NVIDIA GF100 & Friends speculation

[Silus]

The wafers are same but unfortunatly nvidia has a much bigger chip. So yes exactly as i sayed. Nvidias alu takes much more area than ATIs anyway. And to add things together a 1024 sp GF100 would be a monster even on next shrink. They need already a shrink for GF100 if they want to get on ATI level.

I didnt sayed that they will double everything again or that it should be much faster, i just allege that ATI could pull out a 3200sp chip after shrink while nvidia can be happy with a shrinked GF100 with higher clocks, added transistors and die area to made it manufacturable. And this with a 6 month late on market.

Get on ATI level of what ? Die size ?
If GF100 is as fast as the known benchmarks show (granted they are from NVIDIA, so we need to wait for independent ones), the die size difference isn't a big problem, especially when GF100 is also targeted for a much more profitable market, which will help with the costs.

 

[rpg.314]

its 40 to 28 not 32 so why isn't a doubling of ALU's possible given compraible die sizes? Also is it likely that AMD will change its 5D ALU's ? Sure make memory access and sharing more granular etc So 3200 might be a bit to far what about 2400-2800 depending on other changes. If AMD's clock speeds keep trending up as well.

Because, scaling is rarely, if ever perfect. I'll let experts like silent_guy explain in more detail the why's and "how often" bits of it.

5D alu's seem to be a good idea. Overall, AMD has about 30% advantage in flops/mm2, even if there is no "AMD compiler friendly ILP."

a 28nm AMD card H2 this year could be a monster.........................

Yes, but let's not jump the gun. Let it actually come-to-market.

 

[MDolenc]

The wafers are same but unfortunatly nvidia has a much bigger chip. So yes exactly as i sayed. Nvidias alu takes much more area than ATIs anyway. And to add things together a 1024 sp GF100 would be a monster even on next shrink. They need already a shrink for GF100 if they want to get on ATI level.

And how big is this GF100 actually? We have zero info about actual dimensions of the chip and just a lot of FUD about actual performance.
RV870 is also quite a bit larger then RV770 was, so do you think GF100 will be larger then GT200 by the same margin? Stop assuming that GF100 will be about twice the size of RV870 like it was the case with GT200 vs. RV770.

 

[DavidGraham]

TweakTown's own video card guru, Shane Baxtor has acquired some more specific details about NVIDIA's upcoming Fermi powered GTX 470, one of their first two DX11-ready cards to hit the market in answer to AMD's top-end HD 5800/5900 series.

With the very active rumour mill giving a hint that NVIDIA plan to make a big announcement regarding Fermi within the next 12 hrs, Shane gets an early scoop on the pricing and PSU requirements for the GTX 470. He learns that the official launch price looks to be $299 USD which sits in line with AMD's HD 5850, though performance is expected to be more around HD 5870 levels.

Power wise, NVIDIA is apparently slapping on a minimum requirement of 550 Watt, up slightly from AMD's min requirement on the HD 5850 at 500 Watt. In any case, it shouldn't matter to most people wanting a card like this, knowing full well that a quality PSU with plenty of headroom makes for a much more stable gaming rig.

In a separate posting, Shane is confident NVIDIA will announce their launch date for the Fermi cards within the next 12 or so hours which the rumour mill is indicating will be mid to late March. He also feels there may be an availability issue on the horizon with these cards due to the 1 year warranty that NVIDIA gives to its AIBs as opposed to a lot of the AMD cards on the market with 2 year warranties. This will make NV partners reluctant to carry a lot of stock of the cards as they know they'll need to offer 2 year warranties to stay competitive, but have fear of a lot of dead ones coming back to haunt them after the 1 year NV covered warranty period is up.

http://www.tweaktown.com/news/14302/more_details_on_nvidia_fermi_price_psu_requirements_and_more/

 

[rpg.314]

The TSMC reticle size is about ~580 mm2, so you can't make bigger chips even if you wanted to. BTW, is there a relation between reticle size and device sizes?

 

[ferro]

The first GTX470 & 480 prices have appeared. Image of the web site in case it disappears:

The GTX480 must really rock to justify a $679.99 price.

 

[rpg.314]

$299 for 470,

 

[Sontin]

The TSMC reticle size is about ~580 mm2, so you can't make bigger chips even if you wanted to. BTW, is there a relation between reticle size and device sizes?

So GT200 was nearly at the limit?

 

[KimB]

Yeah, that's a couple of exceedingly different price rumors right there...

 

[rpg.314]

That's what I heard. The reticle size might very well be ~10-15mm2 bigger. But the point is that gpu die sizes have for all practical purposes, hit an upper limit.

 

[GZ007]

5D alu's seem to be a good idea. Overall, AMD has about 30% advantage in flops/mm2, even if there is no "AMD compiler friendly ILP."

And with AMD-s clocks rising each generation while nvidias clocks are stalling or even falling down the advantage is increasing.

 

[rpg.314]

And with AMD-s clocks rising each generation while nvidias clocks are stalling or even falling down the advantage is increasing

Yeah, they better fix that pronto. However, let's wait for the final clocks to conclude that they are falling.

 

[Bouncing Zabaglione Bros.]

Yeah, that's a couple of exceedingly different price rumors right there...

<rubs hands>
<waits for AMD to drop prices>

If those prices are true that's gotta hurt Nvidia on such a large chip.

 

[MDolenc]

And with AMD-s clocks rising each generation while nvidias clocks are stalling or even falling down the advantage is increasing.

Again: how much has core clock on Radeon 5870 increased in comparison to 4870/4890?

 

[ZerazaX]

<rubs hands>
<waits for AMD to drop prices>

If those prices are true that's gotta hurt Nvidia on such a large chip.

Well Nvidia has cash, and if this product really is as late as it is, Nvidia is probably thinking: cut our losses, get our architecture out there, and move on.

ATI did the same thing with R600 after all

 

[rpg.314]

If those prices are true that's gotta hurt Nvidia on such a large chip.

I like the "you save : $ -70.00" bit in the screen shot for 470

 

[trinibwoy]

Well Nvidia has cash, and if this product really is as late as it is, Nvidia is probably thinking: cut our losses, get our architecture out there, and move on.

ATI did the same thing with R600 after all

The cash is one thing but at $299 how could they ever hope to keep it in stock? If there's any truth to this rumour I'm putting a stake in the ground that the 470 is a 384-shader part and there will be no 512-shader part on launch day.

 

[Alexko]

The cash is one thing but at $299 how could they ever hope to keep it in stock? If there's any truth to this rumour I'm putting a stake in the ground that the 470 is a 384-shader part and there will be no 512-shader part on launch day.

Well if it's about as fast or a tad slower than the HD 5850, keeping it in stock shouldn't be a problem.

 

[Silus]

Yeah, that's a couple of exceedingly different price rumors right there...

The specs don't even match up with what we know (specifically the GTX 470), so the prices might not be accurate at all.

 

[KimB]

The specs don't even match up with what we know (specifically the GTX 470), so the prices might not be accurate at all.

True enough, though granted, if there's going to be a mistake in a store's listing, it's likely going to be in overestimating the tech specs, not in getting the price wrong.