the positive PR they'll receive.

There will, however, be a price paid for being out of step with the DX release schedule. Microsoft won't let them publicly demo anything using DX9beta1, and won't be there for joint promotion purposes. That's an annoying pothole from a PR perspective.

 

[Username]

Second, having the design done first on 0.15, learning from it, and then tweaking it should enable ATI to be a better "shrink" to 0.13 than if they had just gone to 0.13 straight off. (Similar to GeForce to GeForce2 GTS "shrink".)

Riiiiiiigghht. Designing for a specific size, then re-engineering it is worlds more efficient than actually designing for the specific size in the first place. Makes perfect sense. I mean, damn, why ever make a new architecture at all, you can just keep shrinking your old cores down easy as pie, even with the new features.


[quote="JFinally, the fact that ATI gets their product out FIRST, means that they have at least a few months of higher profit margin on their part. That is the BENEFIT that ATI receives for spending the extra money to do two designs. NV30 will not be able to command the same type of profit margin upon NV30 release when it has to compete with the already existing R-300.

Honeslty, Chalnoth, sometimes your logic defies, well, logic.[/quote]

Its funny, as I was reading this, I was wondering what market history you've (Joe) been looking at. Nvidia has always been able to get a premium for all of thier parts, and have been relatively unaffected by whatever offerings ATI has. Case in point: The Radeon 8500 is selling for LESS than the GF4MX, yet the GF4MX is shipping in much larger volumes. This has been a repeated trend since the TNT series in regards to the margins between Nvidia and ATI products. A simple look at the fiscal performance of both companies clearly illustrates this point.

At any rate, the NV30 is going to seriously throttle the R300, and in a few months the R300 boosters will seem only slightly less ridiculous than the Parhelia banner carries (god bless thier souls). When you have the highest performing products on the market, you can charge a premium. As far as which company "executed" better, we'll see when the OEM wins are announced.

 

[KimB]

Now that is a real stretch. First of all, for anyone to claim NV30 and R300 to be anything other than being based on similar technology is quite funny. From all estimates, both were planned to be released in the same time frame, it's just that ATI executed better than nVidia.

I kind of doubt that. nVidia has always shot for a late August/early September release for their fall line. If ATI is indeed launching tomorrow, it is almost assured that they have planned to for some time.

Also, nVidia has publicly stated that they have always tried to run a very tight ship in terms of not putting products out too early or too late. Too early to market, and your product won't have as many working features or won't perform as well. Basically, nVidia has longer to refine and optimize their design.

Second, having the design done first on 0.15, learning from it, and then tweaking it should enable ATI to be a better "shrink" to 0.13 than if they had just gone to 0.13 straight off. (Similar to GeForce to GeForce2 GTS "shrink".)

That is true...but nVidia will be able to learn just as much if they actually make their next Spring product a true refresh (similar to the GeForce3->GeForce4 transition, as opposed to the GF2 GTS->Ultra/Pro or GeForce3->Ti).

Finally, the fact that ATI gets their product out FIRST, means that they have at least a few months of higher profit margin on their part. That is the BENEFIT that ATI receives for spending the extra money to do two designs. NV30 will not be able to command the same type of profit margin upon NV30 release when it has to compete with the already existing R-300.

To a certain extent. If the Best Buy price I heard earlier is correct (Around $370 for the R300), then ATI won't even be making the type of margins that nVidia made on the GeForce4 Ti 4600 at launch. The problem for ATI's current profitability is that more people have more respect for nVidia currently (similar to an AMD vs. Intel scenario). It takes time to build the respect required in the current marketplace. Unless nVidia really screwed up, I don't believe ATI can quickly take away market share or make lots of quick profit.

 

[SirXcalibur]

At any rate, the NV30 is going to seriously throttle the R300, and in a few months the R300 boosters will seem only slightly less ridiculous than the Parhelia banner carries (god bless thier souls). When you have the highest performing products on the market, you can charge a premium. As far as which company "executed" better, we'll see when the OEM wins are announced.

Seriously , I don't browse Beyond3d for this kind of crap. If I wanted to hurt my intelligence I would watch a soap opera..
There is no way you can back that up with anyway. Neither card is even released yet.

Sorry just had to rant a bit there.

Other than that, we should see lots of Radeon 9700 info in a little while now.

 

[KimB]

What may be interesting is that the lack of DX9 may force ATI to go all OpenGL for their technology demonstrations. Until now, I don't think we've ever seen ATI being possibly much stronger in OpenGL than Direct3D.

 

[phynicle]

it was mean to be a little while a day ago...oh wait i live in australia..we're a day ahead

common where's all the info...!!!!!

 

[Johnny Rotten]

Doom,

R&D costs as absolute measurements are pretty much meaningless. You need to take them in the context of a companys net revenue for the figures to have any real value.

 

[SirPauly]

Its so humorous how people become:
-armchair economists
-armchair sales and marketting
-armchair financiers
-armchair 3d gurus
-armchair psychics
-armchair product engineers
-armchair asic designers

Hehehe........the amazing abilities of the ordinary armchair.

As an ordinary armchair gamer hehe.... I guess I am just glad I have another choice to consider in the near future.

Great quote....

 

[Nagorak]

As usual this thread has started to collect amusing logical leaps as the supporters of both sides go at it. I sort of think making any guesses as to the cost of this stuff, and thus the "better way" of doing things, is a waste of time. I mean, we know nothing, so any opinions we have are just totally contrived.

By the way, whoever was talking about Nvidia getting better margins on the TNT. While that may be true (since at the time ATi had the crappy Rage128), that's sort of revisionist history. At the time it was Nvidia vs 3DFX and their cards were selling for less than 3DFX. Nvidia's only been fetching a premium since 3DFX died as its taken ATi a little while to step up to the plate.

I kind of suspect their gravy train will be over as we move into this next year (not that they won't still make good money, they just won't have the whole market to themselves). Not only have ATi pretty much caught up with them in performance and features (some would say surpassed them in features), but there are a whole slew of other chipsets coming to market. Nvidia got where they are today by capitalizing on their graphics cards and running away with the market. Now, however, you'll notice they are beginning to diversify much more with motherboard chipsets, etc. If only they'd make add in sound cards (I have no understanding as to why they think "sound should be integrated"). Who knows, maybe they will as a way to fight back against Creative coming into their house with the P10.

 

[jvd]

Who knows whats going to happen , all i care about is what card is the fastest , has the best feature set and most important has the price point i want to pay. The only card company i will ot buy from is nvidia ( you all know why i've stated it before) I've been buying ati for awhile since it has give me what i wanted . If a power vr card comes out that give me that i will buy it I was going to buy matrox but it really disapointed me on the price performance ratio. IF ati can give me the same features as nvidia at a lower price but with a speed decrease 3 months before nvidia well heck thats what i will buy . But i really want a new powervr card.

 

[Dio]

When it comes to power or heat, it's not like my Ti4600 is exactly short of excessive heatsinking or small... I could play cricket with it!

 

[Dio]

Aha!

Anandtech has reviews, benchmarks and clockrate. 325MHz.

All those predicting the impossibility of such.... HA!

 

[Tahir2]

At last we have the announcement and it's launched.

The Radeon 9700 has exceeded my expectations by a wide margin..
I cannot believe they have actually got a 110+ million transistor GPU to run @ 325MHz (source Anandtech).

Amazing stuff.