ATI is the Summer King?

[Joe DeFuria]

1. Cheaper to manufacture in the long-run...

Right. But all indications are that the 0.15 R-300 is a "temporary" solution until it can be re-spun (probably next spring) at 0.13. In other words, I don't believe 0.15 R-300 is a "long-term product" for ATI.

Much like nVidia's original GeForce. nVidia produced the original GeForce on a larger process to get it out the door, even though that resulted in a power hungry, "hot" product at a "relatively" low clock speed. But the "long term" GeForce product was the re-spun variants on more advanced process...the GeForce2 GTS. The original GeForce product was disontinued very quickly and replaced by the GTS and it's variant, the "pro."

The advantage to using a mature process vs. a "bleeding edge" process is obvious: getting the product out sooner rather than later.

EDIT: I still find it absolutley amazing if ATI manages to put an 8 pipe, DX9 compliant part with 107 million transistors on a 0.15 micron process, AND manages to clock it a 300 Mhz plus. That is truly astounding IMO.

 

[Tahir2]

Are ATi planning on announcing whatever it is they are planning on announcing (eek what a sentence) @ 23.59hrs in the UK?

I must admit I am a bit disappointed at this anti-climax.

Does anyone have any news? And not the kind of news that your brother-in-laws, Uncle's 2nd cousin removed, who knows someone at ATi kind of news.

Where is the announcement?!

 

[KimB]

EDIT: I still find it absolutley amazing if ATI manages to put an 8 pipe, DX9 compliant part with 107 million transistors on a 0.15 micron process, AND manages to clock it a 300 Mhz plus. That is truly astounding IMO.

Yes, that would be very impressive, which is why I have doubts.

Since it will have eight pixel pipelines, it seems almost assured that it will actually end up running at a lower clock speed than their current Radeon graphics cards.

But we *should* only have one day more to wait to find out for sure...it will be very interesting at the very least.

 

[Mize]

EDIT: I still find it absolutley amazing if ATI manages to put an 8 pipe, DX9 compliant part with 107 million transistors on a 0.15 micron process, AND manages to clock it a 300 Mhz plus. That is truly astounding IMO.

Yes, that would be very impressive, which is why I have doubts.

My bet is for >315 MHz on the high-end product.

Mize

 

[Mize]

Check the front page at www.driverheaven.net - stickers with prices for the 9700 ($350.99) and the 9000 ($149) are out at BestBuy retail stores tonight...maybe it won't be so much of a paper launch?

Mize

 

[KimB]

My bet is for >315 MHz on the high-end product.

Mize

Think about it for a moment. 8 pipelines at 315 MHz would more than double theoretical pixel fillrate. We haven't seen that happen since the inception of modern 3D graphics PC cards (unless you count Voodoo2 SLI vs. the original Voodoo...though even that's not entirely fair since there was a Voodoo SLI setup available before the V2). I see no reason why it would happen now.

But it would cerainly be incredible if ATI could pull it off, though I won't be nearly as impressed if they only do it in very limited quantities.

 

[Tahir2]

Check the front page at www.driverheaven.net - stickers with prices for the 9700 ($350.99) and the 9000 ($149) are out at BestBuy retail stores tonight...maybe it won't be so much of a paper launch?

Mize

Two things:

1) After what Best Buy did with that guy with the GF4 receipt he produced which got him arrested and later thrown out because the Best Buy manager said, "we don't want his custom, he was trespassing" Best Buy can best get lost IMO.

2) ATi have not announced anything yet! And we know what a farce Best Buy is (see above).

They say you shouldn't let 'one experience cloud your judgement'.. to that I say it 'could have been me who got arrested under false charges.'

No ATi announcement no launch. Doesn't matter what some store is claiming.

http://hypothermia.gamershardware.com/articles/bestbuy_gf4deal.html

 

[Mize]

Agreed.
My money says it will be announced tomorrow.
I don't have a spare $351 right now (what kind of pricing is THAT?)

Mize

 

[KimB]

Well, just because you don't like Best Buy doesn't mean they're not going to sell an R300 (btw, I read about that guy, too, and will never purchase from Best Buy again...).

I personally don't doubt that they will be for sale tomorrow (which might explain the delayed announcements).

 

[Tahir2]

I personally don't doubt that they will be for sale tomorrow (which might explain the delayed announcements).

To me that would be the biggest (pleasant) shock of this young millenia.

 

[Randell]

The external connector is only rumoured to be on the ATI top of the range >300mhz part, so natch it will expensive. As for heat, well have you seen the size of the cooler on the leaked pics, & on the Gf4 range?

For all intents and purposes, power consumption=heat dissipation (Yes, there is some infrared radiation, but that is quickly absorbed by other components in a closed case, meaning it is still effectively heat dissipation).

Therefore, if a board requires lots of power, it puts out lots of heat.

thnak you for that physics lesson

 

[Nagorak]

ATi R300: 107m transistors
NV30: 120m transistors

Anyone who thinks they're going to be vastly different is greatly mistaken, IMO. And if I remember the original Radeon had more transistors than the GF2, but that didn't make it faster.

You forgot to mention the smaller die process.

Smaller die process means:

1. Cheaper to manufacture in the long-run (basically means nVidia will have an easier time making money off of the NV30 than ATI can off of R300...won't mean a whole lot for price to the consumer)
2. Higher clock speeds.
3. Lower heat dissipation/power consumption.

The additional transistors won't double performance or anything of the sort, I'm sure, but may well give it many features that ATI simply won't have. Also, the later release date means faster memory will be available, and the .13 micron process may well allow for significantly higher clockspeeds, which could possibly make the NV30 perform quite a bit better.

Fair enough, but anyone expecting 100% speed increase from the NV30 is just going to be disappointed. Has there ever been a 100% speed increase in 3D graphics? Maybe between the V1 and V2, but other than that, not that I can think of.

I also think that it's highly unlikely the NV28 is going to be serious competition for the R300. But we'll see.

 

[Doomtrooper]

Chalnoth,

Nvidia may make more money from the Nv30 @ .13 yet ATI will also move to .13 in the new year....not to mention the record amount of money spent in R&D from Nvidia to make the move to .13...they will need to make up for some of that loss.

 

[KimB]

Put the R&D costs into perspective. nVidia's been doubling their revenues every year for a little while now. Their R&D costs have not doubled since last year, if I remember correctly.

 

[Doomtrooper]

I consider 120-180 million dollars for the Geforce 4 which was a refresh on proven .15 micron a significant amount of money and I assume the Nv30 will be 2 to 3 times that amount..bigger also means more overhead..more engineers, more salaries etc..

 

[phynicle]

wow 120-180 million...for that kind of money why don't they go and buy off some other company which has better innovations and idea than they do

btw where's r300!!!!!!

 

[Ailuros]

wow 120-180 million...for that kind of money why don't they go and buy off some other company which has better innovations and idea than they do

btw where's r300!!!!!!

They spent roughly over a 100M for both 3dfx and Gigapixel. About a million per patent if it sounds better.

 

[phynicle]

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

 

[RussSchultz]

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


All in support of their favorite company.

Please IXNAY ON the "NVIDIA VS ATI"/"MY PENIS IS BIGGER THAN YOUR PENIS" discussions.

 

[KimB]

Okay, here are some realistic numbers. Observe:

2000 revenue: $735 million
2000 profit: $275 million
2000 R&D cost: $86.4 million

2001 revenue: $1,371 million
2001 profit: $518 million
2001 R&D cost: $153 million

While I was a little bit off, you can see that nVidia is hurting in no way from their R&D costs. In fact, R&D is only a relatively small portion of nVidia's budget, which is rather interesting.

Anyway, the GeForce4 is currently making more than enough money for nVidia to spend on the NV30's R&D. There's no reason to believe that nVidia needs to make any back.

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, 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.